DARPA
Article
DARPA is a recurring organization in the Astral Codex Ten archive, appearing 4 times across 4 issues between May 20, 2021 and February 07, 2025. The archive places it in contexts such as “In 2009, DARPA created a digital tutoring system”; “DARPA investigates how prediction markets do vs. expert surveys”; “has lectured at DARPA”. It most often appears alongside 1DaySooner, Bryan Caplan, COVID.
Metadata
- Category: Organizations
- Mention count: 4
- Issue count: 4
- First seen: May 20, 2021
- Last seen: February 07, 2025
Appears In
Related Pages
-
- 1DaySooner (2 shared issues)
-
- Bryan Caplan (2 shared issues)
-
- COVID (2 shared issues)
-
- FDA (2 shared issues)
-
- Metaculus (2 shared issues)
-
- OpenAI (2 shared issues)
-
- Peter Thiel (2 shared issues)
-
- Substack (2 shared issues)
-
- Twitter (2 shared issues)
-
- Tyler Cowen (2 shared issues)
-
- Vitalik Buterin (2 shared issues)
-
- 2016 Washington carbon tax ballot initiative (1 shared issues)
External Links
Source Context
Recovered passages from the original issue text. When the raw archive preserved outbound links inside the source passage, they are listed directly under the quote.
7: Best of Less Wrong: DARPA Digital Tutor: Four Months To Total Technical Expertise? In 2009, DARPA created a digital tutoring system that could adjust lessons based on students’ strong and weak points. After four months, digitally-tutored IT technicians outperformed experienced professionals in DARPA’s tests. How is this different from existing digital learning software, and could we make equally successful programs for other subjects?
— DARPA investigates how prediction markets do vs. expert surveys when guessing the results of social science studies. Answer: neither of them does well. Some suggestive evidence that averaging the price of the prediction market over a while does better than taking the final price, at least in these very non-liquid markets.
Jimmy Koppel, $40,000, to support his work on intelligent tutoring systems. We know 1-on-1 tutoring is the best way to learn, but human tutoring doesn't scale to the number of students who need it. Computer tutoring systems can ask questions, identify areas where people need to improve, and notice/respond to specific error patterns. I was originally skeptical about this but reading things like this essay have gotten me excited. Pure AI tutoring is hard because "it takes 300 hours to develop 1 hour of intelligent tutoring system curriculum", so Jimmy is working on a hybrid model where computers do lots of the work but there's still a human in the loop. Jimmy has a PhD in computer science from MIT and currently runs a company doing advanced training for professional software engineers.
Stuart Buck, $50,000, to help launch the Good Science Project, “a science policy think tank that will focus on essays, blog posts, videos, and other public advocacy about how to improve science funding in the US.” Buck was VP of Research at Arnold Ventures, helped start the Center for Open Science, and has lectured at DARPA and IARPA and written pieces for Science and Nature. You can read more about his philosophy of science funding here or follow @GoodSciProject for updates.
Inline links: here, @GoodSciProject
Release ARPA-H from the NIH: Bhattacharya likes unorthodox things, and the most unorthodox thing you can do in DC is to deliberately decrease the size of your empire. ARPA-H is an innovative government science funder modeled after DARPA. Although it was intended as an independent agency, it got placed within NIH due to bureaucratic machinations. Now it’s in danger of getting shoved into a new National Institute on Innovation and Advanced Research with a lower budget. Separating ARPA-H from the NIH will protect it it from this fate, help it deliver on its intended goals, and help NIH reducing the number of institutes and centers it oversees (granting more research dollars per center).
Inline links: ARPA-H, intended as