National Academies workshop on scientific integrity

I spent most of the day in a (probably ill-advised) attempt to provide a service to the online science community. I followed as much of day 2 of the National Academies workshop on research integrity.

Most of the workshop was good, although I wouldn’t have called it a workshop, because I didn’t see any workshopping. I mostly saw people delivering familiar messages that you would expect based on their background. Publishers talked about the efforts that they are taking to try to screen out problem papers. Others said publishers were the problem and that we should walk away from journals. And so on.

But it was generally good, thoughtful stuff, even if the interaction was low. They did have a form to submit questions. Heck, one of mine got asked of a panel! Ironically, the answer came just at a point I couldn’t watch my laptop, so I didn’t get to hear the answer.

Attendance, from what I saw, fluctuated around a hundred. It was below 100 for the segment that attracted the most attention, a “fireside chat” with NIH and acting CDC director Jay Bhattacharya, moderated by Emily Oster.

To be clear, this was the worst session by far that I saw during the entire day and a half. I’ve watched a few events hosted by the National Academies, and this was the worst I’ve ever seen.

Oster was not an effective moderator. Oster was far too chummy with Bhattacharya, who started by saying that he agreed to appear because he was going to talk to Oster. She was slightly better at this than I feared. She did press Bhattacharya to answer questions he kept digressing away from. She did try ask one question from the audience. At least I thought it came from the audience. They were talking so far it was hard to keep up as I was typing.

Bhattacharya is not an effective communicator. He has a habit that I often share: he starts a sentence without knowing what he is going to say, so he is constantly backtracking and digressing rather than finishing what he is saying. 

I gave this event a chance. Bhattacharya started with some kind of reasonable discussion about replication.

But around the halfway point, his persecution complex started to kick in. And the session just went downhill, picking up speed as it went on.

He started to drop accusations of people having “bad faith” discussions about the NIH, that some information was “fake news,” blamed the Biden administration for... something (I can’t remember what), and near the very end complained about the New York Times and Washington Post of “not telling the truth” about a paper that he blocked the CDC from publishing.

It was about as bad as many critics predicted. 

At the end, organizer Katy Milkman seemed to say that she was “dispirited” but that having Bhattacharya on was “the right call.” I disagree. Even though well intentioned, I think it was a very bad call.

I do want to give a nod to earlier participant, Andrew Gelman, who was about the only person I heard who specifically called out the federal government for anti-vaccine disinformation.

I was able to thread all of my comments, which you can read as a single page on Threadsky here: https://tbsky.app/profile/doctorzen.net/post/3mkam24d2vk2y

(Bluesky users: This is a great tool. Copy the link to the first post in a thread. Paste it into a browser. Put a “t” at the start of the URL. That is, “tbsky” instead of “bsky”. And it unrolls the whole thread.

From what I can tell, the only other person posting about the Bhattacharya event as it happened was Jenna Norton. She did not thread her posts, so I am going to do some very fast copy and paste of some of her comments.

Okay, here we go! Jay Bhattacharya and Emily Oster have started their #NASEM fireside chat. 

Bhattacharya noted that seeing Emily Oster is the facilitator cemented his decision to participate. 

He has a pattern of only speaking with people who have been publicly sympathetic to his views. - https://bsky.app/profile/jenna-m-norton.bsky.social/post/3mkb7puipys2u

Bhattacharya is discussing the problem of replication in research. A major source of failure to replicate is lack of diversity in the original study population. Thus his anti “DEI” efforts – which end up excluding already underrepresented people from research – are undermining his stated goal. - https://bsky.app/profile/jenna-m-norton.bsky.social/post/3mkb7xxwhbk2u 

Emily Oster asked Bhattacharya: “What metrics is he going to use to measure improvement in replication?” 

He’s been talking for several minutes and I’ve not heard an answer to this question. - https://bsky.app/profile/jenna-m-norton.bsky.social/post/3mkbaakfxk32u  

In the #NASEM fireside chat with Jay Bhattacharya, he says there is no political review of grants.

👀

NIH staff being required to censor grants base on a computational text analysis tool beg to differ. https://www.science.org/pb-assets/PDF/News%20PDFs/final_staff_guidance-1765918233.pdf - https://bsky.app/profile/jenna-m-norton.bsky.social/post/3mkbb2tcskd2u

In the #NASEM fireside chat with Bhattacharya, he says the impact score correlates strongly with methods and not innovation. He neglects entirely the existence of summary statements as a source of information to program staff and IC directors who *used to* be able to inform final funding decisions. - https://bsky.app/profile/jenna-m-norton.bsky.social/post/3mkbbjf35yt2u 

In the #NASEM chat with Bhattacharya, he talks about how supporting innovation will support early career scientists. Meanwhile, under his leadership, early career scientists have disproportionately suffered. 

Early career scientists were disproportionately affected by terminations. - https://bsky.app/profile/jenna-m-norton.bsky.social/post/3mkbbp7voit2u 

In the #NASEM chat with Bhattacharya, he suggests that requiring vaccines is “coercive” and undermines trust. He ignores the role concerted vaccine disinformation campaigns have had on trust (and the part he played in them). - https://bsky.app/profile/jenna-m-norton.bsky.social/post/3mkbbvd3avo2u

In the #NASEM fireside chat with Bhattacharya, I will give Emily Oster credit on pushing Jay on refusing to publish the CDC study showing covid vaccine reduce hospitalization, especially when CDC just published something with the same methods on flu. - https://bsky.app/profile/jenna-m-norton.bsky.social/post/3mkbc3p6gx62u

He lied more than once about the political influence happening inside NIH, where ideas are very clearly being constrained by his own admission to “agency priorities” that have been determined not by scientific or patient communities, but by the White House & Russ Vought’s Project 2025 plan. - https://bsky.app/profile/jenna-m-norton.bsky.social/post/3mkbcstwsyc2a 

In the #NASEM chat with Jay Bhattacharya, he falsely accusing the Biden admin of censoring social media companies. In the lawsuit he helped bring, he failed to prove that claim. Yet he continues to repeat it. https://www.bbc.com/news/articles/c100l6jrjvno - https://bsky.app/profile/jenna-m-norton.bsky.social/post/3mkbbxxxjsg2u 

Former NIGMS director Jeremy Berg has said he will have a reaction later. I will update this post when he does. 

Related posts

The National Academies get ratio’d on research integrity workshops

Record number of NSF GRFP awards in 2026, awarded to the usual suspects

Several outlets have run articles with the unexpected good news that the National Science Foundation is giving out more GRFP awards than ever, after a deep reduction in 2025.

Hooray! 

But those awards, and I have written about several times before, are highly concentrated in a small number of institutions.

MIT, with about 4.5 thousand undergrads gets 88 awards, versus the entire University of Texas system  of about 200K undergrads getting ~63 awards.

Harvard University, with about ~7k undergrads, gets 51 awards, versus the entire University of California system of about ~200K undergrads getting ~300 awards.

Again, the NSF does not disclose a lot of data about who applied to the awards, so it could be that Texas students aren’t submitting proposals at the same rate. But I recently ran across an analysis of who gets fellowships in by Dominque Baker, which points out that in the awards examined there, they cluster in people working at universities deemed “prestigious.” She wrote:

Longstanding methods of (fellowship) selection… tend to coincidentally find talent in the same places, year after year.

I suspect but can’t prove this describes the GRFP process. Oh, look, it’s just coincidence that a small number of east coast universities suck up most of the awards. Every. Year.

External links

NSF awards record number of coveted PhD fellowships in surprise move 

NSF names record number of graduate fellows, rebounding from 2025 dip 

 Who gets Guggenheims?

The National Academies get ratio’d on research integrity workshops

The National Academies of Science, Engineering, and Mathematics is showing — again — that they are not up to the moment.

Researcher Katy Milkman has organized a Zoom workshop for the National Academies on research integrity, featuring some speakers who have been accused of having little integrity themselves. 

I examined the full schedule, and I have to say: the program is a strange mix. Some people have great records on speaking about integrity! Elizabeth Bik, the master of image duplication detection? Yes. Ivan Oransky, who runs the Retraction Watch blog? Yes. Michael Dougherty, whose helped to reform career incentives at his department to align with DORA principles? Yes!

Others... do not.

The poster (pictured) mentions Jay Bhattacharya and Emily Oster, both of whom gained prominence for saying institutions overreacted to the covid pandemic. The poster doesn’t mention the format: It is a 45 minute “fireside chat” with Oster interviewing Bhattacharya.

Probably due to this poster featuring these two “covid contrarians” so prominently, the first post was ratio’d quite severely, with over 100 comments, 10 reposts, and about 20 likes as of this writing. 

In response to criticism, Milkman wrote:

To everyone responding with vitriol: please come! Our goal in choosing speakers was to platform both popular and unpopular figures advocating for scientific reforms to spark productive conversations across divides! If this makes you angry, come engage in dialogue — that’s how we improve.

Um. If you’re all about “engaging in dialogue, that’s how we improve,” you may not want to call the comments, “vitriol.”

The ratio suggests Milkman’s stepped in it even worse the second time. 444 comments, 158 reposts, and 24 likes as of this writing.

 A co-organizer, Jeremy Freese, wrote (original emphasis):

To me, the viewpoint of the Director of NIH on the main issues of the workshop has obvious and considerable relevance for contemporary science policy. ... To me, it would have been both an abuse of the role and nutty hubris for a co-organizer to try to assert that the Director of NIH would be an improper participant in this event, or that there should be a “Bluesky veto” over speakers.

Two things. Maybe three.

1. This is supposed to be a workshop on “social and behavioural sciences.” Bhattachayra leads the NIH, which mainly does medical research, not social and behavioural research. Yes, I get that the NIH has a wide remit and supports some social and behavioural research, but it is not the major function of the agency.
2. Why invite someone whose work is agency oversight, which is largely removed from research integrity? Why not invite someone whose whole job is research integrity? The NIH has an entire Office of Research Integrity. It’s listed right there on the label!
3. Saying that people posting criticisms are asking for a “Bluesky veto” is a convenient way to dismiss what your colleagues are telling you.* I haven’t read everything that Freese read, but I haven’t seen any posts yet saying, “The event should be cancelled because Bluesky users say so.” What I have seen are people saying, “This event is extraordinarily unlikely to get the productive dialogue that the organizers say they want. It is more likely that this event will platform people who are actively undermining research.” It is weird that organizers who keep touting the “Let’s air out our differences and find a way forward” keep dismissing people and only minimally engaging in dialogue.

* Old joke: “When one person calls you a jackass, that person is obviously an ignorant fool. When ten people call you a jackass, maybe you should think about getting fitted for a saddle.”

Cole Donovan reposted with the comment:

Again, as a former National Academies program director familiar with the rules for these as well as how things should go, the issue isn’t whether the NIH director is an appropriate invitee or not to an event. The issue is whether the event’s design is one where disagreements result in dialogue.

I get that organizers put a lot of effort into creating what they thought would be a good event on a worthwhile topic. It’s hard to admit mistakes. It’s hard to course correct mere days before the event. I understand that they don’t want to undercut their own work just a couple of weeks before the workshop. 

But the whole thing positively smells like what we have gotten from the National Academies over and over again: absolute refusal to acknowledge that American science is in crisis. And that crisis is being caused by the federal government. And that people like Jay Bhattacharya are simply not leading American science out of the crisis.

Sometimes it feels like the National Academies would rather risk a hundred doctoral programs be cut today rather than risk one congressional representative saying something bad about them five years from now.

Related posts

Ignoring catastrophe: The state of science in 2025, according to the National Academies of Sciences

External links

PDF of workshop program 

Why Do Rightwing Foundations Fund Emily Oster’s Work on COVID and Parenting? 

Summarize your research visually

Word cloud generators are always fun. This one creates a word cloud based on a Google Scholar profile.

That “maple leaf” is one of the shape options is one of my favourite things. Unsurprisingly, the author of this tool is Canadian. 🇨🇦

Another favourite thing is that it creates alt text for the image. Hooray for accessibility!

External links

Scholar Google 

Grammarly’s identity theft

Grammarly is stealing the names of living scientists and science writers. It is using those names to sell “expert review” by generative AI, but without the input or even permission of the listed authors. And there isn’t an easy way to check if your name is being used this was. You more or less have to create a free account and poke around in it.

So far, Richard van Noorden has found in Grammarly’s system scientists and science writers like:

• Ivan Oransky
• Mary Roach
• Rebecca Skloot
• Ed Yong
• Melinda Wenner Moyer
• Deborah Blum
• Michael E. Mann
• Carl Zimmer
• Maryn McKenna
• Carl Bergstrom
• Elisabeth Bik
• Michael Eisen
• Anna Abalkina

To name a few that might be familiar. 

Ingrid Burrington coined what might be the Word of the Year for 2026 for these: “sloppelganger.”

Grammarly has said that you can opt out of being listed as an “Expert reviewer” by emailing: expertoptout@superhuman.com.

If you have an online body of work of any size, I highly recommend you use that email.  

Update: Well, that was short-lived. Surekha Davies is reporting that Grammarly is deactivating the “Expert review” feature tomorrow (12 March 2026).

But they aren’t saying “This was wrong and we should not do this,” no. It’s only gone “while we reimagine how to make it more useful for customers and more respectful of the experts whose work it surfaces.”

I suspect this might be corporate speak for, “We are totally putting this back as soon as the legal department tells us how we can avoid the lawsuits.”

External links

Grammarly will keep using authors’ identities without permission unless they opt out

Artificial intelligence agent takes your class for you

Making the rounds on socials today is a new generative A.I. agent that logs into your class website and takes the course for you. I have no inclination to promote this, so I am removing the name of the service, since it’s easily found if someone wanted to. 

This agent is supposed to allow a student to clock out from taking a course entirely. The description is worth reading.

How can [Agent] do all this?
[Agent] has his own virtual computer with a browser, just like you do. He can navigate websites, watch videos, read documents, type in text fields, click buttons, and submit forms. Anything you can do on a computer, [Agent] can do.

How does [Agent]  access my Canvas?
You link your Canvas account once during setup. [Agent] uses your credentials to log in, view assignments, and submit work on your behalf.

Will my professor know?
[Agent] submits assignments from your account just like you would. The work is original and generated per-assignment — not copied from a database.

As a professional educator, and one who has tried to give students a good online learning experience, you better believe that I have feelings about this. Quick notes.

Some students would never use this and recoil at the thought. We like those students.

Some students would try to use it for everything. Way back in 2020, Ian Bogost noted that for a lot of people, university is about an “experience” rather than an education. Maybe they are there to play sports, maybe they are there to party, maybe they are there hoping to find someone to marry. Passing classes while avoiding most or all of the work has always been possible for students with enough money to hire “essay writing services.” 

Some students would use it for some classes. And this is a group that I sympathize with. We educators should be thoughtful about why students might want to using this, even if they know they are cheating themselves out of learning. 

Some may have very specific interests and don’t see the value of taking courses that they see as unrelated to those interests. Some students would use if if they thought the workload was unreasonable or that the work boring. Some students would use it to not fail, particularly if they were going through a rough spot personally. We’ve loaded up a lot of consequences for students if they fail classes, not the least of which is financial aid.

I want taking a class to be so rewarding that the thought of giving that to someone else is as attractive to a student as a machine that offers to eat your food, drink your beers, watch your favourite movies, and go on dates for you. 

And Charlotte Moore-Lambert noted that a lot of people have to fight, and fight hard, for the privilege of getting homework.

Unfortunately, it’s hard to make any class that engaging. I try, but I have to be honest that I’m not that good. Few are.

Update, 23 February 2026: Oh, here’s something I hadn’t even thought of. Security. Mett Seybold on Bluesky wrote:

Students are going to voluntarily give their campus login info to this thing(.) This thing that I, who fancy myself a pretty top-notch portfolio sleuth, have been researching for two hours & can tell you next to nothing about who made it, who funds it, what other companies it integrates with. 

Caveat emptor.

Update, 24 February 2026: Dr. Vanessa writes that the same people behind the agent above also launched an agent for instructors that... logs into Canvas and does all the marking.

And thus, the circle is complete. 

Update, 25 February 2026: I suppose in the interests of (raises hands to make sarcastic air quotes) ✌️✌️ fairness ✌️✌️, I should point to an interview with the creator of this agent. I still don’t want to dignify it with names.

I do find it interesting that the creator of this agent did not finish a university degree, and pretty clearly views what goes on in university as “labour” that nobody wants to do.

And here’s another interview with the homework agent guy. 

Jacquelyn Gill observed:

Since yesterday, the… website underwent some rebranding. The tagline changed from “(It) does the busywork so you don’t have to,” to “(It) is the personal tutor every student deserves.” In the FAQ, “How does (it) do all this?” became “How does (it) help me learn?”

Update, 27 February 2026: The creator has now received multiple cease and desist orders. One from learning management system Canvas, and one from the caretakers of the name he gave the system.

External links

America will sacrifice anything for the college experience 

The shadow scholar 

Malala dot org

Case dismissed against academic publishing

I’m not going to pretend I’m a legal scholar, but I did call this one from the beginning.

Over a year ago, a lawsuit was filed against several academic publishers, alleging they conspired to exploit scientists.

The case didn’t even make it to a verdict. It was dismissed back a few weeks ago, according to Stat News reporter Jonathan Wosen.

I’m a little surprised that I am only stumbling across this now. I do realize that academics have a lot of more dire events happening every day, but given the complaints about publishing practices, I would have thought to see a little more commentary on academic social media.

External links

Federal judge dismisses lawsuit against academic publishers over unpaid peer review 

Related posts

Scholarly publishers sued