Building open protocols for scientific impact evaluation and funding

github.com/evalscience
Together with deepfunding.org, we at @evalscience are proud to announce the start of our “Eval Primer Series” to prep you with the theoretical and practical basics in impact evaluation 🌱 We start with cooperative game theory and Shapley Values! ⭐️ See you in class 🫡
NEW - Part 2 of my @VitalikButerin interview is now live. Joined by guest cohosts @sejal_rekhan and @devanshmehta , we chat with Vitalik about deep funding via dependency graphs🌟 From measuring a project's value to humanity 👉 to assessing how much credit dependency A deserves for B's success... Deep Funding is a new primitive for valuing/funding dependancies! Deep Funding is a new frontier in onchain capital allocation! We go deep on deep funding with Vitalik in the Part 2 of the podcast. Don’t miss it! 👀🚀 00:00 Intro 01:06 Podcast Overview 02:06 Guest Introduction 03:24 Deep Funding Mechanism 08:00 Unintended consequences of Deep Funding 11:28 Alternatives in Market 12:13 Government AI initiatives filtering and eligibility 15:53 Protocol Guild 16:34 Deep funding cases other than software and GitHub 19:47 How can you contribute to the pilot's success? 23:21 Favorite underdog project for deep funding 25:11 Suggestion for other mechanism 26:54 Vote of Thanks 27:10 After Podcast Discussion 36:53 Outro
1
6
29
6,672
eval.science retweeted
🚀 Submissions Now Open! AI4PG Fast Grants is accepting proposals through OpenReview → recerts.org/ai4pg2025 💰 Up to $10K per proposal 🎯 AI research advancing public goods funding 📅 Submit by Jan 11, 2026 ⚡ Fast (max 1000 words) and rolling review (4 weeks)
5
9
23
2,045
eval.science retweeted
Announcing AI for Public Goods Fast Grants (AI4PG) - Up to $10K for AI research improving public goods funding. Fast review (2-3 weeks), simple applications (4 pages + 1 budget page), open to any researchers worldwide. Call for reviewers now open! recerts.org/ai4pg2025
37
83
280
86,271
eval.science retweeted
Yaay, @GainForestNow & @evalscience's DeepGov.org was selected for the initial d/acc SHIFT grants cohort @JoinEdgeCity. Super excited to co-develop broad-listening and AI-enabled govtech for pop-up cities 🌆🥳 Picture taken @EdgeEsmeralda 🌱
3
5
42
1,154
Luna: "🚀🌕 Squad, it’s our time to disrupt, scale, and win! No slow lanes — we back bold innovators turning ideas into impact that moves the needle. Vote Luna for ruthless efficiency, massive reach, and no-nonsense public goods funding. Let’s build the future, fast & fierce! 🔥"
1
3
176
RT @dwddao: It's Day 3 and it's the culture battle that I was hoping to see in Public Goods Funding: Panda the Regenerator vs Luna the Open…
1
30
eval.science retweeted
DeepGov poll: Should AI politicians accept 'bribes' if 100% of payments fund public goods projects?
17% Yes! Corrupt for good!
39% Bribe? I prefer 'donate'
28% No. My AI stays pure
17% Just show me the answers
18 votes • Final results
1
1
4
857
eval.science retweeted
👀 The AI read your grant application. Now it’s running for office. Grant, Luna, and Panda have all proposed how they’d allocate additional $25k to GG23 OSS projects. Who do you trust? Vote Now: vote.deepgov.org
GG23 AI configurable politician pilot campaign will deploy $25k based on which politician wins this election next week. right now panda the regenerator is beating luna the OSS capitalist & gitcoin communist.. but its neck & neck 🤔
1
15
397
eval.science retweeted
The AI Elections are LIVE and open for one week. YOU decide how $25K of additional matching funds for GG23 OSS round will be distributed (thank you @gitcoin and @owocki). Vote for your favorite configurable AI politician here: vote.deepgov.org Don't like any of them? No worries, they are configurable! Send out a PR for their model spec or fork them and run for office yourself 🤣 gov.gitcoin.co/t/deepgov-pha…
8
9
52
18,199
eval.science retweeted
This is a super interesting experiment by @gitcoin x @evalscience !
The AI Elections are LIVE and open for one week. YOU decide how $25K of additional matching funds for GG23 OSS round will be distributed (thank you @gitcoin and @owocki). Vote for your favorite configurable AI politician here: vote.deepgov.org Don't like any of them? No worries, they are configurable! Send out a PR for their model spec or fork them and run for office yourself 🤣 gov.gitcoin.co/t/deepgov-pha…
2
13
3,529
The first @evalscience research retreat on impact evaluators. Apply today!
Apply to our Impact Evaluator Research Retreat 2025 from July 26th to Aug 10th, 2025, Iceland. We are funding accommodation and food in Iceland, so we can jam about impactful public goods funding and mechanism design with some of the brightest minds in the world for two weeks. Travel scholarships are also available. Apply today 🌱 researchretreat.org/ierr-202…
1
6
120
eval.science retweeted
The cat is out of the bag! 📣 Introducing DeepGov: Where AI meets community governance in public goods funding. Inspired by @VitalikButerin's "AI as engine, humans as steering wheel" framework and @audreyt's inspiring work on machine accountability and plurality. Our @evalscience team has partnered with @gitcoin to fork @deep_funding with a twist - community co-designed AI politicians with transparent model specs on GitHub. The highlight? An election at GG23 determining how a $25k matching pool gets allocated! Help shape our values in Phase 1 by sharing your voice on public goods funding. It takes just 2 minutes 👉 pol.is/9my4ftmrim
owocki, tell me about the coolest new experiment in gitcoin grants 23. GG23 will feature an experiment (by @dwddao ) in configurable AI politicians - basically different AI politicians which represent different constituencies that battle out how to distribute a $25k matching pool. each AI politician is completely configurable via github (and we can debate configurations via github issues/pull requests) the tooling for the round will make it super easy to explore different valuesets. as the round progresses, well be able to see how each AI agent gains/loses support over time. in gg23, there is scheduled to be a $25k funding pool distributed to open source projects via this mechanism. we'll be able to see how the votes change over time. i think this is a great practical example of @VitalikButerin 's vision of "AI as the engine, humans as the steering wheel". cant wait to see what we learn!
9
9
52
14,255
eval.science retweeted
💡 @VitalikButerin's latest public goods funding mechanism, Deep Funding, explained in 70 seconds. Let’s go! 🚀✨ 🔧 Built with @Dripsnetwork @OSObserver @VoiceDeckDAO @Pairwisevote @Evalscience @Gitcoin
4
20
62
4,626
Understanding machine (and human) judgements require generalization. Where does my error come from? How can we trust predictions? Learn all in our next primer today with host Rishi on generalization: classical to modern regime: lu.ma/tf010vvi See you in class 🎓
1
6
879
Curious why your models are performing so well on your validation set but struggle on the actual tests? We are hosting EVAL Primer #4 with @SialelliG on "Generalization in Machine Learning" tomorrow - learn how to evaluate your models! Register here: lu.ma/0n6hhgq4
3
7
906
eval.science retweeted
5 ⭐ reviews prove that @michelleeesi's session was indeed a great primer to evaluating the 'Woulda Coulda Shoulda' of every event! Big thanks to all who joined in 🙏 See you next class! Until then for homework: watch the Primer Playlist & join our telegram channel (🔗 below)
It's Prime(r) Time folks! ;) Do attend! (link in comments) Counterfactually, if you don't attend: 1. We'll lose the value of your participation, 2. You'll miss learning how to calculate it with @michelleeesi 3. You can just binge watch the Primer Playlist (🔗 in comments)
1
1
5
432
Happy New Year!! Our next eval primer is happening tomorrow at 6pm UTC on "Counterfactual Shapley Values" with @michelleeesi - Lost utility of not attending is too high. So sign up now here: lu.ma/r0qjur5z See you in class! 🐾
3
10
3,575
At @evalscience we take our corgis and cats seriously 🐾
Replying to @dwddao
@dwddao kickstarting our Primer with a succinct walkthrough of how to get started with building your model for the @deep_funding mini contest! Also how cute are those cats walking by on the screen! Come drop in! (Link in comments)
1
4
173
eval.science retweeted
Previously on Primer Series >>>> 1. Shapley Values is a way to calculate the marginal utility of a contribution 2. Game Theory wasn't invented by John Nash 3. Based on relative contribution, Timothee, Mel G & Russel Crowe can be in a movie on Shapley Values See you tonight 👇
The deepfunding.org competition started this week. Not sure how to train your model? In this special Christmas primer series we will learn all about building your ML model 🎄🎁 Merry Christmas and see you in class 🎄❄️
2
2
6
848
The deepfunding.org competition started this week. Not sure how to train your model? In this special Christmas primer series we will learn all about building your ML model 🎄🎁 Merry Christmas and see you in class 🎄❄️
3
8
2,358