I sometimes feel like we're living in a spaghetti western when I think about the administrative reforms to the Executive Branch done by the current Administration. The governance landscape has been transformed into a sun-baked, unpredictable frontier where decades of established institutional norms are tossed aside to make way for the new guy in town who is invariably irascible, cynical, self\u2011interested, and governing a corrupt world. So, forgive me for invoking the famous title of Sergio Leone's flick in responding to the Office of Management and Budget's Proposed Regulation for Federal Financial Assistance.
\nI've summarized my response here using this (perhaps regrettable) device. I encourage you to scroll to the bottom to read my detailed response and analysis as submitted to OMB via regulations.gov. You can submit your own comment by the close of business tomorrow here.
\nWait\u2026the good? Yes, in my opinion there are some non-objectionable revisions to the Uniform Guidance in the proposed rulemaking. These are relatively few compared to the terrible bits, but are still worth highlighting:
\nMost of the proposal is very bad. While we've heard a lot in the press about the objections from the scientific community - there are implications that are terrible that go beyond that as well. Here are a few of the parts that I found particularly objectionable and are relevant to my interests:
\nBeyond being bad, there are several components that are ugly. These are provisions that OMB should be embarrassed by - and would have been during my tenure at the agency.
\nChristopher Steven Marcum, PhD
\nOpen Science Advocate
\n7/12/2026
I am writing as a concerned citizen and open science advocate. I was previously a Senior Statistician and Senior Scientist at OMB, had privileged access to earlier drafts of this revision, and worked on the 2024 revision to the Uniform Guidance. \nIn general, while there are a few non-objectionable revisions in this proposed rulemaking (see Section II below for those), it is on balance a terrible revision and I urge OMB not to finalize the rule.
\nThe preamble to the proposed rule adopts an ideological tone that departs from the historically objective, neutral character of administrative modifications to the Uniform Guidance. By expressly incorporating policy frameworks from politically biased external entities, such as the Heritage Foundation, the Office of Management and Budget (OMB) introduces political polarization into a framework governing over $1.1 trillion in federal assistance. It also departs from OMB's traditional evergreen stance on 2 CFR 200 and it should remain insulated from partisan influences to maintain broad public trust in federal funding.
\nOMB's Office of Information and Regulatory Affairs has designated this proposed rule as not economically significant under it's EO 12866 review. Not only was the economic analysis embedded into the regulatory impact analysis (RIA) inadequate, its conclusions that lead to a non-economic significance determination are wildly misaligned with reality.
\nThis proposed rule exposes over $1 trillion in annual federal spending to potential political influence, clawback, and civic lawsuits. The natural outgrowth of such practices will necessarily have trillions in derivative effects that span every sector of society (jobs lost, projects ended mid-development, etc). The failure to conduct a rigorous, data-driven regulatory impact analysis creates severe fiscal uncertainty for states, municipalities, small businesses, and universities. This lack of transparency was explicitly cited by Senator Susan Collins in her formal request for a full delay and withdrawal of these burdensome mandates. Potential economic impacts were also cited by Moody's in their own determination that the proposed rule would likely affect the credit ratings of federal grant recipients. Clearly, whomever conducted the economic analysis component of the RIA did not consider these macro effects. It's a major oversight and a process foul by OMB that could have consequences with respect to non-compliance with the Administrative Procedures Act because OMB did not follow its own policies for conducting a cost-benefit analysis under Circular A-4.
\nI urge OMB to revisit this inadequacy and resubmit the rule with a new regulatory impact analysis that fully considers the underlying macro economics of the proposed rule.
\nThe proposed revisions to Subpart E fail to clarify whether changes to allowable expenses would restrict or eliminate the capacity of university libraries to utilize indirect cost allocations for research journal subscriptions and institutional transformative agreements. As highlighted by the Association of Research Libraries), defunding journal subscriptions within the indirect cost pool would cause severe structural barriers to knowledge access across higher education. Clear and explicit language must be added to confirm that institutional indirect funding remains fully permissible for scholarly communications and institutional database licensing.
\nThe proposed rule introduces restrictions that would preclude federal award recipients from flowing funds to external entities categorized as advocacy groups via subcontracts. In practice, many highly specialized technical consultancies, scientific societies, and civic organizations maintain separate, lawful public advocacy operations and are supported by federal assistance through subcontracts.
\nThe proposed section mandates that a senior political appointee at each federal agency must review and approve every discretionary award prior to issuance. This requirement formally subordinates technical, merit-based peer review to political alignment with presidential policy priorities. As noted by the American Association for the Advancement of Science (AAAS) and the American Public Health Association (APHA), inserting a political layer into the award selection process fundamentally disrupts the objective insulation of scientific and biomedical research from changing political whims. This provision is an egregious assault on scientific integrity for research grantmaking. Moreover, it exposes non-scientific financial assistance (the majority of assistance exposed to 2 CFR 200) to such whims and injects political uncertainty into the risk equations of organizations that rely on federal grants for their programmatic work.
\nThe expansion of agency authority to unilaterally terminate active awards for administrative convenience or shifting national interest creates extreme operational instability. DOGE used this extensively in 2025 to cancel government contracts, sometimes outside the existing terms and conditions contracts between the government and its awardees. It's a terrible idea to codify this rash and irresponsible behavior in reg.
\nScientific and public infrastructure programs are inherently long term, often spanning multiple presidential administrations. Unilaterally stopping active projects because of shifting political priorities results in substantial waste of taxpayer funds, as partially completed research or infrastructure yields minimal, if any, return-on-investment. More broadly, this instability weakens multi-year funding commitments to small businesses and state and local governments, potentially creating a credit negative environment for heavily dependent entities as analyzed by Moody's and reported on by Bond Buyer).
\nThis provision seeks to suppress foreign collaborations in research and development historically limited by the Wolf Amendment to NASA's appropriations from reaching China. Imposing absolute prohibitions on international scientific co-authorship and academic data exchange without clear, targeted statutory directives creates an unnecessary bureaucratic burden. This blanket expansion threatens to suppress international scientific collaboration, which is essential for addressing global challenges in public health, physical, and environmental science.
\nThe proposal disallows Article Processing Charges (APCs) and open access publishing fees as direct allowable expenses. While alternative publishing mechanisms such as preprints and the systematic exercise of the federal purpose license as codified in 2 CFR 200.315 by granting agencies provide a more sustainable framework for public access, an immediate, uncoordinated ban on APC direct charges could have unintended consequences (as I've previously discussed in detail elsewhere).
\nI urge OMB to revise this section to clarify an alternative to traditional scholarly publishing as the mechanism by which research results and data are shared by grantees: including directing federal agencies to assert their federal purpose license over all content supported by their grants and to allow for preprints to satisfy public access requirements.
\nThe proposed section contains provisions that would allow the Department of Justice to formally join or facilitate private, third-party civil society lawsuits against federally funded projects and award recipients for alleged grant noncompliance. This mechanism creates a highly weaponized oversight structure, exposing public entities, universities, and local governments to costly, politically motivated litigation. It is an unprecedented overreach of authority and further politicizes federal grantmaking in a way that will disrupt the continuity of programs benefiting from federal assistance across the timelines of one Administration to another.
\nI support OMB's reclassification of 2 CFR Part 200 from administrative guidance to a regulation. I previously advocated for this move during the last revision process in 2024 when I was at OMB, as it is onerous for the public to be exposed to dozens of deviations in grantmaking policies across the government. As Cole Donovan and I wrote earlier this year, interagency coordination efforts often fail because of decentralized, competing interests across the executive branch. This shift eliminates the administrative fragmentation that occurs when individual federal agencies interpret and implement grantmaking policies inconsistently, creating a single, predictable framework for grantees.
\nAlthough not explicitly articulated as such, the proposed rule introduces data governance language that closely aligns with the GREAT Act, the Open Government Data Act and the related OMB guidance M-25-05, as well as the 2022 OSTP Public Access Policy. I encourage OMB to further clarify that all data, code, and publications produced under a federal award that are acquired by a federal agency should be made immediately publicly accessible and inventoried appropriately consistent with the requirements of M-25-05.
\nThe proposed text correctly maintains and reaffirms the long-standing federal purpose license framework in 2 CFR 200.315. This provision ensures that the federal government retains a royalty-free, nonexclusive, and irrevocable license to reproduce, publish, or otherwise use works developed under a federal award for public purposes. Retaining this language provides a strong, legally established alternative to expensive commercial publishing models and safeguards public access to technical data. OMB should strengthen this by requiring all agencies bounded by the regulation to explicitly assert this license over all data, manuscripts, and code (and other intangible property) resulting from a federal award.
\nThe rule provides a refined and updated definition of Personally Identifiable Information (PII). This is a welcomed change and one that I advocated for during the previous revision to 2 CFR 200 in 2024. This update aligns the definition of PII with other OMB and agency guidance and regulations. It was always mis-defined in 2 CFR 200.
\nThe proposed updates introduce more robust reporting obligations and transparency standards for pass-through entities managing federal funds. There is a need for consistent, systematic transparency regarding how subawards are distributed, monitored, and accounted for on centralized platforms like SAM.gov. This proposal improves financial stewardship of taxpayer monies. It should reduce the risk of waste, fraud, and abuse in federal assistance programs.
\nThe proposed text also offers improved guidance to distinguish whether an external entity is functioning as a contractor or a subrecipient under a federal award. This change is needed because previous guidance was confusing and frequently resulted in misaligned audit findings and inconsistent compliance monitoring.
\nIn addition to the PII definition clarification, I support the clarification that confidential business information (CBI) is protected and cannot be compelled to be released without a legal justification. CBI is one aspect of confidential information that statistical agencies collect and must protect under the Confidential Information Protection and Statistical Efficiency Act that relies on a class E-felony for unlawful disclosure as a deterrent. This clarification helps align the obligations of trust imbued on the statistical agencies with the expectations of grantmaking agencies to protect CBI.
\nHowever, I urge OMB to clarify that both unawarded and awarded proposals and grant applications are NOT considered CBI. The public should have full awareness of what entities are proposing to do with taxpayer monies. Stuart Buck and I wrote about the importance of this for government transparency purposes last year.
","doi":"https://doi.org/10.59350/gpk1d-kvn12","guid":"https://www.chrismarcum.com/marcum-blog/2026/07/12/My%20Response%20to%20Uniform%20Guidance%20Rulemaking","language":"en","license":"https://creativecommons.org/licenses/by/4.0/legalcode","published_at":1783814400,"rid":"x75xm-mj444","summary":"I sometimes feel like we're living in a spaghetti western when I think about the administrative reforms to the Executive Branch done by the current Administration. The governance landscape has been transformed into a sun-baked, unpredictable frontier where decades of established institutional norms are tossed aside to make way for the new guy in town who is invariably irascible, cynical, self\u2011interested, and governing a corrupt world.","tags":["General","Government"],"title":"My Response to OMB's Proposed Regulation for Federal Financial Assistance","updated_at":1783872436,"url":"https://www.chrismarcum.com/marcum-blog/2026/07/12/My-Response-to-Uniform-Guidance-Rulemaking.html","version":"v1"}},{"document":{"authors":[{"affiliation":[{"id":"https://ror.org/02mb95055","name":"Birkbeck, University of London"}],"contributor_roles":[],"family":"Eve","given":"Martin Paul","url":"https://orcid.org/0000-0002-5589-8511"}],"blog":{"authors":[{"name":"Martin Paul Eve","url":"https://orcid.org/0000-0002-5589-8511"}],"community_id":"9224b0d7-fc03-497c-9c6f-85c9fd1e72da","created":1690329600,"current_feed_url":null,"description":null,"favicon":"https://rogue-scholar.org/api/communities/9224b0d7-fc03-497c-9c6f-85c9fd1e72da/logo","feed_format":"application/atom+xml","feed_url":"https://eve.gd/feed_all.xml","filter":null,"generator":"Jekyll","home_page_url":"https://eve.gd","issn":null,"language":"eng","license":"https://creativecommons.org/licenses/by/4.0/legalcode","prefix":"10.59348","relative_url":null,"secure":true,"slug":"eve","status":"active","subfield":"1208","title":"Martin Paul Eve","updated":1783867107,"use_api":true},"blog_name":"Martin Paul Eve","blog_slug":"eve","content_html":"After a recent post by Brennan Kenneth Brown, I decided to investigate what it would take to get this site \u2013 and, then, Knowledge Commons \u2013 onto the ATmosphere, using ATProto. This was a pretty steep learning curve, as I did not, before, understand many of the key concepts in this space. After succeeding, though, in publishing this blog to the ATmosphere (I took Brennan's excellent advice to proceed immediately to Sequoia) I am left pondering.
\nFirst, what is this ATmosphere thing? Well, it's the protocol behind Bluesky and, basically, it provides a decentralized storage system for objects of any type. Bluesky posts have a type of \"app.bsky.feed.post\" while my blog posts are \"site.standard.document\". Various new platforms are emerging that can read and write to the ATmosphere, so it gives a new discovery layer (see e.g. this search), but also, in theory, a decoupled presentation layer; so how the site appears will be in the hands of the receiver, rather than determined by the writer.
\nNow, though, let's see what I have achieved\u2026 If you visit a recent post, you may notice that there are now two link tags in the document head:
\n<link rel=\"site.standard.document\" href=\"at://did:plc:hnpt7ns2lecdujegbi6qkqqm/site.standard.document/3mq7lupu73d2i\">\n<link rel=\"site.standard.publication\" href=\"at://did:plc:hnpt7ns2lecdujegbi6qkqqm/site.standard.publication/3mq4tnwmt3w2i\">\nThese links tie the page to its ATmosphere equivalent. Now, you can't just browse the ATmosphere in a web browser. The content has to have an interface written in front of it, that pulls the content in from this protocol and displays it. However, no fear, because there's an experimental browser for it! So, if you visit the corresponding file, you can, indeed, confirm that I have been successful in getting my content in.
\nSo, interestingly, there are no current standard.site or standard.document aggregators that display the document in their own interface (at least that I could see). Everywhere I tried linked back to the canonical web reference. But, interestingly, they WERE able to find the content and pull it into their aggregation streams.
\nAnother advantage is that, if my current PDS (where the data is stored \u2013 a Personal Data Server \u2013 which is actually on Bluesky's servers, as they run my current storage location) decides to shut down or move, and I no longer wish to be on their system, I can move the files and everything will apparently just be nicely migrated, without needing permission from the old PDS. Well, that sounds good, but I haven't tried it.
\nHowever, the entire reason I did this experiment was to see whether this was something worth pursuing for Knowledge Commons.
\n\u2026 and, for now, I am not sure. At the scale that we operate over there, it would take some effort to make sure we were doing it right. I also think that we would need to run our own PDS. On my personal blog, I hit the daily rate limit when I tried to upload 994 blog posts and associated file blobs (cover images etc.)\u2026 and then tried to modify every single one of them. I was locked out of Bluesky's PDS for 24hrs, which was annoying. However, KC has many, many more blog posts than this that would require initial loading. And if we ever had to do a full forced update of all content (say the canonical URL changed - gasp!), the process would need batching, it would be a pain, etc. But running our own PDS is also a pain. It's a whole new infrastructure that needs tending, opening new security holes, exposing us to the outside world\u2026 and that could go wrong.
\nHence, my conclusion is that, for now, I think we should watch and wait. If there is really substantial traction in this place \u2013 or if lots of people ask for it \u2013 I will certainly reconsider. But I would feel some trepidation about boldly going into this space, right now.
\nInvestigating ATProto here and for Knowledge Commons was originally published by Martin Paul Eve at eve.gd: Martin Paul Eve on July 10, 2026.
","doi":"https://doi.org/10.59348/r3x69-k5d70","guid":"https://doi.org/10.59348/r3x69-k5d70","language":"en","license":"https://creativecommons.org/licenses/by/4.0/legalcode","published_at":1783641600,"rid":"5jfay-xjh33","summary":"After a recent post by Brennan Kenneth Brown, I decided to investigate what it would take to get this site \u2013 and, then, Knowledge Commons \u2013 onto the ATmosphere, using ATProto. This was a pretty steep learning curve, as I did not, before, understand many of the key concepts in this space. After succeeding, though, in publishing this blog to the ATmosphere (I took Brennan's excellent advice to proceed immediately to Sequoia) I am left pondering.","title":"Investigating ATProto here and for Knowledge Commons","updated_at":1783867489,"url":"https://eve.gd/2026/07/10/investigating-atproto-here-and-for-knowledge-commons/","version":"v1"}},{"document":{"authors":[{"affiliation":[{"id":"https://ror.org/02twcfp32","name":"Crossref"}],"contributor_roles":[],"family":"Feeney","given":"Patricia","url":"https://orcid.org/0000-0002-4011-3590"}],"blog":{"authors":[{"name":"Crossref Staff"}],"community_id":"093ada45-3a02-4007-b8b6-be28f221e01d","created":1780876800,"current_feed_url":null,"description":"Recent content in Blog on Crossref","favicon":"https://rogue-scholar.org/api/communities/093ada45-3a02-4007-b8b6-be28f221e01d/logo","feed_format":"application/atom+xml","feed_url":"https://www.crossref.org/blog/feed.xml","filter":null,"generator":"Hugo","home_page_url":"https://www.crossref.org/blog/","issn":null,"language":"eng","license":"https://creativecommons.org/licenses/by/4.0/legalcode","prefix":"10.64000","relative_url":null,"secure":null,"slug":"crossref","status":"active","subfield":"1710","title":"Crossref Blog","updated":1783604547,"use_api":null},"blog_name":"Crossref Blog","blog_slug":"crossref","content_html":"Research is rarely limited to a single contributor performing a single role. Behind every research output are people contributing in various ways: software development, data analyses, methodology design, and much more. Often, the same person contributes in several of these ways. Until now, Crossref metadata could only capture part of that picture, and this is changing now.
\nCrossref Schema 5.5 includes several improvements across different content types, but its most significant enhancement is the expanded support for contributor roles through the introduction of multiple roles per contributor, option to specify the corresponding author, and compatibility with the CRediT (Contributor Roles Taxonomy): a community-owned taxonomy of 14 contributor roles, which has been adopted and made available in multiple languages.
\nThese enhancements allow members to describe research contributions in much greater detail, creating richer metadata that better reflects how research is actually produced, and supporting greater accountability and more comprehensive research assessment.
\nIf your workflow already distinguishes between different kinds of contributions, Schema 5.5 gives you a way to record that detail more accurately using the CRediT taxonomy values. CRediT can be adopted gradually, where it fits your editorial or production workflow.
\n![\"Current](\"https://www.crossref.org/images/blog/2026/current-vs-new-roles-xml.png\")
Figure 1: Until now, contributors could be assigned a single contributor role using Crossref's existing contributor role vocabulary. In Schema 5.5, members can indicate that the same contributor was responsible for different roles, such as corresponding author; writing: reviewing and editing; and data curation.
\nExisting deposits remain fully supported, and members can continue using the current contributor role attribute while planning implementation of the new repeatable role type element. For our members, who have been using CRediT in their workflows already, as ever \u2013 we encourage updating your metadata when practicable.
\nThis update gives more accurate credit to all of the people behind research outputs. Crossref vocabulary includes roles that aren't recognised in CRediT, and vice versa. Capturing richer contributor metadata recognises contributions that may not be visible in a single author line and improves transparency around how research is produced, thereby enabling downstream systems to interpret that information more reliably. The update also offers better interoperability with CRediT, which is well recognised across the scholarly ecosystem.
\n![\"Expanding](\"https://www.crossref.org/images/blog/2026/schema-55-infographic.png\")
Figure 2: Schema 5.5 is an expansion of Crossref contributor metadata. Members can describe contributors using Crossref's existing contributor role vocabulary, as well as the internationally recognised CRediT taxonomy.
\nIn turn, this strengthens metadata reuse across repositories, discovery services, funders, institutions and other infrastructure providers; and supports evaluation, reporting and discovery workflows. Better contributor metadata strengthens the connections that make up the Research Nexus.
\nBeyond the expanded contributor support, Schema 5.5 includes several additional enhancements across the metadata schema.
\nSupport has been added for metadata elements that were previously missing from report series records, including Crossmark, funding, and licence information.
\nPosted content includes preprints, eprints, and other types of content that have been posted to a stewarded host platform. We're all about persistence, so it's vital that everything registered with us be maintained. Note that accepted manuscripts are not considered posted content. Schema 5.5 refreshes posted content sub-types by introducing blog and poster.
\nAt the same time, we are \"retiring\" working paper, dissertation, and report from posted-content sub-types. Over time, these have been developed into separate record types that benefit from richer, dedicated schemas.
\nFinally, archive locations can now also be included for posted content records.
\nA new archive location, CINES, has been added to the list of supported archive providers.
\nClinical trial information is no longer limited to journal articles and conference papers. Schema 5.5 extends support across additional content types, including books, datasets, dissertations, reports, posted content, standards, and pending publications.
\nTaken together, the updates in our latest schema support more holistic recognition of contributions to the research and its communication, as well as greater accountability and integrity in related processes.
\nTo support gradual adoption, Schema 5.5 maintains backwards compatibility with existing deposits. Members can continue using the current contributor_role attribute while preparing to implement the new repeatable role element. We have prepared a migration guide to help members transition to Schema 5.5.
As you prepare to adopt Schema 5.5, we encourage members to include contributor roles whenever they are available from editorial workflows and to use recognised vocabularies consistently, including CRediT roles where appropriate.
","doi":"https://doi.org/10.64000/6xx3c-nmp80","guid":"https://doi.org/10.64000/6xx3c-nmp80","language":"en","license":"https://creativecommons.org/licenses/by/4.0/legalcode","published_at":1783555200,"rid":"zf0xb-7ws79","summary":"Research is rarely limited to a single contributor performing a single role. Behind every research output are people contributing in various ways: software development, data analyses, methodology design, and much more. Often, the same person contributes in several of these ways.","tags":["Crossref","Infrastructure","Interoperability","Metadata","Research Nexus"],"title":"Schema 5.5 now available: adding CRediT, new record types for blogs and posters, and more","updated_at":1783835266,"url":"https://www.crossref.org/blog/schema-5.5-now-available-adding-credit-new-record-types-for-blogs-and-posters-and-more/","version":"v1"}},{"document":{"authors":[{"contributor_roles":[],"family":"Felczak","given":"Michael"}],"blog":{"authors":null,"community_id":"77c8c2e4-ebda-4e7c-9458-6c06b604344b","created":1752192000,"current_feed_url":null,"description":null,"favicon":"https://rogue-scholar.org/api/communities/77c8c2e4-ebda-4e7c-9458-6c06b604344b/logo","feed_format":"application/atom+xml","feed_url":"https://pkp.sfu.ca/feed/atom","filter":null,"generator":"Other","home_page_url":"https://pkp.sfu.ca/","issn":null,"language":"eng","license":"https://creativecommons.org/licenses/by/4.0/legalcode","prefix":"10.59350","relative_url":null,"secure":true,"slug":"pkp","status":"active","subfield":"1710","title":"Public Knowledge Project","updated":1783645240,"use_api":null},"blog_name":"Public Knowledge Project","blog_slug":"pkp","content_html":"![\"Feature](\"https://pkp.sfu.ca/wp-content/uploads/2026/07/Resource-managing-AI-traffic-July-2026-WP-1024x576.jpg\")
PKP's Head of Systems, Michael Felczak, underlines how commercial AI traffic has been putting a strain on hosting providers such as PKP. Read on to learn how PKP is balancing traditional and new tools to protect its hosted journals.
\nWith the rapid growth of commercial AI services and a plethora of applications and agents built upon these services, hosting providers have witnessed an exponential growth in network traffic to their hosted websites. Automated traffic has recently surpassed human-generated traffic and today represents the bulk of access requests to online journals, monograph collections, and preprint servers.\u00a0
\nKeen to train their LLMs, AI companies and service providers view open access scholarly publishing as a free source of data that does not require permission or contractual arrangements typical among commercial operators.\u00a0
\nAs a result, hosting providers are struggling to maintain an acceptable level of service for readers due to new demands from automated traffic, which places high demands on server and network resources. Much of this automated traffic is aggressive, involving requests for pages and URLs at high rates that far exceed human access to published content.
\nTo address the problem of automated traffic, PKP Publishing Services has deployed traditional network management tools alongside new tools that target AI harvesting. These traditional methods include firewalls to restrict origin IP addresses, network throttling to slow down aggressive traffic, and community resources that aggregate information about offending service providers.\u00a0
\nWhile traditional methods are effective to some extent, they are limited to a set of rules that must establish in advance what and who to block. Today AI crawlers often adjust their behaviour, conceal their fingerprints, and switch to a different set of origin IP addresses.\u00a0
\nThe end result is an AI-based iteration of whack-a-mole that requires staff time and resources to monitor traffic, adjust rules, and deploy additional filters to maintain an acceptable level of service for hosted journals and readers. Equally importantly, since AI services are now commonplace for discovery of content and search results, commercial AI services require access to published content and cannot be completely blocked. This additional AI-centric traffic and server activity requires additional server and networking resources.
\nTraditional network management tools co-exist alongside new tools and services that target AI automated traffic. Recently several open source projects have been launched that support dynamic AI crawler filtering and PKP Publishing Services has started to deploy these tools on a limited scale to assess their effectiveness.\u00a0
\nAnubis, the most popular of these projects, filters human traffic from automated traffic by presenting a challenge to website visitors, similar to Google's reCaptcha. Unlike Google's reCaptcha, the Anubis challenge does not require user input and is instead verified by the user's browser.\u00a0
\nEven if AI bots adjust their behaviour or origin IP addresses, if they are unable to complete the Anubis challenge they will continue to be actively blocked. Since it doesn't require user input, from an accessibility perspective the Anubis challenge is more accessible compared to a traditional captcha.\u00a0
\nThis ongoing work represents the commitment of PKP Publishing Services to protect hosted journals from aggressive traffic while balancing the need of emerging AI services to index open access content to ensure that it can be discovered by readers.
\nThe post Managing AI Traffic for PKP Hosted Journals appeared first on Public Knowledge Project.
","doi":"https://doi.org/10.59350/0dm0q-kze91","guid":"https://pkp.sfu.ca/?p=19364","image":"https://pkp.sfu.ca/wp-content/uploads/2026/07/Resource-managing-AI-traffic-July-2026-WP-1024x576.jpg","language":"en","license":"https://creativecommons.org/licenses/by/4.0/legalcode","published_at":1783641600,"rid":"jth0k-cqy48","summary":"PKP's Head of Systems, Michael Felczak, underlines how commercial AI traffic has been putting a strain on hosting providers such as PKP. Read on to learn how PKP is balancing [\u2026] The post Managing AI Traffic for PKP Hosted Journals appeared first on Public Knowledge Project.","tags":["Community Newsletter","News","News For Hosted Clients","AI","Open Journal Systems (OJS)"],"title":"Managing AI Traffic for PKP Hosted Journals","updated_at":1783835229,"url":"https://pkp.sfu.ca/2026/07/09/managing-ai-traffic-for-pkp-hosted-journals/","version":"v1"}},{"document":{"authors":[{"affiliation":[{"id":"https://ror.org/0153tk833","name":"University of Virginia"}],"contributor_roles":[],"family":"Turner","given":"Stephen D.","url":"https://orcid.org/0000-0001-9140-9028"}],"blog":{"authors":[{"name":"Stephen Turner"}],"community_id":"382941a7-2ffa-41df-8bbb-5f772188517f","created":1780876800,"current_feed_url":null,"description":"A practicing data scientist's take on AI, genomics, biosecurity, and the ways AI is reshaping how science gets done. Weekly updates from the field. Occasional notes on programming.","favicon":"https://rogue-scholar.org/api/communities/382941a7-2ffa-41df-8bbb-5f772188517f/logo","feed_format":"application/rss+xml","feed_url":"https://blog.stephenturner.us/feed","filter":null,"generator":"Substack","home_page_url":"https://blog.stephenturner.us","issn":null,"language":"eng","license":"https://creativecommons.org/licenses/by/4.0/legalcode","prefix":"10.59350","relative_url":null,"secure":true,"slug":"stephenturner","status":"active","subfield":"1311","title":"Paired Ends","updated":1783715968,"use_api":null},"blog_name":"Paired Ends","blog_slug":"stephenturner","content_html":"GPT-5.6 came out yesterday. Plenty of other people are writing about it. I'm not yet, because I'm trying to be intentional about my AI use in July (it's not going great \u2014 AI Dry July is more like an AI-damp-kinda-moist July). But, I was able to get a correct one-shot answer to something many smart people I know get consistently wrong, despite all the evidence widely available.
Another six things this week. Stretching to find a common theme here: each of them touch on knowledge in some way or another:
Anthropic and AE Studio's GRAM, and why removing knowledge from weights isn't the same as removing risk.
PhD admissions down, and what funding uncertainty does to a cohort.
A Brown economics class where the take-home midterm averaged near 100% and the in-person final averaged below 50%.
Watching long form reading erode (keep reading \ud83d\ude09)
AI labs becoming biotech firms.
on the \"Bio Mythos\" moment, and building biosecurity assurance.
And a note, with a very special thanks to my paid subscribers (remember, every post here is free and open, regardless of whether you pay): These kinds of recap posts take a lot of time and energy to write. And exceptionally so while avoiding AI for discovery, scanning, summarizing, triage. I also have a major, very major proposal deadline at the end of this month. While I have a few other essays I'll be publishing soon, you won't see another recap like this for a while. Follow some of the other Newsletters I recommend to keep you caught up.
Anthropic and AE Studio published a method this week for giving a model a removable compartment per category of dual-use knowledge.
![\"https://alignment.anthropic.com/2026/modular-pretraining/fig1.png\"](\"https://substackcdn.com/image/fetch/$s_!lRMF!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fde81697c-ec3b-47ad-848e-0acb3b8dea70_1999x611.png\" "\\"https://alignment.anthropic.com/2026/modular-pretraining/fig1.png\\"")
They call it GRAM, for Gradient-Routed Auxiliary Modules: add extra neurons to every layer, route the gradient updates from virology or cybersecurity or nuclear-physics text into that category's module during pretraining, then delete the module at inference when you don't want the capability. One training run yields a model you can reconfigure many ways across domains instead of training many filtered models.
They compared with post-hoc unlearning. When they tried to fine-tune a removed capability back in, the unlearning baseline (MaxEnt) recovered almost fully, while GRAM and plain data filtering held. Unlearning after the fact suppresses more than it removes.
* = The asterisks: everything is measured in next-token loss and not actual downstream performance, the dual-use data was a sliver of the mix, nothing has touched a production Claude model, and some capabilities may be
so entangled with general knowledge that no method can separate them cleanly.
Really cool work here but important to pair this with a well-designed uplift study, and better yet, measuring correlates of uplift. Lower next-token loss on virology tokens isn't the quantity that most people (policymakers, biosecurity folks, general public) care about. Uplift is: can someone holding the ablated model do the dangerous thing any less well than someone with a search engine? So model-editing and unlearning work needs paired human uplift studies, same models, capability toggled on and off, measured against a real baseline. Removal that looks clean in loss space but buys no drop in uplift might look like safety but isn't.
NYT / Vimal Patel: Decline of Ph.D. Admissions Could Imperil a 'Generation of New Talent' Gift Link.
PhD admissions this fall dropped 15% from last year across 55 Association of American Universities members, per the AAU Data Exchange. Those schools confer about half the country's research doctorates. This all stacks on last years numbers: for the 42 schools that reported fall 2025, new enrollments were already down. I.e., Two straight years of contraction at the institutions that produce most of the nation's new scientists.
\u2026developing research talent \"is more like planting an orchard than filling a warehouse.\"
Some are far worse than the average. MIT expects nearly 20% fewer new graduate students, about 500 people, citing federal awards down 20% and the new endowment tax. Caltech is cutting new graduate admissions 40% across the board for fall 2026, and its graduate dean was explicit that the driver is a lack of funding certainty, not any specific cut. UW's astronomy chair took zero new doctoral students this year, the first time since 2016.
I look at this with research dean hat on. You need students to drive research. The cohort you don't admit this year doesn't come back when the budget stabilizes.
IHE / Emma Whitford: Brown Professor Suspects Majority of His Class Used AI to Cheat. https://www.insidehighered.com/news/faculty/learning-assessment/2026/07/08/brown-professor-suspects-most-his-class-used-ai-cheat
Roberto Serrano, who has taught welfare economics at Brown for nearly 20 years, gave a take-home midterm for the first time this spring, partly because students were uneasy in a classroom after the December shooting on campus. The class, normally around 30, had 86. The midterm averaged 96% against a historical 65-to-80 range, on an exam he'd made harder. He ran the questions through ChatGPT, got answers matching his students' work down to the same overcomplicated proof strategy, and told the class he suspected widespread AI use. He moved the final in-person.
18 people dropped, 9 skipped the final altogether, and of those who sat it the average was 48%, the lowest he's recorded, against a prior floor of 65. 19 failed.
A 96% to 48% collapse between take home and proctored is strong circumstantial evidence the take home scores weren't the students' own work. It isn't proof about any individual. Brown's academic-code committee wants separate complaints per student with exam copies. AI detectors would throw false positives and negatives in bulk. The committee's own AI report, out the same week, recommends de-emphasizing punishment. His own summary was blunter:
We cannot choose to become idiots.
Atlantic / Rose Horowitch: The End of Reading Is Here. https://www.theatlantic.com/magazine/2026/08/reading-crisis-postliterate-age/687618/
The Atlantic's August cover essay, by Rose Horowitch, argues America has gone postliterate: not illiterate, but losing the higher-order comprehension sustained reading builds. Fewer than half of adults read a book of any kind in 2022. Reading for pleasure on a given day fell from 28% in 2004 to 16 in 2023, and gambling has passed reading as a leisure activity. Text now thrives inside a shrinking minority, about 20% of adults accounting for more than 80% of books read.
What are you reading now? Last week I finished re-reading Andy Weir's Project Hail Mary (the encore was just as good). I just started reading Stewart Brand's Maintenance of Everything, while listening to David Sedaris's newest collection of essays, The Land and Its People (read by him).
at Scaling Biotech on Anthropic's early-July drug-discovery announcements: Claude Science (which I test-drove here) and an internal effort to find candidates for rare and orphan diseases. His read is that it backfires commercially. Tokens are expensive, open-weight models are closing the gap, and
the whole point of it is to make you use Anthropic's models
instead of self-hosting, so it has to stay pricey. Worse, an in-house drug-discovery team reads to pharma customers as a future competitor trained on their data, however carefully the rare disease problem is chosen.
Related, Amelia Palermo's Nature Biotechnology comment provides some structural context.
Palermo, A. Frontier AI companies as biotech acquirers. Nat Biotechnol (2026). https://doi.org/10.1038/s41587-026-03214-0
In April, Anthropic acquired Coefficient Bio for about $400 million (all stock): roughly 8 months old, <10 computational biologists, no clinical assets. Against a $380B valuation that's about 0.1% dilution. In other words it's a standard tech exit rather than a typical pharma one, with a value Palermo calls
a purchase of foundation model talent and code base.
She sees a new class of buyer, frontier labs acquiring biology platforms for models, data, and people instead of drugs, and predicts several such deals by 2030.
I just took a look at some of the job ads, which point to how serious these companies seem. Anthropic isn't only reselling Claude to pharma. Looks like they're building a wet lab, hiring bench scientists and the people to run it:
OpenAI's public bio hiring leans the a different direction, toward safety, policy, and red-teaming. Which is near and dear to my heart. I can't wait to see what this team does here.
Most of these listings are in the $300k-400k range, plus equity.
and at : Preparing for the \"Bio Mythos\" Moment.
Governments tend to react to AI risk only after a surprise, and Claude Mythos was the cyber case: its ability to find and exploit weaknesses in critical infrastructure software pushed cyber capability to the top of the agenda, and the Fable takedown (reportedly 90 minutes for Anthropic to bar foreign access before it pulled the model worldwide) showed how little assurance infrastructure exists to make those calls deliberately. The controls have since been lifted, though Commerce keeps the right to reimpose them. The argument is that a \"Bio Mythos\" moment, a model crossing into dangerous biological capability, would trigger the same improvisation in a domain where the danger is harder to measure and, in their words,
self-replicating, offense-dominant, and can have enormous societal ramifications.
Their fix is to build the assurance infrastructure before that moment: independent evaluators with real model access and a liability safe harbor, standardized capability tests like their BioTIER, and a default against exposing advanced bio capabilities to unverified users.
![\"\"](\"https://substackcdn.com/image/fetch/$s_!Z1bI!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F42d8172f-ca1e-4403-98d1-02c93ce18b24_2048x1593.png\")
Their Bio Capabilities Index shows the trend rising even as newer models add refusals. This is the missing half of the GRAM story in #1 above: you can build an off switch, but someone has to decide when to flip it, and that call needs more than a back-channel phone call (maybe more well-designed uplift studies, and better yet, measuring correlates of uplift). For my fellow Virginia readers, the state's 2026 legislation already directed its technology commission to study an independent-verification framework.
","doi":"https://doi.org/10.59350/0aew2-jrr03","guid":"206293199","image":"https://substackcdn.com/image/fetch/$s_!Z1bI!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F42d8172f-ca1e-4403-98d1-02c93ce18b24_2048x1593.png","language":"en","license":"https://creativecommons.org/licenses/by/4.0/legalcode","published_at":1783641600,"rid":"8jncm-2x730","summary":"Off switches for dual-use knowledge, the PhD admissions drop, a Brown cheating mess, the end of reading, AI labs at the bench, and planning for a \"Bio Mythos\" moment. 1.7k words, 8 min reading time.","tags":["Papers","Biosecurity","AI"],"title":"Closing my tabs, July 10, 2026","updated_at":1783717232,"url":"https://blog.stephenturner.us/p/closing-my-tabs-july-10-2026","version":"v1"}},{"document":{"authors":[{"contributor_roles":[],"family":"Strauss","given":"Ilan"}],"blog":{"authors":[{"name":"Asimov's Addendum"}],"community_id":"0b81a230-f3a4-4f15-b6db-f262083c3797","created":1781481600,"current_feed_url":"https://asimovaddendum.substack.com/feed/","description":"A publication by Tim O'Reilly and Ilan Strauss on AI commercialization risks and governance. Disclosure and protocols for healthy AI ecosystems.","favicon":"https://rogue-scholar.org/api/communities/0b81a230-f3a4-4f15-b6db-f262083c3797/logo","feed_format":null,"feed_url":"asimovaddendum","filter":null,"generator":"Substack","home_page_url":"https://asimovaddendum.substack.com/","issn":null,"language":"eng","license":"https://creativecommons.org/licenses/by/4.0/legalcode","prefix":"10.59350","relative_url":null,"secure":null,"slug":"asimovaddendum","status":"active","subfield":"1702","title":"Asimov's Addendum","updated":1783685589,"use_api":true},"blog_name":"Asimov's Addendum","blog_slug":"asimovaddendum","content_html":"This is the full version of a five minute \"lightning\" talk given at O'Reilly Media's Foo Camp on Saturday June 27, 2026 at Lighthaven, Berkeley \u2013 co-sponsored by the AI Disclosures Project.
Will or won't AI lead to a \"jobless apocalypse\"? I want to offer one framework for thinking through this question based on Adam Smith and Allyn A. Young \u2014 and Hal Varian on standards.
The question I want to ask you today is as follows: Where does intelligence reside? Where does it live?
The argument I want to make is that the artificial intelligence found in large language models (LLMs) is inert and centralized in the models' weights. This is a very different type of intelligence to that exhibited by markets, which is fundamentally decentralized and constantly being updated.1 Markets already extend and enable machine intelligence; but often sporadically and not always via market exchange. Integrating these two types of intelligence, including through the price mechanism, is vital if AI is to advance specialized production, market exchange, and ultimately employment creation.
To understand market forms of intelligence I want to begin by asking whether this picture below reflects a market-based intelligence. It shows women making men's suits using sewing machines. It looks organized. But according to what principles?
![\"File:RIAN](\"https://substackcdn.com/image/fetch/$s_!SWmy!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ff2cfc27c-125b-4f6d-b8d7-ee1ad94d7bcb_960x970.jpeg\" "\\"File:RIAN")
How do these workers know how many suits to make, what types to make, and with what machinery? By informed guesses of a central committee, it turns out.
The above picture is from Bolshevichka garment factory in Moscow, taken in 1967 when it was part of the Soviet Union. This factory used to produce the majority of men's suits for Russia. Although the Soviet Union excelled in heavy industry production, it ultimately failed in consumer goods production, where market signals were needed to inform changing tastes, quantities, efficiencies in resource utilization, and fundamental innovations. Waste was abundant. Human need could not be met. Production techniques were outdated.2
Bolshevichka still exists as a menswear brand and company today, selling men's suits, but now informed by market signals.
What would it mean if the above factory was instead organized on the basis of a market-based intelligence? This would require two things.
The first is decentralized and permissionless innovation. Markets are a federated system, where information remains dispersed along with the trillions of independent buying and selling actions that it enables. Market-processed information is what enables and disciplines these actions (along with society's conventions and laws).
The second is that current information signals inform market coordination through constantly shifting agents' incentives. Prices and product information are always being replenished as preferences and technology change, and demand and supply decisions respond accordingly. Market signals inform decision making but also respond to it. For this to work, price signals and other relevant information must be widely available (and not gameable).
It appears to me that digital markets have till now often tried to mimic market forms of intelligence. Google Search of old is an example of a market-like intelligence, I would propose. Google Search has:
Permissionless innovation (building on the web's open protocols). Anyone can create a website. Information (websites) remain federated on servers.
Dynamically updated signals. What the user clicks on (consumes) in the search result page is used to improve and constantly update website search result rankings. This in turn improves the next allocations of user attention to information, which in turn informs what information websites and other producers should supply to the market.
The algorithms of this era advanced what Tim O'Reilly calls \"an architecture of participation\".
But will markets create jobs? The truth about market-based systems of intelligence is that humans may not always be needed. Nissan's \"intelligent factory\" below is just one example of labor being made redundant in manufacturing production. But without markets expanding production on a progressive and balanced basis, little prospect for human employment and prosperity exist.
![\"Robots](\"https://substackcdn.com/image/fetch/$s_!nQYq!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ff8c45663-2daf-4070-ae06-35948e5befda_764x432.jpeg\" "\\"Robots")
How do market forms of intelligence compare with present day artificial intelligence in LLMs?
1) An LLM's intelligence is inert and unresponsive at its core. The model's weights are trained on information and feedback after which they become fixed, \"dead\" things. They are unable to dynamically update or learn as they go. That is why LLMs are reliant on markets for a steady stream of new information, and in particular markets online.
Besides for the internet, the other primary source of live information for models is human users and firms. Every time a user has a conversation with a generative AI system, uploads a document, or tells it something, they allow the model's weights to respond to live, up-to-date, information. Post-training also incorporates human judgement and information, organized by firms such as Surge AI. But direct market signals seem to only weakly inform these activities and the piece rates offered to human annotators.
2) The other aspect of present AI is that it aims to be generalized \u2013 as in artificial general intelligence. In doing so, LLM companies strive to be able to do every task that a human can do and at a high level of competency. This potentially undermines the idea of specialized producers existing across a market, each using different and specialized production techniques. The idea of a single generalized producer (\"AGI\") is so radical that it threatens to undermine the notion of markets altogether.
And that is the central contention of this talk: that specialization underpins the ability for a market's intelligence to grow, for permissionless innovation to flourish, and for price signals to function effectively.
Specialized production provides a bridge between market forms of intelligence and artificial intelligence. So if we want to grow the market for AI, as a system of permissionless innovation, with room for human contributions, it needs to permit \u2013 and admit \u2013 that specialized production is necessary and desirable. General intelligence divorced from market forms of intelligence cannot survive, except as a stale set of encyclopedias sitting on the shelf.
Specialized production \u2014 or specialization for short \u2014 is the ultimate basis for market forms of intelligence and its expansion (an idea we owe to Adam Smith). Specialization is what drives production for exchange as for-profit production coordinated via the market.
Allyn Young in 1928 went one step further when he argued that specialization in the market is a self-reinforcing dynamic. The more I specialize as a firm, the more I allow other firms to specialize too, as they can then meet more of their needs (procure more of their inputs) via the market, instead of through internal production. This means that specialization is not only the foundation of market exchange (Adam Smith), but can create an ever-expanding basis for the market itself to grow through increases in output (supply).
Specialization as external economies. The benefits of specialization are best viewed from the industry or ecosystem as a whole that it helps to expand. The early American auto industry is a good example of this. Ford and its rivals began as assemblers, drawing on Detroit's dense ecosystem of specialized machine shops and parts makers. The Dodge Brothers famously supplied Ford's engines and transmissions before ever building a car of their own. It was this external supplier base producing standardized components for a variety of buyers that allowed for mass auto production to arise and prices fall. This grew the size of the market for everyone. Ford later advanced vertically integrated production in its famous River Rouge complex (which Charlie Chaplin's 1936 film Modern Times drew on), until more distributed supplier networks gained favor again. The benefits of specialization, therefore, are about the unfolding division of labor between firms, not about a single firm scaling up its own production, which can potentially be at the exclusion of others.
Specialization and integrated / monolithic producers. Specialization undermines the integrated producer, where a single firm attempts to do everything under its own roof, without procuring standardized inputs via the market (or transferring its intermediate outputs only internally, rather than selling them).3
Advancing specialization online and in LLMs' own production process can have similar benefits. But this can only have a chance of succeeding if model building is no longer undertaken from start to finish in relative secrecy (housed within a single firm). And if information and intelligence can permanently reside outside of a model's weights. Model weights don't only have to consume markets, they can also enable them. Standards and open protocols, as the technical analogues of market-enabled specialized production, are two important mechanisms for advancing such an agenda.
Standards and open protocols are two key mechanisms for advancing market forms of intelligence online that are decentralized and permissionless. The goal is to allow for artificial intelligence to be extended into the living market of information, admitting formal prices and for market-based allocations.
Standards enable specialization by allowing interconnected parts to fit together. In markets with many complementary parts, standards are vital to allowing a single producer to address the whole differentiated market of firms as if it were a single source of demand. This helps the supplier to not be captured by a single producer (\"lock-in\"). As my colleague Tim O'Reilly remarked to me, one much-underappreciated ingredient to the industrial revolution's take-off was specialized inputs and parts being produced to a standard.
Such standards are especially important in networked markets since connecting to the network is vital to tap into its scale and capabilities. This means that it can potentially let average revenue per user grow for any firm with access to the network's expanding utility.
So in technology markets, standards not only facilitate lower costs through specialization / scale, they also inform whether the value of a network is captured by one firm or shared by many.
Open protocols are shared rules that allow independent users, agents, or devices to coordinate and communicate on the basis of a publicly available specification. They are a type of standard.
By hardening an interface at a given layer through an open specification they allow other firms to connect to it reliably and then specialize against it. They are a battering ram against the integrated producer and the monolithic architecture. A single, closed architecture is designed so that potentially separable complements cannot emerge as independent markets. Under a monolithic architecture specialized production is impeded because the interfaces between those functions simply don't exist or are controlled by a single firm. But a modular architecture is the technical enabler of specialized production that can be disassembled safely (an idea that traces back to Herbert Simon's 1962 paper, \"The Architecture of Complexity\".)
Once specialization takes hold at one interface via an open protocol, the integrated architecture of the AI lab or the platform can be challenged via the market as firms can now compete to supply or draw on discrete components (such as memory, retrieval, payments, identity, tool invocation, attribution, safety filters, and so on).
In doing so, an open protocol can drive vertical disintegration. Interfaces become standardized and what used to be one firm's internal operation gets spun out into a specialized industry as the protocol now handles the coordination (or the identification, the encryption, the security) that previously required all functions to be retained inside the same firm. That spinning-out of tasks (in economic terms) is precisely Allyn Young's progressive subdivision of production across specialized firms.
Although not usually described in market-shaping terms, open protocols, therefore, are exactly that. They shape the market intelligence that can be built on top of them.
The Model Context Protocol (MCP) comes to mind here. It helped catalyze specialized production in AI applications and a more permissionless innovation by allowing AI agents (clients) to connect to external tools and context. A large ecosystem of specialized servers, each exposing one or more tools and resources, grew up around it, which any compliant client can use. Such specialization in turn grew the market for other specialized producers. Specialization begot specialization and rival labs eventually adopted the standard.
![\"\"](\"https://substackcdn.com/image/fetch/$s_!7D8t!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F464a1994-62f5-4be2-b1d3-b07c8f476b98_1408x768.png\")
The dream of artificial intelligence being very \"general\" in nature (as in \"AGI\") seems, at the end of the day, to be technically compatible with a specialized division of labor for its production and with the existence of specialized AI producers. The market for AI startup financing seems to think so too, funding different approaches to AGI (and recursive self-improvement) and funding differentiated, highly specialized AI producers at the same time.
Extending machine intelligence into markets, and vice-versa, to allow for human participation and reward via markets (\"an architecture of participation\" for AI) is the focus of the three initiatives we are launching this year at the AI Disclosures Project:
Open Memory Protocol Consortium (OMPC). A multi-stakeholder nonprofit consortium developing open, interoperable specifications for portable AI memory. Membership-based, with open working groups, it provides early-stage technical incubation on specific aspects of agentic memory in specific markets.
The AI Markets Initiative. A cross-institution economics research and data collaboration that galvanizes the best economic minds around AI market design challenges and opportunities. The goal is to create functioning AI markets with architectures of participation.
Skills Marketplace. Launching a skills marketplace with monetization for skills authors based on modular, interoperable, rights-protected skills, with proven uplift to agentic performance. All specifications and code will be open-sourced, to be used as a prototype for others to adopt.
If you are interested in being involved in any of these three initiatives in whatever capacity, we would love to hear from you. Thank you for your support.
This does not preclude the role of firms as owner-producers, hierarchical organizers of capitalist production but still ultimately beholden to the market.
Notes Richard Ericson, in \"The Classical Soviet-Type Economy: Nature of the System and Implications for Reform,\" Journal of Economic Perspectives 5(4), 1991, pg.21: \"When the system pursues a few priority objectives, regardless of sacrifices or losses in lower priority areas, those ultimately responsible cannot know whether the success was worth achieving. The central authorities lack the information and physical capability to monitor all important costs\u2014in particular, true opportunity costs\u2014yet they are the only ones, given the logic of the system, with a true interest in knowing such costs.\"
And then concluding, pg.26: \"Finally, it must be remembered that the ultimate configuration of institutions and interactions is unknowable, a largely unintended consequence of the growth of decentralized agent interaction. Thus, a final lesson for successful reform taught by the nature of the traditional Soviet-style system is to abandon the Faustian urge to control, to know in advance, and thus to allow economic outcomes to arise naturally as the unpredictable consequences of market interaction.\"
So-called agentic orchestration acting on the basis of user memory and context is one attempt to recreate this division of labor arguably, with the user or firm's information providing the specialized ingredients into the produced service \u2013 services are produced as they are consumed. But another approach is simply specialized agents, which implies a more specialized production. The idea that agents lower transaction costs and so might naturally see more production internalized within the firm seems highly speculative. Coordination within a firm is often less efficient / more costly than via the market, noted by Stigler in his 1951 piece (\"The Division of Labor is Limited by the Extent of the Market\"). The boundaries of specialized production are likely in these instances to be decided more by the sensitivity and accessibility of data \u2013 and intellectual property more generally \u2014 along with the ability to have standardized parts and a large enough addressable market.
This article also appeared as a guest post on 1 July 2026 on the blog of the Brandenburg Open Access Networking and Competence Centre (https://open-access-brandenburg.de/gastbeitrag-wertesensible-ansaetze-fuer-die-open-science-beratung/)\n\n\n\n
Hantow, Jonas (2026): \"Guest post: Between FAIRness and responsibility: Value-sensitive approaches to open science consultancy.\" DOI: 10.59350/gepex-r1g49
\"As open as possible, as closed as necessary.\"
\n\n\n\nThis sentence is quoted so often in Open Science contexts that it can almost sound like a solution in itself. It is not. It is a very good starting point, though. But in everyday consulting practice, the fascinating part usually begins with the second half of the sentence: What exactly is necessary? Who decides? And necessary for whom?
\n\n\n\n![\"Website](\"https://bibliocon2026.abstractserver.com/program/img/logo.68727bd8.png\"){kind=logo}
At BiblioCon 2026 (May 19-22, 2026 Berlin), we explored these questions in our hands-on lab \"Between FAIRness and Responsibility: Value-sensitive approaches for Open Science consulting\". The workshop was aimed especially at people working in libraries, data services and research infrastructures who support researchers in publishing data, objects and sources. Our starting point was a tension that many consultants know very well: Open Science encourages us to make data findable, accessible, interoperable and reusable. At the same time, there are cases when full openness is not possible, not appropriate or simply ethically not responsible.
\n\n\n\nThis becomes particularly palpable in object-centred and data-intensive disciplines. A digitised museum object is not just a file. Archaeological data are not just coordinates. Historical sources are not neutral just because they are old. They may be connected to people, communities, cultural heritage, fragile sites, vulnerable species, institutional interests or contested histories.
\n\n\n\nThe FAIR principles helps us to think about technical and infrastructural quality. CARE, developed for Indigenous Data Governance, adds an important ethical and political layer: collective benefit, authority to control, responsibility and ethics. But CARE also has a specific origin and a specific addressee. It was designed for indigenous contexts. Many of its concerns can be conveyed into other contexts, but they should not simply be generalized as if all sensitive cases were the same.
\n\n\n\nThis is where Value Sensitive Design becomes a useful inspiration for us as an additional lens. For now, we propose an adaptation as \"Value Sensitive Data Publishing\". It neither replaces FAIR nor CARE. Rather, it helps to clarify what exactly is at stake in a given publication decision. Which values matter here? Which direct and indirect stakeholders are affected? Which conflicts between values become visible? And how can these conflicts be managed, i.e. translated into practical design decisions, workflows, metadata, licenses or access levels?
\n\n\n\nIn the workshop, we discussed this through three fictional but realistic cases.
\n\n\n\nThe first case dealt with fictional excavation data from G\u00f6beklitepe. Historical handwritten lists had been digitised, made full-text searchable with OCR and enriched with FAIR metadata. From an Open Science perspective, this sounds like a desired outcome. But what if the data contain precise coordinates of places of discovery? What if publication of those coordinates could enable looting? Or what if such data and archaeological interpretations shaped by specific Western epistemologies are used to eclipse local cultural narratives? How to handle scholarly claims and political claims? Here, values such as transparency, reproducibility and reusability meet the need to protect cultural heritage, site preservation and sources. \"Open\" may therefore mean: make metadata visible, document the dataset, contextualize, explain access conditions, but do not publish everything at full resolution and maximum precision by default.
\n\n\n\nThe second scenario focused on sacred objects from Latin America in a German museum. High-resolution 3D models were ready for publication, including information on dimensions, materials and historical descriptions. But one object contained a specific ochre tone that, according to the case description, should not be seen by women in the community of origin for religious reasons. This example shows very clearly that values are not limited to privacy or the protection of individual persons. Cultural goods can require protection. So can religious practices, community knowledge and relationships between institutions and source communities. How can the scholarly and wider public interest in access, availability, and reuse be balanced with culturally specific taboos in this case? And in natural history collections, we may have to think even further: information about nesting sites, for example of a protected bird species, can be sensitive because its availability would affect animals and their habitats. Even just a group of well-intended bird-watchers can pose a disruptive element for a natural setting. Value-sensitive publication therefore needs a broader concept beyond \"personal data\" and privacy alone. This is even more true, when deep text and data mining combined with automatic, AI based synoptic tools allow for broad triangulation of data resources.\u00a0
\n\n\n\nThe third case concerned digitised books from the colonial period of the German Empire. Here, Open Access and the question of proper contextualization can conflict. Should racist images, discriminatory terminology and imperial narratives be made openly available and even provided for possible reuse, as the application of a CC BY licence would entail? We think the more expedient question would be: Under which conditions can Open Access support critical engagement instead of reproducing harm? Contextualisation should not be understood as an obstacle to openness, but as one of its enabling conditions. Users also have an agency and responsibility to engage critically with the material. Infrastructures can help them do so: through usage notes, provenance information, explanatory metadata, essays, teaching materials or links to current research.
\n\n\n\nThis also changes how we think about licences. In many Open Access contexts, CC BY is the default reflex and demand. But when dealing with sensitive historical sources, objects or images, a more restrictive licence such as CC BY-ND may sometimes better support the goal of preventing decontextualised or misleading derivatives. This does not mean that CC BY-ND is always the answer. It means that licence choice is itself a value-sensitive decision.
\n\n\n\nOne important distinction emerged repeatedly: Open Access for research publications and Open Access for sources are not the same thing. A journal article, an interpreted dataset, a 3D scan of a sacred object, a colonial photograph and a precise locality record of a protected species all have different conditions of \u2013 ethical \u2013 openness. Treating them as if they followed the same logic makes consulting easier, but also leaves more space for potentially improper or even malfeasant reuse.
\n\n\n\nWhat did we take away from this session?
\n\n\n\nThat is perhaps the most important point. Context, restrictions and care are not the enemies of Open Science. In many cases, they are what makes Open Science possible.
\n\n\n\n![\"Google](\"https://substackcdn.com/image/fetch/$s_!EKwP!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa6716df6-0653-4f9e-bec7-876e50511cf4_1231x703.webp\" "\\"Google")
![here](\"https://en.wikipedia.org/wiki/Bolshevichka#/media/File:RIAN_archive_901609_Bolshevichka_garment_factory.jpg\"){kind=link}
BiblioCon 2026:\nConference website (in German): https://www.bib-info.de/fortbildung/bibliothekartage/bibliocon-2026-bib\nWorkshop-Session: https://bibliocon2026.abstractserver.com/program/#/details/sessions/151\nAbstract: https://bibliocon2026.abstractserver.com/program/#/details/presentations/145\n\n\nMore information on CC-BY-ND licenes:\nhttps://creativecommons.org/licenses/by-nd/4.0/deed.en\n\nWiNoDa offers a self-paced online course on Regulatory Frameworks for Data Acquisition in Fieldwork on our Moodle platform. Just register on www.moodle.winoda.de and enroll!\nDieser Artikel erschien am 1. Juli 2026 auch als Gastbeitrag im Blog der Vernetzungs-und Kompetenzstelle Open Access Brandenburg (https://open-access-brandenburg.de/gastbeitrag-wertesensible-ansaetze-fuer-die-open-science-beratung/)\n\n\n\n
Hantow, Jonas (2026): \"Gastbeitrag: Zwischen FAIRness und Verantwortung: Wertesensible Ans\u00e4tze f\u00fcr die Open-Science-Beratung.\" DOI: 10.59350/gepex-r1g49
\n\n\n\n\n\n\n\n\n\"So offen wie m\u00f6glich, so geschlossen wie n\u00f6tig.\"
\n
Dieser Satz wird in Open-Science-Kontexten so h\u00e4ufig zitiert, dass er fast selbst wie eine L\u00f6sung klingen kann. Das ist er aber nicht. Er ist jedoch ein sehr guter Ausgangspunkt. Die eigentliche Arbeit beginnt meistens im zweiten Teil des Satzes: Was genau ist n\u00f6tig? Wer entscheidet das? Und n\u00f6tig f\u00fcr wen?
\n\n\n\nAuf der BiblioCon 2026 (19.-22. Mai 2026, Berlin) diskutierten wir diese Fragen in unserem Hands-on-Lab \"Zwischen FAIRness und Verantwortung: Wertesensible Ans\u00e4tze f\u00fcr die Open-Science-Beratung\". Der Workshop richtete sich insbesondere an Personen aus Bibliotheken, Datenservices und Forschungsinfrastrukturen, die Forschende bei der Publikation von Daten, Objekten und Quellen beraten. Ausgangspunkt war ein aus der Beratungspraxis sehr vertrautes Spannungsfeld: Open Science fordert dazu auf, Forschungsdaten, -quellen und -materialien nahezu unbeschr\u00e4nkt auffindbar, zug\u00e4nglich, interoperabel und nachnutzbar zu machen (FAIR). Gleichzeitig gibt es F\u00e4lle, in denen vollst\u00e4ndige Offenheit nicht m\u00f6glich, nicht angemessen oder schlicht ethisch nicht verantwortbar ist.
\n\n\n\nBesonders deutlich wird das in objektzentrierten und datenintensiven Disziplinen. Ein digitalisiertes Museumsobjekt ist nicht einfach nur eine Datei. Arch\u00e4ologische Daten sind nicht einfach nur Koordinaten. Historische Quellen sind nicht neutral, nur weil sie alt sind. Sie k\u00f6nnen mit Personen, Communities, Kulturg\u00fctern, gef\u00e4hrdeten Fundorten, schutzbed\u00fcrftigen Arten, Interessen von Forschenden und Institutionen oder konfliktreichen Geschichten verbunden sein.
\n\n\n\nDie FAIR-Prinzipien helfen uns, \u00fcber technische und infrastrukturelle Qualit\u00e4t nachzudenken. CARE, entwickelt f\u00fcr Indigenous Data Governance, erg\u00e4nzt eine wichtige ethische und politische Perspektive: kollektiver Nutzen, Kontrolle und Verantwortung. Zugleich hat CARE einen spezifischen Ursprung und einen spezifischen Zuschnitt. Die Prinzipien beziehen sich auf indigene Kontexte. Viele der dort formulierten Anliegen sind auch f\u00fcr andere Zusammenh\u00e4nge relevant, sollten aber nicht verallgemeinert werden, als w\u00e4ren alle sensiblen F\u00e4lle gleich.
\n\n\n\nAn dieser Stelle kann der Ansatz des Value Sensitive Design als zus\u00e4tzliche Linse dienen, um die Bewertung sinnvoll zu erweitern. Im vorliegenden Zusammenhang sprechen wir dabei vom \"wertesensiblen Publizieren von Forschungsdaten\". Der Ansatz ersetzt weder FAIR noch CARE. Er hilft vielmehr, erg\u00e4nzend und genauer zu kl\u00e4ren, worum es aus ethischer Sicht bei einer konkreten Publikationsentscheidung geht. Welche Werte sind betroffen? Welche direkten und indirekten Stakeholder m\u00fcssen mitgedacht werden? Welche Wertekonflikte werden sichtbar? Und wie lassen sich diese Konflikte in praktische Entscheidungen zu Workflows, Metadaten, Lizenzen, Zugriffsstufen oder Infrastrukturen \u00fcbersetzen?
\n\n\n\nIm Workshop haben wir dies anhand von drei fiktiven, aber realit\u00e4tsnahen Anwendungsf\u00e4llen erprobt.
\n\n\n\nDer erste Fall behandelte fiktive Grabungsdaten aus G\u00f6beklitepe. Dabei w\u00fcrden historische handschriftliche Listen digitalisiert, mit OCR volltexterschlossen und mit FAIR-Metadaten angereichert werden. Aus Open-Science-Perspektive klingt das zun\u00e4chst absolut begr\u00fc\u00dfenswert. Aber was passiert, wenn die Daten pr\u00e4zise Koordinaten von konkreten Fundstellen enthalten? Was, wenn eine Ver\u00f6ffentlichung Pl\u00fcnderungen erleichtert? Was, wenn diese Daten und die daraus abgeleiteten arch\u00e4ologischen Interpretationen lokale kulturelle Narrative \u00fcberlagern? Wie lassen sich hier wissenschaftliche Erkenntnisziele und m\u00f6gliche politische Anspr\u00fcche austarieren? In solchen F\u00e4llen treffen Werte wie Transparenz, Reproduzierbarkeit und Nachnutzbarkeit auf den Schutz von Kulturg\u00fctern, den Erhalt von Fundorten und Schutz arch\u00e4ologischer Quellen. \"Open\" kann in diesem Fall bedeuten: Metadaten sichtbar machen, den Datensatz dokumentieren und kontextualisieren, Zugangsbedingungen abstufen, also entsprechend nicht alles von vornherein in maximaler Aufl\u00f6sung und Pr\u00e4zision ver\u00f6ffentlichen.
\n\n\n\nDer zweite Musterfall drehte sich um sakrale Objekte aus Lateinamerika in einem deutschen Museum. Hochaufl\u00f6sende 3D-Digitalisate st\u00fcnden zur Publikation bereit, einschlie\u00dflich Informationen zu Ma\u00dfen, Materialien und historischen Beschreibungen. Mindestens ein Objekt w\u00e4re jedoch in einem bestimmten Ockerton gefasst, der in der Herkunftsgesellschaft aus religi\u00f6sen Gr\u00fcnden von Frauen nicht gesehen werden soll. Dieses Beispiel zeigt sehr deutlich: Werte beziehen sich nicht nur auf Privatsph\u00e4re oder den Schutz einzelner Personen. Auch kulturelle Regeln k\u00f6nnen schutzbed\u00fcrftig sein. Gleiches gilt f\u00fcr religi\u00f6se Praktiken, Community-Wissen und Beziehungen zwischen Forschenden, Institutionen und Herkunftsgesellschaften. Wie lassen sich in diesem Fall die Anspr\u00fcche von Wissenschaft und vielleicht auch der Allgemeinheit an einen umfassenden offenen Zugang mit eventuellen kulturellen Tabus vereinbaren? Und in naturwissenschaftlichen Sammlungen m\u00fcssen wir vielleicht noch weiter denken: Informationen \u00fcber Brutpl\u00e4tze gesch\u00fctzter Vogelarten k\u00f6nnen sensibel sein, weil dadurch sensible Lebensr\u00e4ume betroffen sein k\u00f6nnen. Selbst sehr wohlmeinende Freizeitvogelkundler*innen k\u00f6nnten in so einem Setting zum gravierenden St\u00f6rfaktor werden. Eine wertesensible Publikation braucht daher eine konzeptionelle ethische Differenzierung, die \u00fcber \"personenbezogene Daten\" hinausreicht. Dies ist umso relevanter, wenn Text- und Data-Mining-Verfahren mit automatisierten und KI-basierten Zusammenfassungswerkzeugen weitreichende Triangulierungen von Datenquellen erm\u00f6glichen.\u00a0
\n\n\n\nDer dritte Musterfall behandelte digitalisierte B\u00fccher aus der Kolonialzeit des Deutschen Reichs. Hier wirkt es zun\u00e4chst so, als st\u00fcnden Open Access und notwendige Kontextualisierung einander gegen\u00fcber. Sollten rassistische Abbildungen, diskriminierende Begriffe und imperiale Narrative frei zug\u00e4nglich und sogar nachnutzbar gemacht werden, was eine CC-BY-Lizenzierung erm\u00f6glichen w\u00fcrde? Die hilfreichere Frage lautet aus unserer Sicht: Unter welchen Bedingungen erm\u00f6glicht ein offener Zugang eine kritische Auseinandersetzung, anstatt Schaden zu reproduzieren? Kontextualisierung sollte nicht als Hindernis f\u00fcr Offenheit verstanden werden, sondern als eine ihrer Erm\u00f6glichungsbedingungen. Nutzende erhalten so die M\u00f6glichkeit, aber auch die Verantwortung, sich kritisch mit dem Material auseinanderzusetzen. Infrastrukturen k\u00f6nnen sie dabei unterst\u00fctzen: durch Nutzungshinweise, Provenienzinformationen, erl\u00e4uternde Metadaten, Begleittexte, Lehrmaterialien oder Verweise auf aktuelle Forschung.
\n\n\n\nDas ver\u00e4ndert auch den Blick auf Lizenzen. In vielen Open-Access-Kontexten ist CC BY die scheinbar naheliegende und w\u00fcnschenswerte Wahl. Bei sensiblen historischen Quellen, Objekten oder Bildern kann eine restriktivere Lizenz wie CC BY-ND jedoch in bestimmten F\u00e4llen besser dazu beitragen, dekontextualisierte oder irref\u00fchrende Bearbeitungen zu verhindern. Das hei\u00dft nicht, dass CC BY-ND immer die richtige Antwort ist. Es hei\u00dft nur: Auch die Lizenzwahl ist eine wertesensible Entscheidung.
\n\n\n\nEine wichtige Unterscheidung tauchte in der Diskussion immer wieder auf: Open Access f\u00fcr Forschungsarbeiten und Open Access f\u00fcr Quellen sind nicht dasselbe. Ein Zeitschriftenartikel, ein interpretierter Datensatz, ein 3D-Scan eines sakralen Objekts, eine koloniale Fotografie und eine pr\u00e4zise Fundstellenangabe zu einer gesch\u00fctzten Art unterliegen unterschiedlichen Voraussetzungen von Offenheit. Sie so zu behandeln, als g\u00e4lte f\u00fcr alle dieselbe Logik, macht Beratung einfacher, aber nicht besser, sondern erh\u00f6ht das Potenzial f\u00fcr unangemessene oder sogar b\u00f6swillige Nachnutzungen.
\n\n\n\nWas nehmen wir aus der Veranstaltung mit?
\n\n\n\nGenau das ist der wichtigste Punkt. Kontext, Einschr\u00e4nkungen und Sorgfalt sind keine Gegner von Open Science. In vielen F\u00e4llen sind sie das, was Open Science \u00fcberhaupt erst m\u00f6glich macht.
\n\n\n\nWebseite der BiblioCon 2026:\nKonferenz: https://www.bib-info.de/fortbildung/bibliothekartage/bibliocon-2026-bib\nWorkshop-Session: https://bibliocon2026.abstractserver.com/program/#/details/sessions/151\nAbstract: https://bibliocon2026.abstractserver.com/program/#/details/presentations/145\n\nInformationen zur CC-BY-ND Lizenz:\n https://creativecommons.org/licenses/by-nd/4.0/deed.de\n\nWiNoDa bietet einen Selbstlernkurs \u00fcber Regulatory Frameworks for Data Acquisition in Fieldwork (auf Englisch) an. Einfach auf unserer Moodle Plattform www.moodle.winoda.de registrieren und einschreiben. (Gesch\u00e4tzte Zeit: ca. 1 Stunde)![](\"https://medical-education.pages.ub.uni-bielefeld.de/research/mes-blog/posts/2026-07-09-daytime-performance/Bueffel_muede.png\"/)
\"Der fr\u00fche Studi f\u00e4ngt die Punkte \u2026 !?\" Optimierung von Erm\u00fcdung und Passung f\u00fcr die Pr\u00fcfungszeit
\nZwei Kommilitoninnen nach einer 8-Uhr-Klausur. Die eine kommt heraus und sagt: \"Perfekt, ich war hellwach.\" Die andere hat sich durch die erste Stunde gequ\u00e4lt und ist \u00fcberzeugt, dass sie am Nachmittag das Doppelte gewusst h\u00e4tte. Beide haben denselben Stoff gelernt, beide sind gut vorbereitet \u2014 und trotzdem f\u00fchlten sie sich unterschiedlich fit. Ist das nur ein Gef\u00fchl, mit dem man sich die Note sch\u00f6nredet? Oder gibt es die eine, biologisch richtige Pr\u00fcfungszeit, die manche von uns systematisch bevorzugt?
\nDie kurze Antwort vorweg: Die Tageszeit wirkt tats\u00e4chlich messbar auf Pr\u00fcfungsleistung \u2014 aber der Effekt ist kleiner, als das Bauchgef\u00fchl vermutet, und er sieht f\u00fcr \"Lerchen\" und \"Eulen\" unterschiedlich aus.
\nDass Menschen zu verschiedenen Zeiten leistungsf\u00e4hig sind, ist keine Ausrede fauler Langschl\u00e4fer. Der Chronotyp \u2014 die Neigung, eher fr\u00fch (\"Lerche\") oder eher sp\u00e4t (\"Eule\") aktiv zu sein \u2014 verteilt sich in der Bev\u00f6lkerung breit und stabil, von extremen Fr\u00fchtypen bis zu extremen Sp\u00e4ttypen (Roenneberg et al., 2003). Ein sp\u00fcrbarer Teil dieser Unterschiede ist genetisch mitbestimmt: Gro\u00dfe Assoziationsstudien haben mehrere Gene identifiziert, die reproduzierbar mit dem Chronotyp zusammenh\u00e4ngen (Kalmbach et al., 2017).
\nF\u00fcr Euch als Studierende kommt eine biografische T\u00fccke dazu. Im sp\u00e4ten Jugend- und jungen Erwachsenenalter verschiebt sich der innere Fahrplan nach hinten \u2014 ausgerechnet dann, wenn Unterricht, Vorlesungen und Klausuren fr\u00fch beginnen. Diese Schere zwischen biologischer und sozialer Zeit tr\u00e4gt sogar einen eigenen Namen: sozialer Jetlag. Wer unter der Woche gegen die eigene innere Uhr aufsteht, sammelt ein Schlafdefizit an und kompensiert es am Wochenende (Wittmann et al., 2006). Die 8-Uhr-Klausur trifft also nicht alle gleich hart.
\nWenn man die Forschung ordnet, wirken zwei verschiedene Mechanismen \u2014 und sie werden in Alltagsdebatten gern verwechselt.
\nErstens: \u00dcber den Tag erm\u00fcden fast alle. Die sauberste Feldevidenz stammt aus D\u00e4nemark, wo Forschende Millionen standardisierter Testergebnisse auswerten konnten. Der Testzeitpunkt hing dort vom Stundenplan und der Computerverf\u00fcgbarkeit ab, war also nicht von den Sch\u00fclerinnen w\u00e4hlbar. Ergebnis: Pro Stunde sp\u00e4ter am Tag sank die Leistung \u2014 im Sinne eines kleinen, aber verl\u00e4sslichen Abw\u00e4rtstrends, den die Autoren als kumulierende kognitive Erm\u00fcdung deuten (Sievertsen et al., 2016). Auch in gro\u00dfen US-Schuldaten schnitten Lernende im Fach Mathematik am Vormittag besser ab als am Nachmittag (Pope, 2016). Dieser Effekt betrifft im Prinzip jeden, unabh\u00e4ngig vom Chronotyp.
\nZweitens: Es z\u00e4hlt die Passung zwischen innerer Uhr und Pr\u00fcfungszeit. Hier wird es individuell. Fachleute nennen das den Synchronie-Effekt: Wir rufen unsere beste Leistung ab, wenn der Pr\u00fcfungszeitpunkt zur inneren Uhr passt. Den bislang \u00fcberzeugendsten Beleg lieferte eine argentinische Studie, in der 753 Jugendliche per Los einer Morgen-, Nachmittags- oder Abendschicht zugeteilt wurden (eine seltene echte Randomisierung im Bildungskontext!). In der Morgenschicht schnitten Fr\u00fchtypen deutlich besser ab als Sp\u00e4ttypen, am st\u00e4rksten in Mathematik. In der Nachmittagsschicht verschwand dieser Vorteil, und Sp\u00e4ttypen profitierten sogar vom Abendunterricht (Goldin et al., 2020). Genau das erkl\u00e4rt, warum manche morgens gl\u00e4nzen und andere nicht: Es gibt keine f\u00fcr alle beste Uhrzeit, sondern f\u00fcr jeden ein eigenes Zeitfenster.
\nDass die Tageszeit dabei nicht nur \"Wachheit\" ist, zeigen kognitionspsychologische \u00dcbersichten: Aufmerksamkeit, Arbeitsged\u00e4chtnis und exekutive Funktionen schwanken \u00fcber den Tag hinweg systematisch (Valdez et al., 2014). Und Chronotyp-Reviews zeigen ein Muster, das genau dazu passt: Der Zusammenhang zwischen Chronotyp und Noten ist morgens am st\u00e4rksten und l\u00f6st sich nachmittags auf (Zerbini & Merrow, 2017). Das ist die Handschrift des Synchronie-Effekts \u2014 morgens liegen Fr\u00fchtypen im Takt ihrer inneren Uhr und Sp\u00e4ttypen nicht, sodass sich die Leistungen dann am deutlichsten unterscheiden.
\nJetzt kommt eine Einordnung, die in Pr\u00fcfungsdebatten leider meist fehlt. \u00dcber alle Studierenden gemittelt sind die Zusammenh\u00e4nge ziemlich klein. Meta-Analysen finden f\u00fcr die Verbindung von Abendorientierung und schlechteren Noten ziemlich kleine Korrelationen (Preckel et al., 2011; Tonetti et al., 2015). Das hei\u00dft: Der Chronotyp erkl\u00e4rt nur einen winzigen Bruchteil dessen, was Eure Note ausmacht. Vorbereitung, Vorwissen, Motivation und Schlaf wiegen um ein Vielfaches schwerer.
\nKlein hei\u00dft aber nicht bedeutungslos. Ein kleiner, systematischer Nachteil kann an einer Bestehensgrenze oder bei einer knappen Auswahlentscheidung den Ausschlag geben \u2014 und weil er systematisch mit der Pr\u00fcfungszeit zusammenh\u00e4ngt, ist er eine Fairnessfrage und nicht nur Pech. F\u00fcr Euch pers\u00f6nlich lohnt es sich also, die Tageszeit als einen von vielen Stellhebeln zu sehen, nicht als Schicksal. Falls Ihr also die Wahl habt, sucht eine f\u00fcr Euren Chronotyp gut passende Pr\u00fcfungszeit aus.
\nWenn Passung so wichtig ist, warum legt man Pr\u00fcfungen dann nicht einfach in das jeweils optimale Zeitfenster? Weil das f\u00fcr eine ganze Kohorte unm\u00f6glich ist. In demselben H\u00f6rsaal sitzen Lerchen und Eulen nebeneinander. Ein Termin, der den einen entgegenkommt, benachteiligt die anderen. Ein pauschal sp\u00e4terer Pr\u00fcfungsbeginn hilft eben nicht allen \u2014 Sp\u00e4ttypen profitieren, Fr\u00fchtypen verlieren. Dazu kommen Raumkapazit\u00e4ten, Aufsichten, Fristen und die Vergleichbarkeit \u00fcber Pr\u00fcfungstermine hinweg.
\nWas strukturell trotzdem geht, ist gut belegt. Pausen sind der billigste Hebel \u00fcberhaupt: Schon 20 bis 30 Minuten d\u00e4mpfen den Erm\u00fcdungsabfall sp\u00fcrbar (Sievertsen et al., 2016). Sp\u00e4tere Startzeiten verl\u00e4ngern nachweislich den Schlaf von Jugendlichen und gehen tendenziell mit besserem Befinden einher (Alfonsi et al., 2020; Marx et al., 2017). Und Institutionen k\u00f6nnen die Pr\u00fcfungszeit als das behandeln, was sie ist \u2014 eine systematische, dokumentierbare Einflussgr\u00f6\u00dfe, die man bei Auswertungen zumindest im Blick behalten sollte.
\nRealistisch bleibt f\u00fcr Euch ein schmaler, aber nutzbarer Spielraum. Wer bei computerbasierten Pr\u00fcfungen einen Termin w\u00e4hlen darf, kann ihn grob am eigenen Chronotyp ausrichten. Wichtiger als die Uhrzeit ist aber fast immer der Schlaf davor: Ausreichend und regelm\u00e4\u00dfig zu schlafen verschiebt mehr als jede Feinjustierung des Pr\u00fcfungsslots (Nguyen et al., 2025). Und in langen Pr\u00fcfungen (wie zum Beispiel Staatsexamina) sind bewusst genutzte Pausen kein Zeitverlust, sondern messbar investierte Leistung.
\nSo gut die Schul- und Laborevidenz ist \u2014 f\u00fcr den Kern des Medizinstudiums klafft eine L\u00fccke. Direkte Studien dazu, wie sich die Tageszeit auf gro\u00dfe schriftliche Staats- oder Fachexamina auswirkt, fehlen weitgehend. Die belastbaren Zahlen stammen aus Schulkontexten und aus dem Labor. Hinzu kommt, dass sich der Synchronie-Effekt bei jungen Erwachsenen nicht in jeder Studie robust zeigt und in kontrollierten Wiederholungen teils ausbleibt. Und in den gro\u00dfen Felddaten lassen sich Erm\u00fcdung und innere Uhr nur schwer sauber trennen, weil Schlaf, Ern\u00e4hrung und Fachschwierigkeit mit hineinspielen. Wer aus einem kleinen Mittelwertunterschied eine feste Regel f\u00fcr die eigene Pr\u00fcfung ableitet, \u00fcberzieht eventuell die Evidenz.
\n@misc{friederichs2026,\n author = {Friederichs, Hendrik},\n title = {\"Der fr\u00fche Studi f\u00e4ngt die Punkte ...\\,!?\" -\\/- gibt es eine\n optimale Pr\u00fcfungszeit?},\n date = {2026-07-09},\n url = {https://medical-education.pages.ub.uni-bielefeld.de/research/mes-blog/posts/2026-07-09-daytime-performance/},\n langid = {de}\n}\nI sometimes feel like we're living in a spaghetti western when I think about the administrative reforms to the Executive Branch done by the current Administration. The governance landscape has been transformed into a sun-baked, unpredictable frontier where decades of established institutional norms are tossed aside to make way for the new guy in town who is invariably irascible, cynical, self\u2011interested, and governing a corrupt world. So, forgive me for invoking the famous title of Sergio Leone's flick in responding to the Office of Management and Budget's Proposed Regulation for Federal Financial Assistance.
\nI've summarized my response here using this (perhaps regrettable) device. I encourage you to scroll to the bottom to read my detailed response and analysis as submitted to OMB via regulations.gov. You can submit your own comment by the close of business tomorrow here.
\nWait\u2026the good? Yes, in my opinion there are some non-objectionable revisions to the Uniform Guidance in the proposed rulemaking. These are relatively few compared to the terrible bits, but are still worth highlighting:
\nMost of the proposal is very bad. While we've heard a lot in the press about the objections from the scientific community - there are implications that are terrible that go beyond that as well. Here are a few of the parts that I found particularly objectionable and are relevant to my interests:
\nBeyond being bad, there are several components that are ugly. These are provisions that OMB should be embarrassed by - and would have been during my tenure at the agency.
\nChristopher Steven Marcum, PhD
\nOpen Science Advocate
\n7/12/2026
I am writing as a concerned citizen and open science advocate. I was previously a Senior Statistician and Senior Scientist at OMB, had privileged access to earlier drafts of this revision, and worked on the 2024 revision to the Uniform Guidance. \nIn general, while there are a few non-objectionable revisions in this proposed rulemaking (see Section II below for those), it is on balance a terrible revision and I urge OMB not to finalize the rule.
\nThe preamble to the proposed rule adopts an ideological tone that departs from the historically objective, neutral character of administrative modifications to the Uniform Guidance. By expressly incorporating policy frameworks from politically biased external entities, such as the Heritage Foundation, the Office of Management and Budget (OMB) introduces political polarization into a framework governing over $1.1 trillion in federal assistance. It also departs from OMB's traditional evergreen stance on 2 CFR 200 and it should remain insulated from partisan influences to maintain broad public trust in federal funding.
\nOMB's Office of Information and Regulatory Affairs has designated this proposed rule as not economically significant under it's EO 12866 review. Not only was the economic analysis embedded into the regulatory impact analysis (RIA) inadequate, its conclusions that lead to a non-economic significance determination are wildly misaligned with reality.
\nThis proposed rule exposes over $1 trillion in annual federal spending to potential political influence, clawback, and civic lawsuits. The natural outgrowth of such practices will necessarily have trillions in derivative effects that span every sector of society (jobs lost, projects ended mid-development, etc). The failure to conduct a rigorous, data-driven regulatory impact analysis creates severe fiscal uncertainty for states, municipalities, small businesses, and universities. This lack of transparency was explicitly cited by Senator Susan Collins in her formal request for a full delay and withdrawal of these burdensome mandates. Potential economic impacts were also cited by Moody's in their own determination that the proposed rule would likely affect the credit ratings of federal grant recipients. Clearly, whomever conducted the economic analysis component of the RIA did not consider these macro effects. It's a major oversight and a process foul by OMB that could have consequences with respect to non-compliance with the Administrative Procedures Act because OMB did not follow its own policies for conducting a cost-benefit analysis under Circular A-4.
\nI urge OMB to revisit this inadequacy and resubmit the rule with a new regulatory impact analysis that fully considers the underlying macro economics of the proposed rule.
\nThe proposed revisions to Subpart E fail to clarify whether changes to allowable expenses would restrict or eliminate the capacity of university libraries to utilize indirect cost allocations for research journal subscriptions and institutional transformative agreements. As highlighted by the Association of Research Libraries), defunding journal subscriptions within the indirect cost pool would cause severe structural barriers to knowledge access across higher education. Clear and explicit language must be added to confirm that institutional indirect funding remains fully permissible for scholarly communications and institutional database licensing.
\nThe proposed rule introduces restrictions that would preclude federal award recipients from flowing funds to external entities categorized as advocacy groups via subcontracts. In practice, many highly specialized technical consultancies, scientific societies, and civic organizations maintain separate, lawful public advocacy operations and are supported by federal assistance through subcontracts.
\nThe proposed section mandates that a senior political appointee at each federal agency must review and approve every discretionary award prior to issuance. This requirement formally subordinates technical, merit-based peer review to political alignment with presidential policy priorities. As noted by the American Association for the Advancement of Science (AAAS) and the American Public Health Association (APHA), inserting a political layer into the award selection process fundamentally disrupts the objective insulation of scientific and biomedical research from changing political whims. This provision is an egregious assault on scientific integrity for research grantmaking. Moreover, it exposes non-scientific financial assistance (the majority of assistance exposed to 2 CFR 200) to such whims and injects political uncertainty into the risk equations of organizations that rely on federal grants for their programmatic work.
\nThe expansion of agency authority to unilaterally terminate active awards for administrative convenience or shifting national interest creates extreme operational instability. DOGE used this extensively in 2025 to cancel government contracts, sometimes outside the existing terms and conditions contracts between the government and its awardees. It's a terrible idea to codify this rash and irresponsible behavior in reg.
\nScientific and public infrastructure programs are inherently long term, often spanning multiple presidential administrations. Unilaterally stopping active projects because of shifting political priorities results in substantial waste of taxpayer funds, as partially completed research or infrastructure yields minimal, if any, return-on-investment. More broadly, this instability weakens multi-year funding commitments to small businesses and state and local governments, potentially creating a credit negative environment for heavily dependent entities as analyzed by Moody's and reported on by Bond Buyer).
\nThis provision seeks to suppress foreign collaborations in research and development historically limited by the Wolf Amendment to NASA's appropriations from reaching China. Imposing absolute prohibitions on international scientific co-authorship and academic data exchange without clear, targeted statutory directives creates an unnecessary bureaucratic burden. This blanket expansion threatens to suppress international scientific collaboration, which is essential for addressing global challenges in public health, physical, and environmental science.
\nThe proposal disallows Article Processing Charges (APCs) and open access publishing fees as direct allowable expenses. While alternative publishing mechanisms such as preprints and the systematic exercise of the federal purpose license as codified in 2 CFR 200.315 by granting agencies provide a more sustainable framework for public access, an immediate, uncoordinated ban on APC direct charges could have unintended consequences (as I've previously discussed in detail elsewhere).
\nI urge OMB to revise this section to clarify an alternative to traditional scholarly publishing as the mechanism by which research results and data are shared by grantees: including directing federal agencies to assert their federal purpose license over all content supported by their grants and to allow for preprints to satisfy public access requirements.
\nThe proposed section contains provisions that would allow the Department of Justice to formally join or facilitate private, third-party civil society lawsuits against federally funded projects and award recipients for alleged grant noncompliance. This mechanism creates a highly weaponized oversight structure, exposing public entities, universities, and local governments to costly, politically motivated litigation. It is an unprecedented overreach of authority and further politicizes federal grantmaking in a way that will disrupt the continuity of programs benefiting from federal assistance across the timelines of one Administration to another.
\nI support OMB's reclassification of 2 CFR Part 200 from administrative guidance to a regulation. I previously advocated for this move during the last revision process in 2024 when I was at OMB, as it is onerous for the public to be exposed to dozens of deviations in grantmaking policies across the government. As Cole Donovan and I wrote earlier this year, interagency coordination efforts often fail because of decentralized, competing interests across the executive branch. This shift eliminates the administrative fragmentation that occurs when individual federal agencies interpret and implement grantmaking policies inconsistently, creating a single, predictable framework for grantees.
\nAlthough not explicitly articulated as such, the proposed rule introduces data governance language that closely aligns with the GREAT Act, the Open Government Data Act and the related OMB guidance M-25-05, as well as the 2022 OSTP Public Access Policy. I encourage OMB to further clarify that all data, code, and publications produced under a federal award that are acquired by a federal agency should be made immediately publicly accessible and inventoried appropriately consistent with the requirements of M-25-05.
\nThe proposed text correctly maintains and reaffirms the long-standing federal purpose license framework in 2 CFR 200.315. This provision ensures that the federal government retains a royalty-free, nonexclusive, and irrevocable license to reproduce, publish, or otherwise use works developed under a federal award for public purposes. Retaining this language provides a strong, legally established alternative to expensive commercial publishing models and safeguards public access to technical data. OMB should strengthen this by requiring all agencies bounded by the regulation to explicitly assert this license over all data, manuscripts, and code (and other intangible property) resulting from a federal award.
\nThe rule provides a refined and updated definition of Personally Identifiable Information (PII). This is a welcomed change and one that I advocated for during the previous revision to 2 CFR 200 in 2024. This update aligns the definition of PII with other OMB and agency guidance and regulations. It was always mis-defined in 2 CFR 200.
\nThe proposed updates introduce more robust reporting obligations and transparency standards for pass-through entities managing federal funds. There is a need for consistent, systematic transparency regarding how subawards are distributed, monitored, and accounted for on centralized platforms like SAM.gov. This proposal improves financial stewardship of taxpayer monies. It should reduce the risk of waste, fraud, and abuse in federal assistance programs.
\nThe proposed text also offers improved guidance to distinguish whether an external entity is functioning as a contractor or a subrecipient under a federal award. This change is needed because previous guidance was confusing and frequently resulted in misaligned audit findings and inconsistent compliance monitoring.
\nIn addition to the PII definition clarification, I support the clarification that confidential business information (CBI) is protected and cannot be compelled to be released without a legal justification. CBI is one aspect of confidential information that statistical agencies collect and must protect under the Confidential Information Protection and Statistical Efficiency Act that relies on a class E-felony for unlawful disclosure as a deterrent. This clarification helps align the obligations of trust imbued on the statistical agencies with the expectations of grantmaking agencies to protect CBI.
\nHowever, I urge OMB to clarify that both unawarded and awarded proposals and grant applications are NOT considered CBI. The public should have full awareness of what entities are proposing to do with taxpayer monies. Stuart Buck and I wrote about the importance of this for government transparency purposes last year.
","doi":"https://doi.org/10.59350/gpk1d-kvn12","guid":"https://www.chrismarcum.com/marcum-blog/2026/07/12/My%20Response%20to%20Uniform%20Guidance%20Rulemaking","language":"en","license":"https://creativecommons.org/licenses/by/4.0/legalcode","published_at":1783814400,"rid":"x75xm-mj444","summary":"I sometimes feel like we're living in a spaghetti western when I think about the administrative reforms to the Executive Branch done by the current Administration. The governance landscape has been transformed into a sun-baked, unpredictable frontier where decades of established institutional norms are tossed aside to make way for the new guy in town who is invariably irascible, cynical, self\u2011interested, and governing a corrupt world.","tags":["General","Government"],"title":"My Response to OMB's Proposed Regulation for Federal Financial Assistance","updated_at":1783872436,"url":"https://www.chrismarcum.com/marcum-blog/2026/07/12/My-Response-to-Uniform-Guidance-Rulemaking.html","version":"v1"},{"authors":[{"affiliation":[{"id":"https://ror.org/02mb95055","name":"Birkbeck, University of London"}],"contributor_roles":[],"family":"Eve","given":"Martin Paul","url":"https://orcid.org/0000-0002-5589-8511"}],"blog":{"authors":[{"name":"Martin Paul Eve","url":"https://orcid.org/0000-0002-5589-8511"}],"community_id":"9224b0d7-fc03-497c-9c6f-85c9fd1e72da","created":1690329600,"current_feed_url":null,"description":null,"favicon":"https://rogue-scholar.org/api/communities/9224b0d7-fc03-497c-9c6f-85c9fd1e72da/logo","feed_format":"application/atom+xml","feed_url":"https://eve.gd/feed_all.xml","filter":null,"generator":"Jekyll","home_page_url":"https://eve.gd","issn":null,"language":"eng","license":"https://creativecommons.org/licenses/by/4.0/legalcode","prefix":"10.59348","relative_url":null,"secure":true,"slug":"eve","status":"active","subfield":"1208","title":"Martin Paul Eve","updated":1783867107,"use_api":true},"blog_name":"Martin Paul Eve","blog_slug":"eve","content_html":"After a recent post by Brennan Kenneth Brown, I decided to investigate what it would take to get this site \u2013 and, then, Knowledge Commons \u2013 onto the ATmosphere, using ATProto. This was a pretty steep learning curve, as I did not, before, understand many of the key concepts in this space. After succeeding, though, in publishing this blog to the ATmosphere (I took Brennan's excellent advice to proceed immediately to Sequoia) I am left pondering.
\nFirst, what is this ATmosphere thing? Well, it's the protocol behind Bluesky and, basically, it provides a decentralized storage system for objects of any type. Bluesky posts have a type of \"app.bsky.feed.post\" while my blog posts are \"site.standard.document\". Various new platforms are emerging that can read and write to the ATmosphere, so it gives a new discovery layer (see e.g. this search), but also, in theory, a decoupled presentation layer; so how the site appears will be in the hands of the receiver, rather than determined by the writer.
\nNow, though, let's see what I have achieved\u2026 If you visit a recent post, you may notice that there are now two link tags in the document head:
\n<link rel=\"site.standard.document\" href=\"at://did:plc:hnpt7ns2lecdujegbi6qkqqm/site.standard.document/3mq7lupu73d2i\">\n<link rel=\"site.standard.publication\" href=\"at://did:plc:hnpt7ns2lecdujegbi6qkqqm/site.standard.publication/3mq4tnwmt3w2i\">\nThese links tie the page to its ATmosphere equivalent. Now, you can't just browse the ATmosphere in a web browser. The content has to have an interface written in front of it, that pulls the content in from this protocol and displays it. However, no fear, because there's an experimental browser for it! So, if you visit the corresponding file, you can, indeed, confirm that I have been successful in getting my content in.
\nSo, interestingly, there are no current standard.site or standard.document aggregators that display the document in their own interface (at least that I could see). Everywhere I tried linked back to the canonical web reference. But, interestingly, they WERE able to find the content and pull it into their aggregation streams.
\nAnother advantage is that, if my current PDS (where the data is stored \u2013 a Personal Data Server \u2013 which is actually on Bluesky's servers, as they run my current storage location) decides to shut down or move, and I no longer wish to be on their system, I can move the files and everything will apparently just be nicely migrated, without needing permission from the old PDS. Well, that sounds good, but I haven't tried it.
\nHowever, the entire reason I did this experiment was to see whether this was something worth pursuing for Knowledge Commons.
\n\u2026 and, for now, I am not sure. At the scale that we operate over there, it would take some effort to make sure we were doing it right. I also think that we would need to run our own PDS. On my personal blog, I hit the daily rate limit when I tried to upload 994 blog posts and associated file blobs (cover images etc.)\u2026 and then tried to modify every single one of them. I was locked out of Bluesky's PDS for 24hrs, which was annoying. However, KC has many, many more blog posts than this that would require initial loading. And if we ever had to do a full forced update of all content (say the canonical URL changed - gasp!), the process would need batching, it would be a pain, etc. But running our own PDS is also a pain. It's a whole new infrastructure that needs tending, opening new security holes, exposing us to the outside world\u2026 and that could go wrong.
\nHence, my conclusion is that, for now, I think we should watch and wait. If there is really substantial traction in this place \u2013 or if lots of people ask for it \u2013 I will certainly reconsider. But I would feel some trepidation about boldly going into this space, right now.
\nInvestigating ATProto here and for Knowledge Commons was originally published by Martin Paul Eve at eve.gd: Martin Paul Eve on July 10, 2026.
","doi":"https://doi.org/10.59348/r3x69-k5d70","guid":"https://doi.org/10.59348/r3x69-k5d70","language":"en","license":"https://creativecommons.org/licenses/by/4.0/legalcode","published_at":1783641600,"rid":"5jfay-xjh33","summary":"After a recent post by Brennan Kenneth Brown, I decided to investigate what it would take to get this site \u2013 and, then, Knowledge Commons \u2013 onto the ATmosphere, using ATProto. This was a pretty steep learning curve, as I did not, before, understand many of the key concepts in this space. After succeeding, though, in publishing this blog to the ATmosphere (I took Brennan's excellent advice to proceed immediately to Sequoia) I am left pondering.","title":"Investigating ATProto here and for Knowledge Commons","updated_at":1783867489,"url":"https://eve.gd/2026/07/10/investigating-atproto-here-and-for-knowledge-commons/","version":"v1"},{"authors":[{"affiliation":[{"id":"https://ror.org/02twcfp32","name":"Crossref"}],"contributor_roles":[],"family":"Feeney","given":"Patricia","url":"https://orcid.org/0000-0002-4011-3590"}],"blog":{"authors":[{"name":"Crossref Staff"}],"community_id":"093ada45-3a02-4007-b8b6-be28f221e01d","created":1780876800,"current_feed_url":null,"description":"Recent content in Blog on Crossref","favicon":"https://rogue-scholar.org/api/communities/093ada45-3a02-4007-b8b6-be28f221e01d/logo","feed_format":"application/atom+xml","feed_url":"https://www.crossref.org/blog/feed.xml","filter":null,"generator":"Hugo","home_page_url":"https://www.crossref.org/blog/","issn":null,"language":"eng","license":"https://creativecommons.org/licenses/by/4.0/legalcode","prefix":"10.64000","relative_url":null,"secure":null,"slug":"crossref","status":"active","subfield":"1710","title":"Crossref Blog","updated":1783604547,"use_api":null},"blog_name":"Crossref Blog","blog_slug":"crossref","content_html":"Research is rarely limited to a single contributor performing a single role. Behind every research output are people contributing in various ways: software development, data analyses, methodology design, and much more. Often, the same person contributes in several of these ways. Until now, Crossref metadata could only capture part of that picture, and this is changing now.
\nCrossref Schema 5.5 includes several improvements across different content types, but its most significant enhancement is the expanded support for contributor roles through the introduction of multiple roles per contributor, option to specify the corresponding author, and compatibility with the CRediT (Contributor Roles Taxonomy): a community-owned taxonomy of 14 contributor roles, which has been adopted and made available in multiple languages.
\nThese enhancements allow members to describe research contributions in much greater detail, creating richer metadata that better reflects how research is actually produced, and supporting greater accountability and more comprehensive research assessment.
\nIf your workflow already distinguishes between different kinds of contributions, Schema 5.5 gives you a way to record that detail more accurately using the CRediT taxonomy values. CRediT can be adopted gradually, where it fits your editorial or production workflow.
\nFigure 1: Until now, contributors could be assigned a single contributor role using Crossref's existing contributor role vocabulary. In Schema 5.5, members can indicate that the same contributor was responsible for different roles, such as corresponding author; writing: reviewing and editing; and data curation.
\nExisting deposits remain fully supported, and members can continue using the current contributor role attribute while planning implementation of the new repeatable role type element. For our members, who have been using CRediT in their workflows already, as ever \u2013 we encourage updating your metadata when practicable.
\nThis update gives more accurate credit to all of the people behind research outputs. Crossref vocabulary includes roles that aren't recognised in CRediT, and vice versa. Capturing richer contributor metadata recognises contributions that may not be visible in a single author line and improves transparency around how research is produced, thereby enabling downstream systems to interpret that information more reliably. The update also offers better interoperability with CRediT, which is well recognised across the scholarly ecosystem.
\nFigure 2: Schema 5.5 is an expansion of Crossref contributor metadata. Members can describe contributors using Crossref's existing contributor role vocabulary, as well as the internationally recognised CRediT taxonomy.
\nIn turn, this strengthens metadata reuse across repositories, discovery services, funders, institutions and other infrastructure providers; and supports evaluation, reporting and discovery workflows. Better contributor metadata strengthens the connections that make up the Research Nexus.
\nBeyond the expanded contributor support, Schema 5.5 includes several additional enhancements across the metadata schema.
\nSupport has been added for metadata elements that were previously missing from report series records, including Crossmark, funding, and licence information.
\nPosted content includes preprints, eprints, and other types of content that have been posted to a stewarded host platform. We're all about persistence, so it's vital that everything registered with us be maintained. Note that accepted manuscripts are not considered posted content. Schema 5.5 refreshes posted content sub-types by introducing blog and poster.
\nAt the same time, we are \"retiring\" working paper, dissertation, and report from posted-content sub-types. Over time, these have been developed into separate record types that benefit from richer, dedicated schemas.
\nFinally, archive locations can now also be included for posted content records.
\nA new archive location, CINES, has been added to the list of supported archive providers.
\nClinical trial information is no longer limited to journal articles and conference papers. Schema 5.5 extends support across additional content types, including books, datasets, dissertations, reports, posted content, standards, and pending publications.
\nTaken together, the updates in our latest schema support more holistic recognition of contributions to the research and its communication, as well as greater accountability and integrity in related processes.
\nTo support gradual adoption, Schema 5.5 maintains backwards compatibility with existing deposits. Members can continue using the current contributor_role attribute while preparing to implement the new repeatable role element. We have prepared a migration guide to help members transition to Schema 5.5.
As you prepare to adopt Schema 5.5, we encourage members to include contributor roles whenever they are available from editorial workflows and to use recognised vocabularies consistently, including CRediT roles where appropriate.
","doi":"https://doi.org/10.64000/6xx3c-nmp80","guid":"https://doi.org/10.64000/6xx3c-nmp80","language":"en","license":"https://creativecommons.org/licenses/by/4.0/legalcode","published_at":1783555200,"rid":"zf0xb-7ws79","summary":"Research is rarely limited to a single contributor performing a single role. Behind every research output are people contributing in various ways: software development, data analyses, methodology design, and much more. Often, the same person contributes in several of these ways.","tags":["Crossref","Infrastructure","Interoperability","Metadata","Research Nexus"],"title":"Schema 5.5 now available: adding CRediT, new record types for blogs and posters, and more","updated_at":1783835266,"url":"https://www.crossref.org/blog/schema-5.5-now-available-adding-credit-new-record-types-for-blogs-and-posters-and-more/","version":"v1"},{"authors":[{"contributor_roles":[],"family":"Felczak","given":"Michael"}],"blog":{"authors":null,"community_id":"77c8c2e4-ebda-4e7c-9458-6c06b604344b","created":1752192000,"current_feed_url":null,"description":null,"favicon":"https://rogue-scholar.org/api/communities/77c8c2e4-ebda-4e7c-9458-6c06b604344b/logo","feed_format":"application/atom+xml","feed_url":"https://pkp.sfu.ca/feed/atom","filter":null,"generator":"Other","home_page_url":"https://pkp.sfu.ca/","issn":null,"language":"eng","license":"https://creativecommons.org/licenses/by/4.0/legalcode","prefix":"10.59350","relative_url":null,"secure":true,"slug":"pkp","status":"active","subfield":"1710","title":"Public Knowledge Project","updated":1783645240,"use_api":null},"blog_name":"Public Knowledge Project","blog_slug":"pkp","content_html":"PKP's Head of Systems, Michael Felczak, underlines how commercial AI traffic has been putting a strain on hosting providers such as PKP. Read on to learn how PKP is balancing traditional and new tools to protect its hosted journals.
\nWith the rapid growth of commercial AI services and a plethora of applications and agents built upon these services, hosting providers have witnessed an exponential growth in network traffic to their hosted websites. Automated traffic has recently surpassed human-generated traffic and today represents the bulk of access requests to online journals, monograph collections, and preprint servers.\u00a0
\nKeen to train their LLMs, AI companies and service providers view open access scholarly publishing as a free source of data that does not require permission or contractual arrangements typical among commercial operators.\u00a0
\nAs a result, hosting providers are struggling to maintain an acceptable level of service for readers due to new demands from automated traffic, which places high demands on server and network resources. Much of this automated traffic is aggressive, involving requests for pages and URLs at high rates that far exceed human access to published content.
\nTo address the problem of automated traffic, PKP Publishing Services has deployed traditional network management tools alongside new tools that target AI harvesting. These traditional methods include firewalls to restrict origin IP addresses, network throttling to slow down aggressive traffic, and community resources that aggregate information about offending service providers.\u00a0
\nWhile traditional methods are effective to some extent, they are limited to a set of rules that must establish in advance what and who to block. Today AI crawlers often adjust their behaviour, conceal their fingerprints, and switch to a different set of origin IP addresses.\u00a0
\nThe end result is an AI-based iteration of whack-a-mole that requires staff time and resources to monitor traffic, adjust rules, and deploy additional filters to maintain an acceptable level of service for hosted journals and readers. Equally importantly, since AI services are now commonplace for discovery of content and search results, commercial AI services require access to published content and cannot be completely blocked. This additional AI-centric traffic and server activity requires additional server and networking resources.
\nTraditional network management tools co-exist alongside new tools and services that target AI automated traffic. Recently several open source projects have been launched that support dynamic AI crawler filtering and PKP Publishing Services has started to deploy these tools on a limited scale to assess their effectiveness.\u00a0
\nAnubis, the most popular of these projects, filters human traffic from automated traffic by presenting a challenge to website visitors, similar to Google's reCaptcha. Unlike Google's reCaptcha, the Anubis challenge does not require user input and is instead verified by the user's browser.\u00a0
\nEven if AI bots adjust their behaviour or origin IP addresses, if they are unable to complete the Anubis challenge they will continue to be actively blocked. Since it doesn't require user input, from an accessibility perspective the Anubis challenge is more accessible compared to a traditional captcha.\u00a0
\nThis ongoing work represents the commitment of PKP Publishing Services to protect hosted journals from aggressive traffic while balancing the need of emerging AI services to index open access content to ensure that it can be discovered by readers.
\nThe post Managing AI Traffic for PKP Hosted Journals appeared first on Public Knowledge Project.
","doi":"https://doi.org/10.59350/0dm0q-kze91","guid":"https://pkp.sfu.ca/?p=19364","image":"https://pkp.sfu.ca/wp-content/uploads/2026/07/Resource-managing-AI-traffic-July-2026-WP-1024x576.jpg","language":"en","license":"https://creativecommons.org/licenses/by/4.0/legalcode","published_at":1783641600,"rid":"jth0k-cqy48","summary":"PKP's Head of Systems, Michael Felczak, underlines how commercial AI traffic has been putting a strain on hosting providers such as PKP. Read on to learn how PKP is balancing [\u2026] The post Managing AI Traffic for PKP Hosted Journals appeared first on Public Knowledge Project.","tags":["Community Newsletter","News","News For Hosted Clients","AI","Open Journal Systems (OJS)"],"title":"Managing AI Traffic for PKP Hosted Journals","updated_at":1783835229,"url":"https://pkp.sfu.ca/2026/07/09/managing-ai-traffic-for-pkp-hosted-journals/","version":"v1"},{"authors":[{"affiliation":[{"id":"https://ror.org/0153tk833","name":"University of Virginia"}],"contributor_roles":[],"family":"Turner","given":"Stephen D.","url":"https://orcid.org/0000-0001-9140-9028"}],"blog":{"authors":[{"name":"Stephen Turner"}],"community_id":"382941a7-2ffa-41df-8bbb-5f772188517f","created":1780876800,"current_feed_url":null,"description":"A practicing data scientist's take on AI, genomics, biosecurity, and the ways AI is reshaping how science gets done. Weekly updates from the field. Occasional notes on programming.","favicon":"https://rogue-scholar.org/api/communities/382941a7-2ffa-41df-8bbb-5f772188517f/logo","feed_format":"application/rss+xml","feed_url":"https://blog.stephenturner.us/feed","filter":null,"generator":"Substack","home_page_url":"https://blog.stephenturner.us","issn":null,"language":"eng","license":"https://creativecommons.org/licenses/by/4.0/legalcode","prefix":"10.59350","relative_url":null,"secure":true,"slug":"stephenturner","status":"active","subfield":"1311","title":"Paired Ends","updated":1783715968,"use_api":null},"blog_name":"Paired Ends","blog_slug":"stephenturner","content_html":"GPT-5.6 came out yesterday. Plenty of other people are writing about it. I'm not yet, because I'm trying to be intentional about my AI use in July (it's not going great \u2014 AI Dry July is more like an AI-damp-kinda-moist July). But, I was able to get a correct one-shot answer to something many smart people I know get consistently wrong, despite all the evidence widely available.
Another six things this week. Stretching to find a common theme here: each of them touch on knowledge in some way or another:
Anthropic and AE Studio's GRAM, and why removing knowledge from weights isn't the same as removing risk.
PhD admissions down, and what funding uncertainty does to a cohort.
A Brown economics class where the take-home midterm averaged near 100% and the in-person final averaged below 50%.
Watching long form reading erode (keep reading \ud83d\ude09)
AI labs becoming biotech firms.
on the \"Bio Mythos\" moment, and building biosecurity assurance.
And a note, with a very special thanks to my paid subscribers (remember, every post here is free and open, regardless of whether you pay): These kinds of recap posts take a lot of time and energy to write. And exceptionally so while avoiding AI for discovery, scanning, summarizing, triage. I also have a major, very major proposal deadline at the end of this month. While I have a few other essays I'll be publishing soon, you won't see another recap like this for a while. Follow some of the other Newsletters I recommend to keep you caught up.
Anthropic and AE Studio published a method this week for giving a model a removable compartment per category of dual-use knowledge.
They call it GRAM, for Gradient-Routed Auxiliary Modules: add extra neurons to every layer, route the gradient updates from virology or cybersecurity or nuclear-physics text into that category's module during pretraining, then delete the module at inference when you don't want the capability. One training run yields a model you can reconfigure many ways across domains instead of training many filtered models.
They compared with post-hoc unlearning. When they tried to fine-tune a removed capability back in, the unlearning baseline (MaxEnt) recovered almost fully, while GRAM and plain data filtering held. Unlearning after the fact suppresses more than it removes.
* = The asterisks: everything is measured in next-token loss and not actual downstream performance, the dual-use data was a sliver of the mix, nothing has touched a production Claude model, and some capabilities may be
so entangled with general knowledge that no method can separate them cleanly.
Really cool work here but important to pair this with a well-designed uplift study, and better yet, measuring correlates of uplift. Lower next-token loss on virology tokens isn't the quantity that most people (policymakers, biosecurity folks, general public) care about. Uplift is: can someone holding the ablated model do the dangerous thing any less well than someone with a search engine? So model-editing and unlearning work needs paired human uplift studies, same models, capability toggled on and off, measured against a real baseline. Removal that looks clean in loss space but buys no drop in uplift might look like safety but isn't.
NYT / Vimal Patel: Decline of Ph.D. Admissions Could Imperil a 'Generation of New Talent' Gift Link.
PhD admissions this fall dropped 15% from last year across 55 Association of American Universities members, per the AAU Data Exchange. Those schools confer about half the country's research doctorates. This all stacks on last years numbers: for the 42 schools that reported fall 2025, new enrollments were already down. I.e., Two straight years of contraction at the institutions that produce most of the nation's new scientists.
\u2026developing research talent \"is more like planting an orchard than filling a warehouse.\"
Some are far worse than the average. MIT expects nearly 20% fewer new graduate students, about 500 people, citing federal awards down 20% and the new endowment tax. Caltech is cutting new graduate admissions 40% across the board for fall 2026, and its graduate dean was explicit that the driver is a lack of funding certainty, not any specific cut. UW's astronomy chair took zero new doctoral students this year, the first time since 2016.
I look at this with research dean hat on. You need students to drive research. The cohort you don't admit this year doesn't come back when the budget stabilizes.
IHE / Emma Whitford: Brown Professor Suspects Majority of His Class Used AI to Cheat. https://www.insidehighered.com/news/faculty/learning-assessment/2026/07/08/brown-professor-suspects-most-his-class-used-ai-cheat
Roberto Serrano, who has taught welfare economics at Brown for nearly 20 years, gave a take-home midterm for the first time this spring, partly because students were uneasy in a classroom after the December shooting on campus. The class, normally around 30, had 86. The midterm averaged 96% against a historical 65-to-80 range, on an exam he'd made harder. He ran the questions through ChatGPT, got answers matching his students' work down to the same overcomplicated proof strategy, and told the class he suspected widespread AI use. He moved the final in-person.
18 people dropped, 9 skipped the final altogether, and of those who sat it the average was 48%, the lowest he's recorded, against a prior floor of 65. 19 failed.
A 96% to 48% collapse between take home and proctored is strong circumstantial evidence the take home scores weren't the students' own work. It isn't proof about any individual. Brown's academic-code committee wants separate complaints per student with exam copies. AI detectors would throw false positives and negatives in bulk. The committee's own AI report, out the same week, recommends de-emphasizing punishment. His own summary was blunter:
We cannot choose to become idiots.
Atlantic / Rose Horowitch: The End of Reading Is Here. https://www.theatlantic.com/magazine/2026/08/reading-crisis-postliterate-age/687618/
The Atlantic's August cover essay, by Rose Horowitch, argues America has gone postliterate: not illiterate, but losing the higher-order comprehension sustained reading builds. Fewer than half of adults read a book of any kind in 2022. Reading for pleasure on a given day fell from 28% in 2004 to 16 in 2023, and gambling has passed reading as a leisure activity. Text now thrives inside a shrinking minority, about 20% of adults accounting for more than 80% of books read.
What are you reading now? Last week I finished re-reading Andy Weir's Project Hail Mary (the encore was just as good). I just started reading Stewart Brand's Maintenance of Everything, while listening to David Sedaris's newest collection of essays, The Land and Its People (read by him).
at Scaling Biotech on Anthropic's early-July drug-discovery announcements: Claude Science (which I test-drove here) and an internal effort to find candidates for rare and orphan diseases. His read is that it backfires commercially. Tokens are expensive, open-weight models are closing the gap, and
the whole point of it is to make you use Anthropic's models
instead of self-hosting, so it has to stay pricey. Worse, an in-house drug-discovery team reads to pharma customers as a future competitor trained on their data, however carefully the rare disease problem is chosen.
Related, Amelia Palermo's Nature Biotechnology comment provides some structural context.
Palermo, A. Frontier AI companies as biotech acquirers. Nat Biotechnol (2026). https://doi.org/10.1038/s41587-026-03214-0
In April, Anthropic acquired Coefficient Bio for about $400 million (all stock): roughly 8 months old, <10 computational biologists, no clinical assets. Against a $380B valuation that's about 0.1% dilution. In other words it's a standard tech exit rather than a typical pharma one, with a value Palermo calls
a purchase of foundation model talent and code base.
She sees a new class of buyer, frontier labs acquiring biology platforms for models, data, and people instead of drugs, and predicts several such deals by 2030.
I just took a look at some of the job ads, which point to how serious these companies seem. Anthropic isn't only reselling Claude to pharma. Looks like they're building a wet lab, hiring bench scientists and the people to run it:
OpenAI's public bio hiring leans the a different direction, toward safety, policy, and red-teaming. Which is near and dear to my heart. I can't wait to see what this team does here.
Most of these listings are in the $300k-400k range, plus equity.
and at : Preparing for the \"Bio Mythos\" Moment.
Governments tend to react to AI risk only after a surprise, and Claude Mythos was the cyber case: its ability to find and exploit weaknesses in critical infrastructure software pushed cyber capability to the top of the agenda, and the Fable takedown (reportedly 90 minutes for Anthropic to bar foreign access before it pulled the model worldwide) showed how little assurance infrastructure exists to make those calls deliberately. The controls have since been lifted, though Commerce keeps the right to reimpose them. The argument is that a \"Bio Mythos\" moment, a model crossing into dangerous biological capability, would trigger the same improvisation in a domain where the danger is harder to measure and, in their words,
self-replicating, offense-dominant, and can have enormous societal ramifications.
Their fix is to build the assurance infrastructure before that moment: independent evaluators with real model access and a liability safe harbor, standardized capability tests like their BioTIER, and a default against exposing advanced bio capabilities to unverified users.
Their Bio Capabilities Index shows the trend rising even as newer models add refusals. This is the missing half of the GRAM story in #1 above: you can build an off switch, but someone has to decide when to flip it, and that call needs more than a back-channel phone call (maybe more well-designed uplift studies, and better yet, measuring correlates of uplift). For my fellow Virginia readers, the state's 2026 legislation already directed its technology commission to study an independent-verification framework.
","doi":"https://doi.org/10.59350/0aew2-jrr03","guid":"206293199","image":"https://substackcdn.com/image/fetch/$s_!Z1bI!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F42d8172f-ca1e-4403-98d1-02c93ce18b24_2048x1593.png","language":"en","license":"https://creativecommons.org/licenses/by/4.0/legalcode","published_at":1783641600,"rid":"8jncm-2x730","summary":"Off switches for dual-use knowledge, the PhD admissions drop, a Brown cheating mess, the end of reading, AI labs at the bench, and planning for a \"Bio Mythos\" moment. 1.7k words, 8 min reading time.","tags":["Papers","Biosecurity","AI"],"title":"Closing my tabs, July 10, 2026","updated_at":1783717232,"url":"https://blog.stephenturner.us/p/closing-my-tabs-july-10-2026","version":"v1"},{"authors":[{"contributor_roles":[],"family":"Strauss","given":"Ilan"}],"blog":{"authors":[{"name":"Asimov's Addendum"}],"community_id":"0b81a230-f3a4-4f15-b6db-f262083c3797","created":1781481600,"current_feed_url":"https://asimovaddendum.substack.com/feed/","description":"A publication by Tim O'Reilly and Ilan Strauss on AI commercialization risks and governance. Disclosure and protocols for healthy AI ecosystems.","favicon":"https://rogue-scholar.org/api/communities/0b81a230-f3a4-4f15-b6db-f262083c3797/logo","feed_format":null,"feed_url":"asimovaddendum","filter":null,"generator":"Substack","home_page_url":"https://asimovaddendum.substack.com/","issn":null,"language":"eng","license":"https://creativecommons.org/licenses/by/4.0/legalcode","prefix":"10.59350","relative_url":null,"secure":null,"slug":"asimovaddendum","status":"active","subfield":"1702","title":"Asimov's Addendum","updated":1783685589,"use_api":true},"blog_name":"Asimov's Addendum","blog_slug":"asimovaddendum","content_html":"This is the full version of a five minute \"lightning\" talk given at O'Reilly Media's Foo Camp on Saturday June 27, 2026 at Lighthaven, Berkeley \u2013 co-sponsored by the AI Disclosures Project.
Will or won't AI lead to a \"jobless apocalypse\"? I want to offer one framework for thinking through this question based on Adam Smith and Allyn A. Young \u2014 and Hal Varian on standards.
The question I want to ask you today is as follows: Where does intelligence reside? Where does it live?
The argument I want to make is that the artificial intelligence found in large language models (LLMs) is inert and centralized in the models' weights. This is a very different type of intelligence to that exhibited by markets, which is fundamentally decentralized and constantly being updated.1 Markets already extend and enable machine intelligence; but often sporadically and not always via market exchange. Integrating these two types of intelligence, including through the price mechanism, is vital if AI is to advance specialized production, market exchange, and ultimately employment creation.
To understand market forms of intelligence I want to begin by asking whether this picture below reflects a market-based intelligence. It shows women making men's suits using sewing machines. It looks organized. But according to what principles?
How do these workers know how many suits to make, what types to make, and with what machinery? By informed guesses of a central committee, it turns out.
The above picture is from Bolshevichka garment factory in Moscow, taken in 1967 when it was part of the Soviet Union. This factory used to produce the majority of men's suits for Russia. Although the Soviet Union excelled in heavy industry production, it ultimately failed in consumer goods production, where market signals were needed to inform changing tastes, quantities, efficiencies in resource utilization, and fundamental innovations. Waste was abundant. Human need could not be met. Production techniques were outdated.2
Bolshevichka still exists as a menswear brand and company today, selling men's suits, but now informed by market signals.
What would it mean if the above factory was instead organized on the basis of a market-based intelligence? This would require two things.
The first is decentralized and permissionless innovation. Markets are a federated system, where information remains dispersed along with the trillions of independent buying and selling actions that it enables. Market-processed information is what enables and disciplines these actions (along with society's conventions and laws).
The second is that current information signals inform market coordination through constantly shifting agents' incentives. Prices and product information are always being replenished as preferences and technology change, and demand and supply decisions respond accordingly. Market signals inform decision making but also respond to it. For this to work, price signals and other relevant information must be widely available (and not gameable).
It appears to me that digital markets have till now often tried to mimic market forms of intelligence. Google Search of old is an example of a market-like intelligence, I would propose. Google Search has:
Permissionless innovation (building on the web's open protocols). Anyone can create a website. Information (websites) remain federated on servers.
Dynamically updated signals. What the user clicks on (consumes) in the search result page is used to improve and constantly update website search result rankings. This in turn improves the next allocations of user attention to information, which in turn informs what information websites and other producers should supply to the market.
The algorithms of this era advanced what Tim O'Reilly calls \"an architecture of participation\".
But will markets create jobs? The truth about market-based systems of intelligence is that humans may not always be needed. Nissan's \"intelligent factory\" below is just one example of labor being made redundant in manufacturing production. But without markets expanding production on a progressive and balanced basis, little prospect for human employment and prosperity exist.
How do market forms of intelligence compare with present day artificial intelligence in LLMs?
1) An LLM's intelligence is inert and unresponsive at its core. The model's weights are trained on information and feedback after which they become fixed, \"dead\" things. They are unable to dynamically update or learn as they go. That is why LLMs are reliant on markets for a steady stream of new information, and in particular markets online.
Besides for the internet, the other primary source of live information for models is human users and firms. Every time a user has a conversation with a generative AI system, uploads a document, or tells it something, they allow the model's weights to respond to live, up-to-date, information. Post-training also incorporates human judgement and information, organized by firms such as Surge AI. But direct market signals seem to only weakly inform these activities and the piece rates offered to human annotators.
2) The other aspect of present AI is that it aims to be generalized \u2013 as in artificial general intelligence. In doing so, LLM companies strive to be able to do every task that a human can do and at a high level of competency. This potentially undermines the idea of specialized producers existing across a market, each using different and specialized production techniques. The idea of a single generalized producer (\"AGI\") is so radical that it threatens to undermine the notion of markets altogether.
And that is the central contention of this talk: that specialization underpins the ability for a market's intelligence to grow, for permissionless innovation to flourish, and for price signals to function effectively.
Specialized production provides a bridge between market forms of intelligence and artificial intelligence. So if we want to grow the market for AI, as a system of permissionless innovation, with room for human contributions, it needs to permit \u2013 and admit \u2013 that specialized production is necessary and desirable. General intelligence divorced from market forms of intelligence cannot survive, except as a stale set of encyclopedias sitting on the shelf.
Specialized production \u2014 or specialization for short \u2014 is the ultimate basis for market forms of intelligence and its expansion (an idea we owe to Adam Smith). Specialization is what drives production for exchange as for-profit production coordinated via the market.
Allyn Young in 1928 went one step further when he argued that specialization in the market is a self-reinforcing dynamic. The more I specialize as a firm, the more I allow other firms to specialize too, as they can then meet more of their needs (procure more of their inputs) via the market, instead of through internal production. This means that specialization is not only the foundation of market exchange (Adam Smith), but can create an ever-expanding basis for the market itself to grow through increases in output (supply).
Specialization as external economies. The benefits of specialization are best viewed from the industry or ecosystem as a whole that it helps to expand. The early American auto industry is a good example of this. Ford and its rivals began as assemblers, drawing on Detroit's dense ecosystem of specialized machine shops and parts makers. The Dodge Brothers famously supplied Ford's engines and transmissions before ever building a car of their own. It was this external supplier base producing standardized components for a variety of buyers that allowed for mass auto production to arise and prices fall. This grew the size of the market for everyone. Ford later advanced vertically integrated production in its famous River Rouge complex (which Charlie Chaplin's 1936 film Modern Times drew on), until more distributed supplier networks gained favor again. The benefits of specialization, therefore, are about the unfolding division of labor between firms, not about a single firm scaling up its own production, which can potentially be at the exclusion of others.
Specialization and integrated / monolithic producers. Specialization undermines the integrated producer, where a single firm attempts to do everything under its own roof, without procuring standardized inputs via the market (or transferring its intermediate outputs only internally, rather than selling them).3
Advancing specialization online and in LLMs' own production process can have similar benefits. But this can only have a chance of succeeding if model building is no longer undertaken from start to finish in relative secrecy (housed within a single firm). And if information and intelligence can permanently reside outside of a model's weights. Model weights don't only have to consume markets, they can also enable them. Standards and open protocols, as the technical analogues of market-enabled specialized production, are two important mechanisms for advancing such an agenda.
Standards and open protocols are two key mechanisms for advancing market forms of intelligence online that are decentralized and permissionless. The goal is to allow for artificial intelligence to be extended into the living market of information, admitting formal prices and for market-based allocations.
Standards enable specialization by allowing interconnected parts to fit together. In markets with many complementary parts, standards are vital to allowing a single producer to address the whole differentiated market of firms as if it were a single source of demand. This helps the supplier to not be captured by a single producer (\"lock-in\"). As my colleague Tim O'Reilly remarked to me, one much-underappreciated ingredient to the industrial revolution's take-off was specialized inputs and parts being produced to a standard.
Such standards are especially important in networked markets since connecting to the network is vital to tap into its scale and capabilities. This means that it can potentially let average revenue per user grow for any firm with access to the network's expanding utility.
So in technology markets, standards not only facilitate lower costs through specialization / scale, they also inform whether the value of a network is captured by one firm or shared by many.
Open protocols are shared rules that allow independent users, agents, or devices to coordinate and communicate on the basis of a publicly available specification. They are a type of standard.
By hardening an interface at a given layer through an open specification they allow other firms to connect to it reliably and then specialize against it. They are a battering ram against the integrated producer and the monolithic architecture. A single, closed architecture is designed so that potentially separable complements cannot emerge as independent markets. Under a monolithic architecture specialized production is impeded because the interfaces between those functions simply don't exist or are controlled by a single firm. But a modular architecture is the technical enabler of specialized production that can be disassembled safely (an idea that traces back to Herbert Simon's 1962 paper, \"The Architecture of Complexity\".)
Once specialization takes hold at one interface via an open protocol, the integrated architecture of the AI lab or the platform can be challenged via the market as firms can now compete to supply or draw on discrete components (such as memory, retrieval, payments, identity, tool invocation, attribution, safety filters, and so on).
In doing so, an open protocol can drive vertical disintegration. Interfaces become standardized and what used to be one firm's internal operation gets spun out into a specialized industry as the protocol now handles the coordination (or the identification, the encryption, the security) that previously required all functions to be retained inside the same firm. That spinning-out of tasks (in economic terms) is precisely Allyn Young's progressive subdivision of production across specialized firms.
Although not usually described in market-shaping terms, open protocols, therefore, are exactly that. They shape the market intelligence that can be built on top of them.
The Model Context Protocol (MCP) comes to mind here. It helped catalyze specialized production in AI applications and a more permissionless innovation by allowing AI agents (clients) to connect to external tools and context. A large ecosystem of specialized servers, each exposing one or more tools and resources, grew up around it, which any compliant client can use. Such specialization in turn grew the market for other specialized producers. Specialization begot specialization and rival labs eventually adopted the standard.
The dream of artificial intelligence being very \"general\" in nature (as in \"AGI\") seems, at the end of the day, to be technically compatible with a specialized division of labor for its production and with the existence of specialized AI producers. The market for AI startup financing seems to think so too, funding different approaches to AGI (and recursive self-improvement) and funding differentiated, highly specialized AI producers at the same time.
Extending machine intelligence into markets, and vice-versa, to allow for human participation and reward via markets (\"an architecture of participation\" for AI) is the focus of the three initiatives we are launching this year at the AI Disclosures Project:
Open Memory Protocol Consortium (OMPC). A multi-stakeholder nonprofit consortium developing open, interoperable specifications for portable AI memory. Membership-based, with open working groups, it provides early-stage technical incubation on specific aspects of agentic memory in specific markets.
The AI Markets Initiative. A cross-institution economics research and data collaboration that galvanizes the best economic minds around AI market design challenges and opportunities. The goal is to create functioning AI markets with architectures of participation.
Skills Marketplace. Launching a skills marketplace with monetization for skills authors based on modular, interoperable, rights-protected skills, with proven uplift to agentic performance. All specifications and code will be open-sourced, to be used as a prototype for others to adopt.
If you are interested in being involved in any of these three initiatives in whatever capacity, we would love to hear from you. Thank you for your support.
This does not preclude the role of firms as owner-producers, hierarchical organizers of capitalist production but still ultimately beholden to the market.
Notes Richard Ericson, in \"The Classical Soviet-Type Economy: Nature of the System and Implications for Reform,\" Journal of Economic Perspectives 5(4), 1991, pg.21: \"When the system pursues a few priority objectives, regardless of sacrifices or losses in lower priority areas, those ultimately responsible cannot know whether the success was worth achieving. The central authorities lack the information and physical capability to monitor all important costs\u2014in particular, true opportunity costs\u2014yet they are the only ones, given the logic of the system, with a true interest in knowing such costs.\"
And then concluding, pg.26: \"Finally, it must be remembered that the ultimate configuration of institutions and interactions is unknowable, a largely unintended consequence of the growth of decentralized agent interaction. Thus, a final lesson for successful reform taught by the nature of the traditional Soviet-style system is to abandon the Faustian urge to control, to know in advance, and thus to allow economic outcomes to arise naturally as the unpredictable consequences of market interaction.\"
So-called agentic orchestration acting on the basis of user memory and context is one attempt to recreate this division of labor arguably, with the user or firm's information providing the specialized ingredients into the produced service \u2013 services are produced as they are consumed. But another approach is simply specialized agents, which implies a more specialized production. The idea that agents lower transaction costs and so might naturally see more production internalized within the firm seems highly speculative. Coordination within a firm is often less efficient / more costly than via the market, noted by Stigler in his 1951 piece (\"The Division of Labor is Limited by the Extent of the Market\"). The boundaries of specialized production are likely in these instances to be decided more by the sensitivity and accessibility of data \u2013 and intellectual property more generally \u2014 along with the ability to have standardized parts and a large enough addressable market.
This article also appeared as a guest post on 1 July 2026 on the blog of the Brandenburg Open Access Networking and Competence Centre (https://open-access-brandenburg.de/gastbeitrag-wertesensible-ansaetze-fuer-die-open-science-beratung/)\n\n\n\n
Hantow, Jonas (2026): \"Guest post: Between FAIRness and responsibility: Value-sensitive approaches to open science consultancy.\" DOI: 10.59350/gepex-r1g49
\"As open as possible, as closed as necessary.\"
\n\n\n\nThis sentence is quoted so often in Open Science contexts that it can almost sound like a solution in itself. It is not. It is a very good starting point, though. But in everyday consulting practice, the fascinating part usually begins with the second half of the sentence: What exactly is necessary? Who decides? And necessary for whom?
\n\n\n\nAt BiblioCon 2026 (May 19-22, 2026 Berlin), we explored these questions in our hands-on lab \"Between FAIRness and Responsibility: Value-sensitive approaches for Open Science consulting\". The workshop was aimed especially at people working in libraries, data services and research infrastructures who support researchers in publishing data, objects and sources. Our starting point was a tension that many consultants know very well: Open Science encourages us to make data findable, accessible, interoperable and reusable. At the same time, there are cases when full openness is not possible, not appropriate or simply ethically not responsible.
\n\n\n\nThis becomes particularly palpable in object-centred and data-intensive disciplines. A digitised museum object is not just a file. Archaeological data are not just coordinates. Historical sources are not neutral just because they are old. They may be connected to people, communities, cultural heritage, fragile sites, vulnerable species, institutional interests or contested histories.
\n\n\n\nThe FAIR principles helps us to think about technical and infrastructural quality. CARE, developed for Indigenous Data Governance, adds an important ethical and political layer: collective benefit, authority to control, responsibility and ethics. But CARE also has a specific origin and a specific addressee. It was designed for indigenous contexts. Many of its concerns can be conveyed into other contexts, but they should not simply be generalized as if all sensitive cases were the same.
\n\n\n\nThis is where Value Sensitive Design becomes a useful inspiration for us as an additional lens. For now, we propose an adaptation as \"Value Sensitive Data Publishing\". It neither replaces FAIR nor CARE. Rather, it helps to clarify what exactly is at stake in a given publication decision. Which values matter here? Which direct and indirect stakeholders are affected? Which conflicts between values become visible? And how can these conflicts be managed, i.e. translated into practical design decisions, workflows, metadata, licenses or access levels?
\n\n\n\nIn the workshop, we discussed this through three fictional but realistic cases.
\n\n\n\nThe first case dealt with fictional excavation data from G\u00f6beklitepe. Historical handwritten lists had been digitised, made full-text searchable with OCR and enriched with FAIR metadata. From an Open Science perspective, this sounds like a desired outcome. But what if the data contain precise coordinates of places of discovery? What if publication of those coordinates could enable looting? Or what if such data and archaeological interpretations shaped by specific Western epistemologies are used to eclipse local cultural narratives? How to handle scholarly claims and political claims? Here, values such as transparency, reproducibility and reusability meet the need to protect cultural heritage, site preservation and sources. \"Open\" may therefore mean: make metadata visible, document the dataset, contextualize, explain access conditions, but do not publish everything at full resolution and maximum precision by default.
\n\n\n\nThe second scenario focused on sacred objects from Latin America in a German museum. High-resolution 3D models were ready for publication, including information on dimensions, materials and historical descriptions. But one object contained a specific ochre tone that, according to the case description, should not be seen by women in the community of origin for religious reasons. This example shows very clearly that values are not limited to privacy or the protection of individual persons. Cultural goods can require protection. So can religious practices, community knowledge and relationships between institutions and source communities. How can the scholarly and wider public interest in access, availability, and reuse be balanced with culturally specific taboos in this case? And in natural history collections, we may have to think even further: information about nesting sites, for example of a protected bird species, can be sensitive because its availability would affect animals and their habitats. Even just a group of well-intended bird-watchers can pose a disruptive element for a natural setting. Value-sensitive publication therefore needs a broader concept beyond \"personal data\" and privacy alone. This is even more true, when deep text and data mining combined with automatic, AI based synoptic tools allow for broad triangulation of data resources.\u00a0
\n\n\n\nThe third case concerned digitised books from the colonial period of the German Empire. Here, Open Access and the question of proper contextualization can conflict. Should racist images, discriminatory terminology and imperial narratives be made openly available and even provided for possible reuse, as the application of a CC BY licence would entail? We think the more expedient question would be: Under which conditions can Open Access support critical engagement instead of reproducing harm? Contextualisation should not be understood as an obstacle to openness, but as one of its enabling conditions. Users also have an agency and responsibility to engage critically with the material. Infrastructures can help them do so: through usage notes, provenance information, explanatory metadata, essays, teaching materials or links to current research.
\n\n\n\nThis also changes how we think about licences. In many Open Access contexts, CC BY is the default reflex and demand. But when dealing with sensitive historical sources, objects or images, a more restrictive licence such as CC BY-ND may sometimes better support the goal of preventing decontextualised or misleading derivatives. This does not mean that CC BY-ND is always the answer. It means that licence choice is itself a value-sensitive decision.
\n\n\n\nOne important distinction emerged repeatedly: Open Access for research publications and Open Access for sources are not the same thing. A journal article, an interpreted dataset, a 3D scan of a sacred object, a colonial photograph and a precise locality record of a protected species all have different conditions of \u2013 ethical \u2013 openness. Treating them as if they followed the same logic makes consulting easier, but also leaves more space for potentially improper or even malfeasant reuse.
\n\n\n\nWhat did we take away from this session?
\n\n\n\nThat is perhaps the most important point. Context, restrictions and care are not the enemies of Open Science. In many cases, they are what makes Open Science possible.
\n\n\n\nBiblioCon 2026:\nConference website (in German): https://www.bib-info.de/fortbildung/bibliothekartage/bibliocon-2026-bib\nWorkshop-Session: https://bibliocon2026.abstractserver.com/program/#/details/sessions/151\nAbstract: https://bibliocon2026.abstractserver.com/program/#/details/presentations/145\n\n\nMore information on CC-BY-ND licenes:\nhttps://creativecommons.org/licenses/by-nd/4.0/deed.en\n\nWiNoDa offers a self-paced online course on Regulatory Frameworks for Data Acquisition in Fieldwork on our Moodle platform. Just register on www.moodle.winoda.de and enroll!\nDieser Artikel erschien am 1. Juli 2026 auch als Gastbeitrag im Blog der Vernetzungs-und Kompetenzstelle Open Access Brandenburg (https://open-access-brandenburg.de/gastbeitrag-wertesensible-ansaetze-fuer-die-open-science-beratung/)\n\n\n\n
Hantow, Jonas (2026): \"Gastbeitrag: Zwischen FAIRness und Verantwortung: Wertesensible Ans\u00e4tze f\u00fcr die Open-Science-Beratung.\" DOI: 10.59350/gepex-r1g49
\n\n\n\n\n\n\n\n\n\"So offen wie m\u00f6glich, so geschlossen wie n\u00f6tig.\"
\n
Dieser Satz wird in Open-Science-Kontexten so h\u00e4ufig zitiert, dass er fast selbst wie eine L\u00f6sung klingen kann. Das ist er aber nicht. Er ist jedoch ein sehr guter Ausgangspunkt. Die eigentliche Arbeit beginnt meistens im zweiten Teil des Satzes: Was genau ist n\u00f6tig? Wer entscheidet das? Und n\u00f6tig f\u00fcr wen?
\n\n\n\nAuf der BiblioCon 2026 (19.-22. Mai 2026, Berlin) diskutierten wir diese Fragen in unserem Hands-on-Lab \"Zwischen FAIRness und Verantwortung: Wertesensible Ans\u00e4tze f\u00fcr die Open-Science-Beratung\". Der Workshop richtete sich insbesondere an Personen aus Bibliotheken, Datenservices und Forschungsinfrastrukturen, die Forschende bei der Publikation von Daten, Objekten und Quellen beraten. Ausgangspunkt war ein aus der Beratungspraxis sehr vertrautes Spannungsfeld: Open Science fordert dazu auf, Forschungsdaten, -quellen und -materialien nahezu unbeschr\u00e4nkt auffindbar, zug\u00e4nglich, interoperabel und nachnutzbar zu machen (FAIR). Gleichzeitig gibt es F\u00e4lle, in denen vollst\u00e4ndige Offenheit nicht m\u00f6glich, nicht angemessen oder schlicht ethisch nicht verantwortbar ist.
\n\n\n\nBesonders deutlich wird das in objektzentrierten und datenintensiven Disziplinen. Ein digitalisiertes Museumsobjekt ist nicht einfach nur eine Datei. Arch\u00e4ologische Daten sind nicht einfach nur Koordinaten. Historische Quellen sind nicht neutral, nur weil sie alt sind. Sie k\u00f6nnen mit Personen, Communities, Kulturg\u00fctern, gef\u00e4hrdeten Fundorten, schutzbed\u00fcrftigen Arten, Interessen von Forschenden und Institutionen oder konfliktreichen Geschichten verbunden sein.
\n\n\n\nDie FAIR-Prinzipien helfen uns, \u00fcber technische und infrastrukturelle Qualit\u00e4t nachzudenken. CARE, entwickelt f\u00fcr Indigenous Data Governance, erg\u00e4nzt eine wichtige ethische und politische Perspektive: kollektiver Nutzen, Kontrolle und Verantwortung. Zugleich hat CARE einen spezifischen Ursprung und einen spezifischen Zuschnitt. Die Prinzipien beziehen sich auf indigene Kontexte. Viele der dort formulierten Anliegen sind auch f\u00fcr andere Zusammenh\u00e4nge relevant, sollten aber nicht verallgemeinert werden, als w\u00e4ren alle sensiblen F\u00e4lle gleich.
\n\n\n\nAn dieser Stelle kann der Ansatz des Value Sensitive Design als zus\u00e4tzliche Linse dienen, um die Bewertung sinnvoll zu erweitern. Im vorliegenden Zusammenhang sprechen wir dabei vom \"wertesensiblen Publizieren von Forschungsdaten\". Der Ansatz ersetzt weder FAIR noch CARE. Er hilft vielmehr, erg\u00e4nzend und genauer zu kl\u00e4ren, worum es aus ethischer Sicht bei einer konkreten Publikationsentscheidung geht. Welche Werte sind betroffen? Welche direkten und indirekten Stakeholder m\u00fcssen mitgedacht werden? Welche Wertekonflikte werden sichtbar? Und wie lassen sich diese Konflikte in praktische Entscheidungen zu Workflows, Metadaten, Lizenzen, Zugriffsstufen oder Infrastrukturen \u00fcbersetzen?
\n\n\n\nIm Workshop haben wir dies anhand von drei fiktiven, aber realit\u00e4tsnahen Anwendungsf\u00e4llen erprobt.
\n\n\n\nDer erste Fall behandelte fiktive Grabungsdaten aus G\u00f6beklitepe. Dabei w\u00fcrden historische handschriftliche Listen digitalisiert, mit OCR volltexterschlossen und mit FAIR-Metadaten angereichert werden. Aus Open-Science-Perspektive klingt das zun\u00e4chst absolut begr\u00fc\u00dfenswert. Aber was passiert, wenn die Daten pr\u00e4zise Koordinaten von konkreten Fundstellen enthalten? Was, wenn eine Ver\u00f6ffentlichung Pl\u00fcnderungen erleichtert? Was, wenn diese Daten und die daraus abgeleiteten arch\u00e4ologischen Interpretationen lokale kulturelle Narrative \u00fcberlagern? Wie lassen sich hier wissenschaftliche Erkenntnisziele und m\u00f6gliche politische Anspr\u00fcche austarieren? In solchen F\u00e4llen treffen Werte wie Transparenz, Reproduzierbarkeit und Nachnutzbarkeit auf den Schutz von Kulturg\u00fctern, den Erhalt von Fundorten und Schutz arch\u00e4ologischer Quellen. \"Open\" kann in diesem Fall bedeuten: Metadaten sichtbar machen, den Datensatz dokumentieren und kontextualisieren, Zugangsbedingungen abstufen, also entsprechend nicht alles von vornherein in maximaler Aufl\u00f6sung und Pr\u00e4zision ver\u00f6ffentlichen.
\n\n\n\nDer zweite Musterfall drehte sich um sakrale Objekte aus Lateinamerika in einem deutschen Museum. Hochaufl\u00f6sende 3D-Digitalisate st\u00fcnden zur Publikation bereit, einschlie\u00dflich Informationen zu Ma\u00dfen, Materialien und historischen Beschreibungen. Mindestens ein Objekt w\u00e4re jedoch in einem bestimmten Ockerton gefasst, der in der Herkunftsgesellschaft aus religi\u00f6sen Gr\u00fcnden von Frauen nicht gesehen werden soll. Dieses Beispiel zeigt sehr deutlich: Werte beziehen sich nicht nur auf Privatsph\u00e4re oder den Schutz einzelner Personen. Auch kulturelle Regeln k\u00f6nnen schutzbed\u00fcrftig sein. Gleiches gilt f\u00fcr religi\u00f6se Praktiken, Community-Wissen und Beziehungen zwischen Forschenden, Institutionen und Herkunftsgesellschaften. Wie lassen sich in diesem Fall die Anspr\u00fcche von Wissenschaft und vielleicht auch der Allgemeinheit an einen umfassenden offenen Zugang mit eventuellen kulturellen Tabus vereinbaren? Und in naturwissenschaftlichen Sammlungen m\u00fcssen wir vielleicht noch weiter denken: Informationen \u00fcber Brutpl\u00e4tze gesch\u00fctzter Vogelarten k\u00f6nnen sensibel sein, weil dadurch sensible Lebensr\u00e4ume betroffen sein k\u00f6nnen. Selbst sehr wohlmeinende Freizeitvogelkundler*innen k\u00f6nnten in so einem Setting zum gravierenden St\u00f6rfaktor werden. Eine wertesensible Publikation braucht daher eine konzeptionelle ethische Differenzierung, die \u00fcber \"personenbezogene Daten\" hinausreicht. Dies ist umso relevanter, wenn Text- und Data-Mining-Verfahren mit automatisierten und KI-basierten Zusammenfassungswerkzeugen weitreichende Triangulierungen von Datenquellen erm\u00f6glichen.\u00a0
\n\n\n\nDer dritte Musterfall behandelte digitalisierte B\u00fccher aus der Kolonialzeit des Deutschen Reichs. Hier wirkt es zun\u00e4chst so, als st\u00fcnden Open Access und notwendige Kontextualisierung einander gegen\u00fcber. Sollten rassistische Abbildungen, diskriminierende Begriffe und imperiale Narrative frei zug\u00e4nglich und sogar nachnutzbar gemacht werden, was eine CC-BY-Lizenzierung erm\u00f6glichen w\u00fcrde? Die hilfreichere Frage lautet aus unserer Sicht: Unter welchen Bedingungen erm\u00f6glicht ein offener Zugang eine kritische Auseinandersetzung, anstatt Schaden zu reproduzieren? Kontextualisierung sollte nicht als Hindernis f\u00fcr Offenheit verstanden werden, sondern als eine ihrer Erm\u00f6glichungsbedingungen. Nutzende erhalten so die M\u00f6glichkeit, aber auch die Verantwortung, sich kritisch mit dem Material auseinanderzusetzen. Infrastrukturen k\u00f6nnen sie dabei unterst\u00fctzen: durch Nutzungshinweise, Provenienzinformationen, erl\u00e4uternde Metadaten, Begleittexte, Lehrmaterialien oder Verweise auf aktuelle Forschung.
\n\n\n\nDas ver\u00e4ndert auch den Blick auf Lizenzen. In vielen Open-Access-Kontexten ist CC BY die scheinbar naheliegende und w\u00fcnschenswerte Wahl. Bei sensiblen historischen Quellen, Objekten oder Bildern kann eine restriktivere Lizenz wie CC BY-ND jedoch in bestimmten F\u00e4llen besser dazu beitragen, dekontextualisierte oder irref\u00fchrende Bearbeitungen zu verhindern. Das hei\u00dft nicht, dass CC BY-ND immer die richtige Antwort ist. Es hei\u00dft nur: Auch die Lizenzwahl ist eine wertesensible Entscheidung.
\n\n\n\nEine wichtige Unterscheidung tauchte in der Diskussion immer wieder auf: Open Access f\u00fcr Forschungsarbeiten und Open Access f\u00fcr Quellen sind nicht dasselbe. Ein Zeitschriftenartikel, ein interpretierter Datensatz, ein 3D-Scan eines sakralen Objekts, eine koloniale Fotografie und eine pr\u00e4zise Fundstellenangabe zu einer gesch\u00fctzten Art unterliegen unterschiedlichen Voraussetzungen von Offenheit. Sie so zu behandeln, als g\u00e4lte f\u00fcr alle dieselbe Logik, macht Beratung einfacher, aber nicht besser, sondern erh\u00f6ht das Potenzial f\u00fcr unangemessene oder sogar b\u00f6swillige Nachnutzungen.
\n\n\n\nWas nehmen wir aus der Veranstaltung mit?
\n\n\n\nGenau das ist der wichtigste Punkt. Kontext, Einschr\u00e4nkungen und Sorgfalt sind keine Gegner von Open Science. In vielen F\u00e4llen sind sie das, was Open Science \u00fcberhaupt erst m\u00f6glich macht.
\n\n\n\nWebseite der BiblioCon 2026:\nKonferenz: https://www.bib-info.de/fortbildung/bibliothekartage/bibliocon-2026-bib\nWorkshop-Session: https://bibliocon2026.abstractserver.com/program/#/details/sessions/151\nAbstract: https://bibliocon2026.abstractserver.com/program/#/details/presentations/145\n\nInformationen zur CC-BY-ND Lizenz:\n https://creativecommons.org/licenses/by-nd/4.0/deed.de\n\nWiNoDa bietet einen Selbstlernkurs \u00fcber Regulatory Frameworks for Data Acquisition in Fieldwork (auf Englisch) an. Einfach auf unserer Moodle Plattform www.moodle.winoda.de registrieren und einschreiben. (Gesch\u00e4tzte Zeit: ca. 1 Stunde)
\"Der fr\u00fche Studi f\u00e4ngt die Punkte \u2026 !?\" Optimierung von Erm\u00fcdung und Passung f\u00fcr die Pr\u00fcfungszeit
\nZwei Kommilitoninnen nach einer 8-Uhr-Klausur. Die eine kommt heraus und sagt: \"Perfekt, ich war hellwach.\" Die andere hat sich durch die erste Stunde gequ\u00e4lt und ist \u00fcberzeugt, dass sie am Nachmittag das Doppelte gewusst h\u00e4tte. Beide haben denselben Stoff gelernt, beide sind gut vorbereitet \u2014 und trotzdem f\u00fchlten sie sich unterschiedlich fit. Ist das nur ein Gef\u00fchl, mit dem man sich die Note sch\u00f6nredet? Oder gibt es die eine, biologisch richtige Pr\u00fcfungszeit, die manche von uns systematisch bevorzugt?
\nDie kurze Antwort vorweg: Die Tageszeit wirkt tats\u00e4chlich messbar auf Pr\u00fcfungsleistung \u2014 aber der Effekt ist kleiner, als das Bauchgef\u00fchl vermutet, und er sieht f\u00fcr \"Lerchen\" und \"Eulen\" unterschiedlich aus.
\nDass Menschen zu verschiedenen Zeiten leistungsf\u00e4hig sind, ist keine Ausrede fauler Langschl\u00e4fer. Der Chronotyp \u2014 die Neigung, eher fr\u00fch (\"Lerche\") oder eher sp\u00e4t (\"Eule\") aktiv zu sein \u2014 verteilt sich in der Bev\u00f6lkerung breit und stabil, von extremen Fr\u00fchtypen bis zu extremen Sp\u00e4ttypen (Roenneberg et al., 2003). Ein sp\u00fcrbarer Teil dieser Unterschiede ist genetisch mitbestimmt: Gro\u00dfe Assoziationsstudien haben mehrere Gene identifiziert, die reproduzierbar mit dem Chronotyp zusammenh\u00e4ngen (Kalmbach et al., 2017).
\nF\u00fcr Euch als Studierende kommt eine biografische T\u00fccke dazu. Im sp\u00e4ten Jugend- und jungen Erwachsenenalter verschiebt sich der innere Fahrplan nach hinten \u2014 ausgerechnet dann, wenn Unterricht, Vorlesungen und Klausuren fr\u00fch beginnen. Diese Schere zwischen biologischer und sozialer Zeit tr\u00e4gt sogar einen eigenen Namen: sozialer Jetlag. Wer unter der Woche gegen die eigene innere Uhr aufsteht, sammelt ein Schlafdefizit an und kompensiert es am Wochenende (Wittmann et al., 2006). Die 8-Uhr-Klausur trifft also nicht alle gleich hart.
\nWenn man die Forschung ordnet, wirken zwei verschiedene Mechanismen \u2014 und sie werden in Alltagsdebatten gern verwechselt.
\nErstens: \u00dcber den Tag erm\u00fcden fast alle. Die sauberste Feldevidenz stammt aus D\u00e4nemark, wo Forschende Millionen standardisierter Testergebnisse auswerten konnten. Der Testzeitpunkt hing dort vom Stundenplan und der Computerverf\u00fcgbarkeit ab, war also nicht von den Sch\u00fclerinnen w\u00e4hlbar. Ergebnis: Pro Stunde sp\u00e4ter am Tag sank die Leistung \u2014 im Sinne eines kleinen, aber verl\u00e4sslichen Abw\u00e4rtstrends, den die Autoren als kumulierende kognitive Erm\u00fcdung deuten (Sievertsen et al., 2016). Auch in gro\u00dfen US-Schuldaten schnitten Lernende im Fach Mathematik am Vormittag besser ab als am Nachmittag (Pope, 2016). Dieser Effekt betrifft im Prinzip jeden, unabh\u00e4ngig vom Chronotyp.
\nZweitens: Es z\u00e4hlt die Passung zwischen innerer Uhr und Pr\u00fcfungszeit. Hier wird es individuell. Fachleute nennen das den Synchronie-Effekt: Wir rufen unsere beste Leistung ab, wenn der Pr\u00fcfungszeitpunkt zur inneren Uhr passt. Den bislang \u00fcberzeugendsten Beleg lieferte eine argentinische Studie, in der 753 Jugendliche per Los einer Morgen-, Nachmittags- oder Abendschicht zugeteilt wurden (eine seltene echte Randomisierung im Bildungskontext!). In der Morgenschicht schnitten Fr\u00fchtypen deutlich besser ab als Sp\u00e4ttypen, am st\u00e4rksten in Mathematik. In der Nachmittagsschicht verschwand dieser Vorteil, und Sp\u00e4ttypen profitierten sogar vom Abendunterricht (Goldin et al., 2020). Genau das erkl\u00e4rt, warum manche morgens gl\u00e4nzen und andere nicht: Es gibt keine f\u00fcr alle beste Uhrzeit, sondern f\u00fcr jeden ein eigenes Zeitfenster.
\nDass die Tageszeit dabei nicht nur \"Wachheit\" ist, zeigen kognitionspsychologische \u00dcbersichten: Aufmerksamkeit, Arbeitsged\u00e4chtnis und exekutive Funktionen schwanken \u00fcber den Tag hinweg systematisch (Valdez et al., 2014). Und Chronotyp-Reviews zeigen ein Muster, das genau dazu passt: Der Zusammenhang zwischen Chronotyp und Noten ist morgens am st\u00e4rksten und l\u00f6st sich nachmittags auf (Zerbini & Merrow, 2017). Das ist die Handschrift des Synchronie-Effekts \u2014 morgens liegen Fr\u00fchtypen im Takt ihrer inneren Uhr und Sp\u00e4ttypen nicht, sodass sich die Leistungen dann am deutlichsten unterscheiden.
\nJetzt kommt eine Einordnung, die in Pr\u00fcfungsdebatten leider meist fehlt. \u00dcber alle Studierenden gemittelt sind die Zusammenh\u00e4nge ziemlich klein. Meta-Analysen finden f\u00fcr die Verbindung von Abendorientierung und schlechteren Noten ziemlich kleine Korrelationen (Preckel et al., 2011; Tonetti et al., 2015). Das hei\u00dft: Der Chronotyp erkl\u00e4rt nur einen winzigen Bruchteil dessen, was Eure Note ausmacht. Vorbereitung, Vorwissen, Motivation und Schlaf wiegen um ein Vielfaches schwerer.
\nKlein hei\u00dft aber nicht bedeutungslos. Ein kleiner, systematischer Nachteil kann an einer Bestehensgrenze oder bei einer knappen Auswahlentscheidung den Ausschlag geben \u2014 und weil er systematisch mit der Pr\u00fcfungszeit zusammenh\u00e4ngt, ist er eine Fairnessfrage und nicht nur Pech. F\u00fcr Euch pers\u00f6nlich lohnt es sich also, die Tageszeit als einen von vielen Stellhebeln zu sehen, nicht als Schicksal. Falls Ihr also die Wahl habt, sucht eine f\u00fcr Euren Chronotyp gut passende Pr\u00fcfungszeit aus.
\nWenn Passung so wichtig ist, warum legt man Pr\u00fcfungen dann nicht einfach in das jeweils optimale Zeitfenster? Weil das f\u00fcr eine ganze Kohorte unm\u00f6glich ist. In demselben H\u00f6rsaal sitzen Lerchen und Eulen nebeneinander. Ein Termin, der den einen entgegenkommt, benachteiligt die anderen. Ein pauschal sp\u00e4terer Pr\u00fcfungsbeginn hilft eben nicht allen \u2014 Sp\u00e4ttypen profitieren, Fr\u00fchtypen verlieren. Dazu kommen Raumkapazit\u00e4ten, Aufsichten, Fristen und die Vergleichbarkeit \u00fcber Pr\u00fcfungstermine hinweg.
\nWas strukturell trotzdem geht, ist gut belegt. Pausen sind der billigste Hebel \u00fcberhaupt: Schon 20 bis 30 Minuten d\u00e4mpfen den Erm\u00fcdungsabfall sp\u00fcrbar (Sievertsen et al., 2016). Sp\u00e4tere Startzeiten verl\u00e4ngern nachweislich den Schlaf von Jugendlichen und gehen tendenziell mit besserem Befinden einher (Alfonsi et al., 2020; Marx et al., 2017). Und Institutionen k\u00f6nnen die Pr\u00fcfungszeit als das behandeln, was sie ist \u2014 eine systematische, dokumentierbare Einflussgr\u00f6\u00dfe, die man bei Auswertungen zumindest im Blick behalten sollte.
\nRealistisch bleibt f\u00fcr Euch ein schmaler, aber nutzbarer Spielraum. Wer bei computerbasierten Pr\u00fcfungen einen Termin w\u00e4hlen darf, kann ihn grob am eigenen Chronotyp ausrichten. Wichtiger als die Uhrzeit ist aber fast immer der Schlaf davor: Ausreichend und regelm\u00e4\u00dfig zu schlafen verschiebt mehr als jede Feinjustierung des Pr\u00fcfungsslots (Nguyen et al., 2025). Und in langen Pr\u00fcfungen (wie zum Beispiel Staatsexamina) sind bewusst genutzte Pausen kein Zeitverlust, sondern messbar investierte Leistung.
\nSo gut die Schul- und Laborevidenz ist \u2014 f\u00fcr den Kern des Medizinstudiums klafft eine L\u00fccke. Direkte Studien dazu, wie sich die Tageszeit auf gro\u00dfe schriftliche Staats- oder Fachexamina auswirkt, fehlen weitgehend. Die belastbaren Zahlen stammen aus Schulkontexten und aus dem Labor. Hinzu kommt, dass sich der Synchronie-Effekt bei jungen Erwachsenen nicht in jeder Studie robust zeigt und in kontrollierten Wiederholungen teils ausbleibt. Und in den gro\u00dfen Felddaten lassen sich Erm\u00fcdung und innere Uhr nur schwer sauber trennen, weil Schlaf, Ern\u00e4hrung und Fachschwierigkeit mit hineinspielen. Wer aus einem kleinen Mittelwertunterschied eine feste Regel f\u00fcr die eigene Pr\u00fcfung ableitet, \u00fcberzieht eventuell die Evidenz.
\n@misc{friederichs2026,\n author = {Friederichs, Hendrik},\n title = {\"Der fr\u00fche Studi f\u00e4ngt die Punkte ...\\,!?\" -\\/- gibt es eine\n optimale Pr\u00fcfungszeit?},\n date = {2026-07-09},\n url = {https://medical-education.pages.ub.uni-bielefeld.de/research/mes-blog/posts/2026-07-09-daytime-performance/},\n langid = {de}\n}\n