IMO tactics that emerged in war time became simple dispositions of ideological intolerance. It’s fairly common in psychology to breeze past theory you don’t find aesthetically pleasing in a paper to accept it if it shows something interesting. Not nearly as much in linguistics.

For me it's 99.9% in linguistics. Psych ppl are mostly critical but nice

,

6. Month 21-25: Resubmit and wait
7. See 5.

collecting new data (though they won't test the hypothesis), citing Reviewer, and workshopping individual sentences to avoid triggering the reviewer, wondering if any your ideas are actually any good or if you really are an imposter, while telling students they DEFINITELY ARE NOT.
...

Project Joy/Dread Lifepan:
1. Month 1 (study conception): Pure joy
2. Months 2-4: pilots, methods, pre-reg. Hard, but so fun.
3. Months 5-7: Data collection, analysis writing: I'm in heaven
4. Months 8 - 14: Waiting in anxiety
5. Months 15-20: Dread, healing, reconsidering life choices,
...

YOU'RE RUINING OPEN SCIENCE!

I want an metric that automatically scores the severity of Reviewer 2's remarks, and then weights the significance of actually publishing the paper on this basis. This would start to reflect the true work that goes into publishing!

Right - and the worst part is, when you reduce a pool this profoundly, you end up missing out on the best candidates, most creative work, and so on. The best money right now is on hiring in any area BUT ai.

I do not object to any of this in this context. The issue is the hiring fad and what it will do to research programs, grants, and students who think that similar jobs will exist in 10-20 years. Our field thrives under diversity of views, not when we go all in on one idea.

Exactly.

We can reap the benefits of everything LLMs have to teach us without hiring LLM “people”.

It’s an amazing engineering feat. And in some areas there’s a neat ”in principle“ point to be made about what’s learnable given lots of data. But the idea is as old as psychology & at the end of the day we’re an empirical field. What we need is content experts who can use tools, not tool experts.

Seriously though it's insanely depressing for current PhD students, and is driving everyone to think they need to dump LLMs into their papers just to get a job. That's not what we want here.

Here is a link to the entire special issue: royalsocietypublishing.org/toc/rstb/202...

This is a first foray, but maybe a neat direction to go w/ this literature, to probe the extent to which putative universals of number word learning might actually be by-products of contingently universal cultural practices - rather than of cognition itself.

Many could compose small number words using a conjunctive rule & some older kids could also comprehend novel multiplicative expressions—e.g. ‘two twos of bananas’. Exp 2 extended this w/ novel "monkey numbers" (e.g., zo-zi / 3-1) & found that kids could learn rules in a novel base 3 system, too.

In Exp 1 we asked kids to give numbers w/ concatenated numbers using either overt conjunctive morphology (which is featured in many attested number systems) - e.g., give two&two bananas - or w/o explicit morphology - give two-two. We also tested multiplicative rules.