santiago salinas

How long until we get a Sokal-type hoax for AI papers?

This is an amazing study

Fascinating work from @ssalinas.bsky.social and colleagues

Thanks!

Ha, funny you say that! @cbo.bsky.social, James Holehouse, and I are working on something to make that case

🤣

Shit, a Jon compliment is 🔥

12/12 Finally, some easter eggs (couldn't write a soccer paper without repping Newell's Old Boys): red and black in Fig. 1 and a dedication to Bielsa in the acknowledgments! 🔴⚫

11/12 Shoutout to co-authors Shun Yonehara and Miyani Sonera (@kalamazoocollege.bsky.social students) for all the work running games and collecting data! 🙌

Paper here: authors.elsevier.com/a/1mUusArnpe...

10/12 The takeaway for scouting/analytics: we might be undervaluing players whose contributions emerge through team interactions rather than individual brilliance.

9/12 Caveats: 3v3 isn't 11v11, sample is collegiate not pro, and ~65% of variance was residual (opponents, randomness, etc.). But the pattern was consistent across two independent datasets (men's and women's teams).

8/12 For the men, teams with three different player 'archetypes' (e.g., goal scorer + catalyst + defensive specialist) outperformed less diverse teams. Complementary roles seem to matter, though this one needs more work to confirm.

7/12 We also found that players' rankings from these game contexts correlated only weakly with their scores on standard unopposed skill tests (passing, dribbling, shooting). Being good in drills ≠ being effective in games.

6/12 The result? Teammate combinations explained 20-23% of variance in goal differential. Individual player effects? Only 11-12%. Teammate effects were roughly TWICE as large. And this held for both men and women.

5/12 To get around this, we had 31 players compete in 3v3 matches where we systematically shuffled team compositions. 78 matches, lots of different teammate combinations per player.

4/12 Biologists have tools to separate 'nature' from 'nurture.' We used the same logic: how much of performance is the player vs. the teammates around them? (In biology terms, players = genotypes, teammates = environment, performance = phenotype.)

3/12 The problem is that in regular football data, the same players appear together over and over. You can't separate individual skill from teammate effects. So we borrowed an idea from quantitative genetics...

2/12 The question: what matters more for team success, individual talent or who you play with? Coaches talk about "chemistry" all the time, but it's been hard to actually measure.

Photo of 3-versus-3 football game

1/12 Really proud of this paper! We borrowed tools from biology to ask a simple question: does who you play with actually matter in football? (Spoiler: it does, a lot.) 🧬 🤝 ⚽

Paper here: authors.elsevier.com/a/1mUusArnpe...

Wow! In the ~1.5 years since we launched the EEB section of microPublication Biology, we have published 69 articles with another 10 in review or revision. 🎉 🧪

Thanks to authors and reviewers for supporting the effort! 🙏🏼

Social media scientific poster

Love it! We need more creativity in scientific presentations

people.kzoo.edu/santiago/ass...

as much as a SLAC person.

I started using E&E ideas to study soccer performance just because I thought it'd be fun. It's been surprising me how much more I'm enjoying doing my 'normal' fish work now, too, so there's an interaction effect there as well.

The incentives issue I agree with, but luckily it doesn't impact me ...

Mothers of disappeared people during three last dictatorship in Argentina holding posters of their kids’ faces during one of their typical marches

I was born in Argentina during the last dictatorship. This is *all* eerily similar to the way things were back then

career milestone unlocked: frantically finishing my conference talk hours before my presentation

It may not be much, but @kalamazoocollege.bsky.social is the only Michigan SLAC on this list...

They could also walk over to the expert on foreign policy in the department of American Studies

Fuck Tom, I'm so sorry to hear. Hugs to my fave Steady Eddy

🎉microPublications EEB is looking for new editors!

-Editorial load is extremely light (but we work quickly)
-Awesome format and journal, serves the EEB community well IMO
-We cover all of EEB so we're looking for breath/generalists/complimentary expertise
-DM me if interested (or suggest someone)!