Phill Jones

Cheers Nikesh,
The tension between the need to increase the diversity of training materials and the need not appropriate works, given how much of the training data is assumed to have been acquired would make an excellent @scholarlykitchen.bsky.social post.

We need to divert our resources to sustaining the infrastructure that makes it possible for the community to establish new norms. We need to encourage and reward the behaviour or practices that meet the standards we intentionally set in policy. #r2rconf

It's taken until now for someone to note that in a discussion about researcher incentives and guidelines for AI use in academia, Funders are conspicuously absent.
HT Simone Ragavooloo from @frontiersmedia.bsky.social
#r2rconf

Researchers don't report or share AI failures because there's a stigma against its use.
#r2rconf.

Solutions from @nikeshgo.bsky.social #r2rconf.

AI isn't inherently unfair, it's optimised for a very narrow user group. Each iteration will become more biased, unless we intervene. #r2rconf

This. If we refuse to engage with AI, we lose our ability to influence the outcomes. It's irresponsible to try to wait until the tech goes away.

A brilliant presentation by @nikeshgo.bsky.social from @cactusglobal.bsky.social #r2rconf
Outlining the many ways in which AI amplifies the biases built into scholarly publishing.

Well there goes my weekend plans.

Why isn't Peer Review Commons more popular than it is even though it helps researchers by reducing the number of review cycles and time to publish?

Researchers publish strategically. They make conscious choices about where to publish based on audience and career considerations.
#r2rconf #r2r2026

There are claims that DeepSeek and other Chinese models are 'better' than US models like GPT and Claude depite being smaller.
It's worth noting that there is a live debate in AI circles about how to benchmark LLMs and many who say that the evals do not reflect real-world usefulness well. #r2rconf

Governance of AI is chaotic with various governments actively regulating while others retreat to a laissez-faire position.
#r2rconf

Which future are we building by @cmkeithw.bsky.social #r2rconf. Opens by describing the zero-click 'problem' and asks if it is a problem? Does it matter?

The ‘what should we do to fix peer review’ debate doesn’t appear to have moved on much in 20 years #R2RConf

Keynote by @mcintold.bsky.social at #r2rconf
Focusing on impostors and impersonators, showing how insecure the scholarly record has become.
Our security and QA processes are inadequate because they rely on a definition of trust that predate information warfare.

UK loses measles elimination status Decision made after outbreaks in 2024, when there were nearly 3,000 cases in England and Wales.

Measles is no longer officially eliminated in the UK

There's no cure or effective treatment for measles. 1 in 5 kids that get it require a hospital visit. Since its introduction in 1968, the vaccine has saved an estimated 4,500 lives

Anti-vax nonsense costs lives

www.bbc.co.uk/news/article...

26 months since the cyberattack on @britishlibrary.bsky.social Only now have some services been restored with launch of Alma & PrimoVE from @clarivate.com
1) Well done to both teams on getting this done, and
2) This is a cautionary tale of the dangers of not having a proper tech / security strategy

Why We Must Work Together to Harden the Scholarly Supply Chain - The Scholarly Kitchen At the STM innovation and Integrity days in London last week, it's clear that research integrity has become an increasingly pressing issue. Many publishers are reporting significant increases in submi...

"The systems that we currently have aren't robust enough for the level of quality assurance we need."

How might libraries help with this problem?

scholarlykitchen.sspnet.org/2025/12/17/w...

via @phillbjones.bsky.social

Why We Must Work Together to Harden the Scholarly Supply Chain - The Scholarly Kitchen At the STM innovation and Integrity days in London last week, it's clear that research integrity has become an increasingly pressing issue. Many publishers are reporting significant increases in submi...

Last week I attended the #STMInnovation and #STMIntegrity days in London.
I believe the scholarly research ecosystem needs to change in a pretty substantial way if we're to face in to the societal threats that have emerged.
scholarlykitchen.sspnet.org/2025/12/17/w...

Congratulations to this year's winners of the #vessaliusinnovationawards

1. Profectus.academy
2. Safespace research
3. Thesify

Well done to everyone on the #kargerteam. This year's award was the best yet and the hardest to judge.

Greetings from #London!

The #KargerTeam is attending the #VesaliusInnovationAward ceremony tomorrow at the STM Innovator Fair.

Who will be the #winner this year?

⭐ @authentisci.bsky.social
⭐ Profectus Academy
⭐ Safespace Research
⭐ Scitility PBC
⭐ @thesify.bsky.social

Find out tomorrow!

A lot of #scholcomm #publishing startups focus on using #AI to detect bad manuscripts at the point of review.
Publishing is a late stage step in the scholalry supply change. Trying to detect what has happened at each upstream step is a fool's errand. The whole chain needs hardening #STMinnovation

Trust in scientists and their role in society across 68 countries - Nature Human Behaviour What is the state of trust in scientists around the world? To answer this question, the authors surveyed 71,922 respondents in 68 countries and found that trust in scientists is moderately high.

The opening keynote at #STMInnovation, Rachel Gubermann-Hill of U Bristol and UKCORI says there's no real evidence that the public doesn't trust scientists
Here's the key article she recomended: doi.org/10.1038/s415...

I'm at the STM trends meeting this morning. Thanks to @springernature.com for hosting us. I used to work in this office and it's nice to be back.

I really like Google's pageless layout feature. Unless it's intended to be printed, page breaks are a distraction from the semantic structure of the doc.
I only wish somebody had told them that line lengths longer than about 100 characters are really hard to read
'Narrow' should be 'Medium'

Oxford University Press announces agreement to acquire Karger - Oxford University Press Oxford University Press has signed a definitive agreement to acquire Karger Publishers. The transaction is expected to complete in December.

Big news.
@kargerpublishers.bsky.social to be acquired by Oxford University Press @oxfordacademic.bsky.social

corp.oup.com/news/oxford-...

This label will put your journal's research integrity in black and white - Public Knowledge Project The Publication Facts Label is now freely available for the 34,000 journals using OJS (V. 3.3 – 3.5)

PKP released the Publication Facts Label for OJS 3.3–3.5, a transparency tool to strengthen journal integrity and reader trust.

Built as a plugin, it helps journals demonstrate their quality and accountability.

Learn more: pkp.sfu.ca/2025/10/31/p...
#OpenAccess #OJS #MediaLiteracy #ScholarlyComms

Could research security measures reshape open science? In recent years, research security has become a critical concern for governments, funders, research institutes and, to some extent, publishers.

Research security is an under-discussed topic in #scholcomm and #academicpublishing circles. I think it's time to pay a bit more attention because the mandates and requirements will be coming.

Could research security measures reshape open science?
open.substack.com/pub/scholarl...

The Royal Society Moves to Subscribe to Open The Royal Society is switching eight subscription journals to a Subscribe to Open model. Here’s why and how.

Twenty years after publishing their first #OA article @royalsociety.org are going full #OpenAccess using #SubscribeToOpen as the funding model.
katinamagazine.org/content/arti...

Launch of Scholarly Futures Join us as we try to work out which futures are worth discussing

A new substack from @jontreadway.bsky.social @roobina.bsky.social Sarah Greaves and @pgarner.bsky.social

I've already subscribed and you should too.

scholarlyfutures.substack.com/p/launch-of-...

Three Years After the Launch of ChatGPT, Do We Know Where This Is Heading? - The Scholarly Kitchen Nearly three years after ChatGPT’s debut, generative AI continues to reshape scholarly publishing. The sector has moved from experimentation toward integration, with advances in ethical writing tools,...

I have a new post out on @scholarlykitchen.bsky.social where I look at the evolving AI tools market in scholarly publishing.

Spoiler: It's still maturing and is a little bit messy.