If your test for ground-breaking discoveries can’t detect the discovery of RNAi, or of CRISPR-Cas9, or the cryoEM resolution revolution or Alphafold2, maybe it’s not a very good test. www.nature.com/articles/d41... - ThreadSky

scurry.bsky.social • 10 days ago

If your test for ground-breaking discoveries can’t detect the discovery of RNAi, or of CRISPR-Cas9, or the cryoEM resolution revolution or Alphafold2, maybe it’s not a very good test.
https://www.nature.com/articles/d41586-025-01548-4

Comments

resurgequality.bsky.social•7 days ago

An index of how often reviews of a field are cited. This thing is just confounded.

deniswirtz.bsky.social•9 days ago

Exactly

brooklynkid53bskys.bsky.social•9 days ago

yes yes a 1,000 times yes

I speculate that the issue is that these people are so dumb that they are measuring "progress" by tons of cement produced

dabiophysicist.bsky.social•8 days ago

It's the number of "focal" papers which are more root-like in the citation graph. IMO partly reflects volume, partly faster communication & more parallel efforts, also changes in culture of scholarship (we expect broad citation). All reflective of "focalness" and not impact

dabiophysicist.bsky.social•8 days ago

No paywall

laurentwada.bsky.social•8 days ago

How did this get past peer review?

dabiophysicist.bsky.social•8 days ago

Our paradigms have become so broad and flexible that they are rarely overturned even by the most impactful work

dabiophysicist.bsky.social•8 days ago

Under this CD metric the many chemotherapy publications of the 20th century are (conceptually) less "focal" because cisplatin was discovered in the 19th (1845!) and the therapeutic window in the 16th

dabiophysicist.bsky.social•8 days ago

(I say "conceptually" because the CD scores are actually calculated only within a relatively short temporal window around each publication)

easternblot.bsky.social•8 days ago

And not even human genome sequencing! Of course older papers are building more influence over time, but shouldn’t there be at least a blip around 2000 for all the projects that came directly out of the Human Genome Project?

wheelerlab.org•8 days ago

The measure appears to be: "if a paper is disruptive if people who cite it rarely cite its preceding work" (it's a citation bottleneck).

What if it's just gotten increasingly easy to get ahold of older papers and cite them? So, uh, library science and the internet disrupted the disruption index...

betsythemuffin.bsky.social•8 days ago

That would explain the deep dive in 1996 and why it's most prominent in the biosciences. PubMed launch date.

sjorsscheres.bsky.social•9 days ago

By measuring an average score over all papers, the plot may just reflect a growing number of papers with little impact. Would it be more interesting to look at the tails of the distribution?

scurry.bsky.social•9 days ago

I bow to your greater command of stats but aren’t ‘disruptive’ papers likely to be outliers, so isn’t looking at averages looking in the wrong place? The chart looks/feeld misleading to me.

sjorsscheres.bsky.social•9 days ago

Fully agree! That's why this chart is not interesting at all. It indirectly reflects the volume of fields, rather than their disruptiveness.

colebatty.bsky.social•9 days ago

Technology

cajackson.org•8 days ago

According to this plot the last groundbreaking paper in biology was in 1996

Must have been microarrays

ksbakes.bsky.social•8 days ago

Is this not also something to do with it being the inverse of the trend in overall scientific publishing volume? (With the qualifier that association is not causation, etc)

chloeroffay.bsky.social•8 days ago

adamjkucharski.bsky.social•10 days ago

The argument seems to be “disruptive = overturning prior hypotheses” (aka AlphaFold not “disruptive”) - but as scientific understanding about the world improves, it would make sense that there are proportionally fewer dominant-but-incorrect theories out there to be proved wrong?

scurry.bsky.social•10 days ago

Maybe so but it’s a rather vague definition and not clear to me that the method can detect what they’re looking for. Arguably, Alphafold2 overturned the working hypothesis that protein structure prediction was a fool’s errand! ;-)

kaplan.bio•7 days ago

And IIRC there metric for “disruption” was just a citation proxy which didn’t account for changes in citation practices. Like modern papers just cite way more than older papers

dabiophysicist.bsky.social•8 days ago

Nobody thought comparative modeling was a fool's errand, though!

scurry.bsky.social•8 days ago

If you mean homology modelling, then I agree. Used it more than once to create models of phasing X-Ray data by molecular replacement. But protein structure predictors had been over-promising since the 1970s.

dabiophysicist.bsky.social•8 days ago

Yes - "comparative" b/c there may not be an evolutionary relationship. I'm just saying the daunting folding problem approach shifted to sequence-based prediction, now to (highly sensitive, multi-scale, impressive, etc - but easier) comparative modeling with coevolution & structure data + templates

hallben.bsky.social•9 days ago

All other things aside, that’s a great way to frame alphafold successes

ijayas.bsky.social•8 days ago

This. Also, in our corner of the plant, they look at you like you’ve broken wind in an elevator if you describe yourself as a fundamental sciences researcher.

martinreich.bsky.social•8 days ago

Wow, what a nonsense is this yellow line please 🤨

arnau-de-torroja.bsky.social•7 days ago

The paper does not mainly seem to involve any method to identify ground-breaking discoveries, but rather to assess the evolution of disruptive papers in terms of how many were published each year, over a period of time, ending in 2010: before CRISPR and Alphafold2 made news.

scurry.bsky.social•7 days ago

Missed that - fair point, though the earlier discoveries/innovations should have popped out.

Comments

Posting Rules

Reply