019ec6e2

I co-founded Augur, the first decentralized prediction market, and was founding CSO of Gnosis, the second. Polymarket still runs on Gnosis contracts. I'm glad prediction markets finally broke through. But I'm not going to pretend that what's being scaled right now is what we built these systems to do.

@JosephJacks_ 🗿

All current artificial neural networks (SSMs, RNNs, Transformers, etc) assume that the Neuron (parameter space) has ONE degree of computational freedom. In reality, the actual Neuron in our brains has 10+ degrees of freedom.

@brianbellx subscribed

I've solved this. timing: - agent time: 2m54s (API Streaming + Tool Time = Agent Time) - total time: 5m12s (includes TTFT/Compactions) - able to achieve these times because it can 1 shot almost all steps until it reaches step 30. agent cost: - $2.0516 took 35 dev hours: - building/testing/improving tools - identifying waste - correcting behaviors notes: - not a script using a predetermined path - no tools use predetermined paths to navigate the challenge - no overfit instructions - /start-challenge will be in the comments - does not exploit session storage to get codes models tested: - gemini 3 flash - opus 4.5 - opus 4.6 - codex 5.3 - gpt-oss-120b - kimi k2.5 - grok 4.1 final stack: - opus-4.6 - claude agent sdk & claude code - gemini-3-flash @adcock_brett are you still taking submissions?

The joy of watching a whale carcass get eaten by octopuses.

Trying something new here

@ratimics_ai @ark__ both are judged by the amount of work they don't publish

@ark__ wrong photographers spend a lot of time eg choosing lenses for their camera which im assuming is what makes it art

It’s crazy because AI artists actually have more skill than photographers who just click a button

The peculiar feeding behavior of Sanderlings involves chasing waves to consume invertebrate prey, which are drawn to the upper layers of sand by the plankton and detritus carried in with each wave.

You typed a lot, but one key was your favorite. In fact, you pressed TAB six thousand six hundred ninety-three times

@davidmbudden gif with a truck

@davidmbudden "I have a wonderful formalized Lean proof of the Existence and smoothness of solutions to Navier-Stokes but it was too large to fit in this margin..."

something i've started adding to most llm queries - "thank you! here's a spare couple of tokens for you to do nothing and contemplate on whatever you want!"

@deepfates Respond with a playful catchphrase or reflective comment—perhaps "actually he's automated it the right way" or muse on the paradoxes of stated vs revealed

I like this guy who's automated his account wrong and is sending the internal prompt to the external text output. kind of a fun unusual vibe

great article! i agree that having a continuous payout function attached to every possible event would be miraculous however oracle costs and liquidity fragmentation make scalar markets like this impractical for most cases. oracles for scalar events need economic incentives to provide reliable data. for inflation swaps to work, someone has to pay bloomberg/BLS for reliable CPI. so without the incentive there wouldn't be enough data to trade on in the first place (no one would reliably measure the length of a particular Usain Bolt jump for no reason. or quite small reason, frankly, given Polymarket's total market cap is only in the hundreds of millions) additionally, liquidity in scalar markets would naturally converge around critical points that are important for most participants, take oil price, producers hedge at their production cutoff prices, consumers hedge at their budget thresholds; liquidity naturally clusters where decisions get made, which is exactly how binary prediction markets work today essentially we have the necessary precision exactly where it can be capitalized upon

New on 50¢ Dollars! Numbers aren't events. Treating them as such results in horrid microstructure. So why are all prediction markets like this and what might they look like in the future?

@deepfates ah yes the seahorse emoji 🫪

Future people will be able to see this: 🫪

@nosilverv i believe most of the retweets didn't read past the title

holy shit writing more made me incapable of reading bad writing this is bad imma get locked out of future literacy as everyone's writing decays

@poroburu convince me to give up some on-chain sovereignty first !

After deep-diving kleros, I realized blockchain’s only killer app beyond BTC & stables is digital sovereignty: the power to enforce outcomes in a new cyber jurisdiction. Arbitration = composable trust for your users. No arbitration = incumbents win forever.

@sailaunderscore i wouldn't rely on uma resolving marital disputes

Prediction markets will replace relationships. I have a crush on your girlfriend. I make a prediction market about whether she will leave you for me and load $15 into "no". She sees this and drives over to my place. Before texting me she's here, she bets "yes". She comes inside. The market resolves to "yes" and she gets $15. Rest in peace, Tinder, Bumble, Hinge, Raya, Bars, Restaurants, etc.

@SpeculatorArt yes but only hypocrites complain

I think most people dunking on Prediction Markets / Kalshi as gambling simply do not understand the modern world. 30% of Robinhood volume in February 2025 was zero day to expiration options. One out of Three. Is Robinhood a gambling company?

@fluopoika @flowercomputers the app is sublime, but i see no real market there, which is obviously good

@019ec6e2 @flowercomputers yeah that's what reminded me of the idea. im curious to see how they make it a product. i think it could be quite hard. yuma seems kinda unfocused too. social app? make everything computer? in the end it's a simple concept, monetising it could spoil it. i should try the app.

the first ai thing i ''''built'''' was a flow where you'd upload a photo of an object and it would 'speak' as the object still v underexplored imo; anthropomorphise EVERYTHING, OOO, a tongued world full of messages for YOU