@kawillis
Systems biologist, data scientist, policy wonk, runner. Former NIH-funded investigator, former NIH program official. Founder, Woodley Park Institute. Writing about understanding and accelerating scientific progress at https://theprogressbarwpi.org
Peer review can tell you if something is a sincere effort to engage with a problem and improve our understanding of the question under study. It can surface inconsistencies and suggest new avenues to pursue.
It isn’t a certification that a paper, a grant, or an idea is error-free.
This is the problem with arguing that we need peer review to determine what’s correct.
It’s important that we don’t allow sky-is-falling narratives about loss of trust in science and medicine to drive us into a bubble of self-fulfilling prophecy
Important to remember: a supermajority (67%) have confidence in career scientists at federal agencies
86% have confidence in their primary care provider
NIH Office of Extramural Research announcements of feedback sessions for NIH-wide strategic plan for March 16 and April 8.
IMPORTANT---IMPORTANT---IMPORTANT
NIH is seeking input for the NIH-wide Strategic Plan. Open to everyone but Registration is Required!
Your chance to listen and get your 2 or more cents in.
grants.nih.gov/news-events/...
dm me if I can help
ok I’ll bite, way out of my wheelhouse too, but an R01 is an R01, and I can flag general issues with the shape
My kids can verify. It's a running joke in our house.
Even if, most of the time you arrive at the same place everyone else did. 😅
I suspect it’s culture independent; I suspect it’s that humans are a social species, and Chat lights up those neural pathways. For some people more than others.
I quite like an audiobook and a podcast myself, actually .. great for doing chores or on a long run. I am admittedly a multi-tasking addict, although I probably shouldn’t be. But personally I could never uptake scientific literature that way. Not enough space for cogitation.
I think I can see something similar with Chat. It’s slipped into that “ask-down-the-hall” niche
… and it’s easier, to only look for and accept the collective wisdom of the tribe. But it’s limiting. Great discoveries imo come from seeing what the tribe missed, and you only get that by going out into the wilderness alone.
This is something I feel like I’ve seen amongst scientists pre-Chat; that desire to ask the authority figure down the hall for an Answer, rather than wrestle out an understanding from the literature on your own. Some more so than others. I get it, it’s faster..
I saw something recently (unfortunately didn’t bookmark) about how for most of human history, knowledge was transmitted orally, and hypothesizing that to some degree, humans are hardwired to have this preference. We want, maybe, to assimilate knowledge through conversation, not books.
Yes! We literally cannot shut up about our work.
This is legit why our method for detecting breakthroughs works. When a whole bunch of scientists are suddenly and simultaneously unable to shut up about the same thing, look out.
.. streamlining manufacturing, refining existing treatments, repurposing approved drugs.
Good stuff, but what if there was an LLM-independent way to direct early-stage investments to the research areas most likely to bear fruit? Right now, that's the work that occupies me.
Finally, and this is super interesting to me, there's a great discussion of the challenges associated with choosing a research direction that will produce a return on investment. My takeaway is that pharma has concentrated on making the economics of discovery work by optimizing the later steps ...
Third, Thompson's guest, Eli Lilly CEO Dave Ricks, tells us that the 2025 budget for R&D at his company was $14 billion. In comparison, the FY2025 budget for NIH was $48 billion.
NIH is indisputably a major player in this space (although I'm curious what the sector's total investment was in R&D).
Second point, and I am surprised by how rarely I see this: one reason LLM-based AI agents are not very good at promoting breakthroughs in biomedicine is because our knowledge of biology is still so sparse. You can't connect and remix what isn't there.
Great piece by @dkthomp.bsky.social on GLPs, AI, and how we make progress in biomedicine. First thing to notice: without early studies on the biology of anglerfish and gila monsters, this trillion-dollar breakthrough likely wouldn't exist.
open.substack.com/pub/derektho...
I am both laughing and weeping
Please don’t get me started on the lack of coherent definitions in research administration 😩
Hope you’re also enjoying the food! Go to Morning Call for beignets like a native! For dinner I recommend Irene’s.
For no particular reason, I find myself thinking this morning of this recent @davidlazer.bsky.social paper, suggesting that seeing themselves represented amongst scientists improves trust in science for a variety of demographics
www.nature.com/articles/s41...
Ah, the good old fashioned divide between programmatic intent and programmatic reality.
It is … not an easy listen
listening now
From 1981.
"Chance and Consensus in Peer Review"
www.science.org/doi/10.1126/...