Mark Fabian

As UKRI explores using LLMs to review grants, it's a good time to revisit Bryan Wilder's excellent blog post.

There are a lot of naive reasons to oppose AI review ("you'll never automate human intuition!"). But there are also good reasons, including the *load-bearing role of human disagreement.*

The philosophers were so preoccupied with whether they should that they didn't stop to think if they could.

You're a hero don't stop. Those PIs doing <6 in 12 months may be problematic.

I think a rule of thumb is to do 3 for every paper you submit because that's how many reviews you generate with each submission.

A PI with a lab could be generating ~30 reviews/year; doing 6 would be villainous.

This is so good!

Yeah especially when the data shows (iirc) young people leaving social media faster than the elderly.

Yes good point and I hate it when the questions do that! Ask me to make some long and tenuous causal connection to wellbeing.

Rather than bans, I'm more sympathetic to regulations targeting the aspects of social media that the firms themselves admit to be problematic but profitable, like the scrolling slot machine, Facebook emotional contagion, LLM sexy/sycophantic/suicidal chatbots, ragebait for engagement, etc.

I agree on urge to control.

On the other hand, this reminds me a bit of the pearl clutching about nudges being an imposition on my autonomy. Yes, you could trust me to self-nudge by placing the chocolate out of sight (I do). But I also don't mind regulation freeing up my cognitive bandwidth.

let's get this graphic into public finance class

bsky.app/profile/852d...

I always find the wellbeing panel survey awkward to answer because it's typically asking for a personal opinion on something for which we have pretty limited science. And the policy implications are often phrased in quite big terms, like here with the bans.

That said, speaking as a voter not a scientist...
1. Some things could obviously be banned, like the slot machine mechanisms
2. I'm not really seeing the strong benefit of many forms of social media as against the big risks, notably instagram/tiktok/shorts. Social connection is a reddit phenomenon.

I don't remember what I answered for attention. I don't know the literature. My *personal* exp is that it's bad for my attention span.

On the policy stuff I was certainly in the 'hold your horses' camp because I know the evidence is limited and mixed. I suspect there is a lot of heterogeneity.

It does feel like a critical mass of psychopaths who just do everything in game theory instead of playing the litigious metagame through the institutions. But perhaps I am underappreciating the cost and time involved in suing your primary funder.

I can't understand how this isn't transparently extortion and thus illegal. Like, are you being sued? Because if not why are you paying what appears to be a fine? And if it is extortion, why can't you sue for extortion? It's just very strange to me to see this in the world's most litigious country.

Students applying for grad school, or reaching out to professors. I have an important piece of advice for you: STOP DOING THIS 👇 (a thread) #STEM #PhD #gradschool #academictips

😂

1 billion is a big number.

Not to be an economist, but I think if matters ever genuinely got dire our values would shift and our behaviours too.

It's no coincidence that one of the only rich nations with fertility around 3 is Israel - they need to repopulate!

Yeah and for most of us that academic work is publicly funded.

I'd like to know the history of how flagship journals came to be for profit. Like journal of political philosophy editorial board resigned on mass to start a new OA journal, but I'm curious why JPP was w/ a for profit to begin with.

Interesting. I think im lucky that our field is growing. But I have found it challenging to unearth potential reviewers who aren't prominent because of how page rank and citation count search works. Even someone at a solid US state university is basically invisible to cold search for keywords.

Oh wow that's next level.

I've had something orthogonal where I submit a very suitable paper that keeps getting blocked by an outsourced wiley person with a checklist that it doesn't fit e.g. no data statement for a theory paper. I eventually have to email the editor.

I think this must implicitly be practiced in narrow fields but it's awkward for people like me who publish all over the place. I generally try to accept review requests from places I have published.

When charity is compensated it ceases to be charitable and loses its moral qualities. There is reams of literature on this from the study of care.

Trying to fix the commercialisation of academia by commoditising peer review is like trying to cure termites with a controlled house fire.

Science and the humanities are characterised by public goods, large positive externalities, and merit goods, making them Ill-suited to market solutions. Community based solutions are the appropriate policy tool, and "social" media is a fine place to spread the good word.

Mate I *am* doing something about it. I volunteer my time at an open access non profit journal being the structurally change we want. The "whining" is me imploring my trapped colleagues to work with me to get us out of the BS. Social media is the water cooler of academia & so a good place to preach.

I'm surprised that an account with as much anti capitalist content as yours would advise introducing even more intense market forces into an already highly hierarchical and capitalistic industry, one that also happens to produce public goods.

Nope.

Do you think that would improve the scientific ecosystem?

Yeah that's fair. Obviously more egregious at places like MPDi but I got one from scientific reports the other day that just didn't need to be published. I suspect the authors were just trying to hit the absurd quotas in US psych departments.

Why are market forces the most appropriate framework for this policy?

If the publisher is for profit OK, but the profit model is precisely the crux of the problem.

We are a diamond open access non profit journal.

Commercially unprofitable truths aren't obvious candidates for a market approach.

Also that for every paper you submit you should do 3 reviews because that's how many you generate. Certainly that's our in house rule.

The people pumping out 10 scientific reports a year from bland regressions would stop that quick smart if this rule was enforced across journals.

Yes.

Unfortunately, the 'top' journals are often for profit. People review there to publish there.

I'm asking from a diamond open access journal entirely non profit. We can't displace the for profits if we can't find reviewers.

The system is cooked. Base issue is volume of needless papers.

Yes I hate it when I get a request to one email when I already have an account at the journal with a different email. Queue 15 minutes of faff to merge them.