Dávalos Lab

@labdavalos.bsky.social

Principal Investigator 👉🏽 https://lmdavalos.github.io/ Editor-in-Chief 👉🏽 @qrb.bsky.social/ recent Fulbright Amazonia scholar (https://www.fulbrightprogram.org/amazonia/), news from lab and beyond

I have spent the last day and a half reading and considering "Science: A New Golden Age" from OSTP. My short take home: this is not a serious science policy document. My long take went about 2,300 words and should be published tomorrow.

10. Hard to even know where to start in commenting on this report because it's such a jumble of grievances, lies, and accelerationist fever dreams. But this paragraph floors me. Not only is it talking about AI agents doing the science, it's utterly ignorant of the information economics of science.

n such a world, experimental information becomes a tradeable commodity,
and price mechanisms replace slow institutional coordination. Markets could
form to support the scientific enterprise, such as prediction markets informing
grantmakers about technologies on the horizon, bounty markets directing re-
sources toward unsolved problems, and reputation markets tracking which
agents produce reliable results. Agents would interact directly, exchanging data,
hypotheses, and compute time through microtransactions. The whole system
runs continuously, at speeds no human institution could match, but is guided by
human judgment about which breakthroughs merit large bounties, and which
questions require framing that machines cannot yet provide.

And guess what? Focusing on the welfare of your imaginary friends and away from the harm to actual people (and our environment, thus all living things) done in the process of building them is not improving that track record. >>