My current & recent interests:

• reducing AI risk by making AI prove its outputs certain properties
  • this is what I spend most of my time on at Atlas Computing
• “revolutionary coordination systems”
  • This was the focus of Network Goods, the venture studio I ran at Protocol Labs
  • We thought about the overlap of mechanism design, social choice theory, and metascience

Prior work at Protocol Labs

What was Network Goods:

• I created and led a team (of at one point 25 people including contractors) with a budget of up to ~$7M/yr
• In 2 years, we started and spun out
  • hypercerts (non-profit)
  • Funding the Commons (PBC, conference series)
  • The Open Impact Foundation (for-profit)
  • gov4git (non-profit)
  • Open Source Observer (for-profit)
• We accelerated Discourse Graphs
• We supported davidad during the initial proposal of the Open Agency Architecture

What was PL Research:

I started the long-term research and metascience arm of Protocol Labs and ran it for ~5 years, which included

• creating the PL Research RFP program, advancing the cutting edge of networking, distributed systems, and cryptography research with $5M in grants to academic researchers. Also, hiring and managing a 3-person team to run the program.
• leading a team of up to 5 independent researchers nicknamed the AbstractionLab
• supporting Nadia Eghbal’s research and creation of Working in Public

Selected talks

• Oct 13, 2022 – DevCon 6 – Hypercerts for regenerative cryptoeconomics (30 min)
  • In which I present Hypercerts and motivate retrospective funding for public good (slides)
• Feb 12, 2022 – DeSci at Eth Denver – How to self-assemble an atomic bomb project (15 min)
  • In which I provide context on the field of metascience to people from crypto and the Decentralized Science (DeSci) movement. (slides)
• June 23, 2023 – Summer of Protocols – Putting protocols into practice
  • In which I talk about Filecoin (and FIL+), the projects of the Network Goods, and the open agency architecture, not as products, but as protocols.
  • slides
• Nov 11, 2022 – Santa Fe Institute – Emergently Engineered Economic Experiments transcript
  • In which I present Filecoin (and FIL+), SourceCred, and Hypercerts as emergent economic experiments(slides)

Education

• In 2017, I completed my PhD at Caltech as Andrei Faraon’s first grad student, where I was working on a scalable optical quantum memory (thesis).
• My formative college years included a BS and MS in Materials Science and Engineering at Stanford, lots of courses in math, CS, and philosophy, and perhaps a surprising amount of set design and construction.

Published writings

• See my Google Scholar profile for a fairly comprehensive list of my academic writings.
• I also frequently reference this blogpost I wrote about the SourceCred project’s method of providing algorithmic impact measurment of open source software in an effort to create permissionless employment.
• I think this blogpost is still a pretty comprehensive summary of why we need Discourse Graphs as a micropublication standard for immediately and continuously open science.
• And I feel like this list of questions (for new grad students picking a group) have stood up pretty well since I was a grad student.

Little things

Real-world backlinks from people or insitutions I deeply respect (roughly chronologically)

• I’ve gotten multiple shout-outs from 3blue1brown (am I a hand model?)
• I’ve been thanked in multiple pieces by Michael Nielsen
• Vitalik has publicly name-checked two projects started on my team (hypercerts and Open Agency Architecture)
• I’ve organized a metascience workshop at the Santa Fe Institute
• Andrew Critch and davidad jointly suggested I start what is now Atlas Computing

Contact

If you want to talk about something related to Atlas Computing, use the contact link or email on the atlas website (I’m on the other end of it).

If you want to talk about something NOT related to my current job, email me at [first name].[second half of last name]@gmail.com but note that it might take me a while to respond.

I’m also probably more responsive to LinkedIn messages and twitter DMs than email thanks to beeper, but I tend not to accept LinkedIn connections that don’t have a message unless we’ve met.