Interesting how irrelevant Binance has become to the crypto trenches. None of the major runners are listed on Binance and all price discovery happens onchain. And perp dexes taking their perp volumes too. Plus, tokenized stock meta requires licenses and regulations that Binance lacks so their regulatory arbitrage isn’t paying off in this meta. CEXs are mostly good for fiat onramp/offramp. But even for this Binance isn’t an option in the US and the EU.

Interesting relative strength here from $PONS compared to $CASHCAT and $AI Think we might see a decent catchup bid from both, especially $AI over the coming days if this move from $PONS is sustained. Still believe $CASHCAT breaking ATH will lead the next wave of attention on RH chain which takes $PONS to $1B+

Interesting to contrast the relative silence on my feed about a 4% move up in gold miners today with the cacophony on my feed that accompanied a 5% move up in BTC yesterday. 🤔

Interesting work on long-horizon research agents. PrimeScientists can decide where a research agent spends its budget. Achieves 10.3% more reward with 50.6% fewer research attempts, under the same budget. That is PrimeScientist against AutoResearch on 12 AI research tasks. They treat deciding where to spend a research agent's budget as part of the agent's job. PrimeScientist keeps an executable plan tree of competing research directions and their outcomes. An adaptive MCTS policy reads the experimental feedback and the remaining budget and chooses whether to explore a new direction or continue a promising one. The gains also hold on systems, code optimization and ML engineering tasks. If your research agent can propose more experiments than you can afford to run, this is a concrete method for choosing among them. Paper:

Interesting choice of words by .@FredLambert to use “crashes” in an automotive article when nothing crashed. It feels deliberately provocative and tells you a lot about the framing the writer chose.

Interesting to see how the AI companies are trying to make their apps feel more human, less alien. Muse has this cute little animated guy smiling all the time, ready to help you with anything. It feels friendlier and more relatable than a blank, sterile screen with an input. ChatGPT Computer Use shows a moving cursor as it’s working. It’s not technically required, but it makes the whole process less scary and easier to follow. It’s like the early iPhone days when things used to immitate real world objects so that an unfamiliar technology felt more intuitive and relatable. In this case we imitate presence, trying to give the AI a more human feel.