18 chunks · 17 episodes
...at to cozy communities. Cozy communities have high trust and expect discretion. Higher quality discourse can happen in cozy communities, but it also means that insights are not necessarily shared outside the group. The insights are meaningful s...
...gained that would have made this run easier. This makes the next run faster and higher quality. This is where the feedback loop closes and becomes a meta, compounding loop.[jr] That LEARNINGS.md is a file that a human can help curate and steer....
...ey run if they have tools to ground themselves (e.g. tests and playwright), the higher quality things they have for you to review when it's time. Safe rooms for agents allows autonomy, where if they explode there's limited collateral damage.
...e lack of curation. Collaboration is good--because you can curate and distill a higher quality output. But consensus is "collaboration with no curation". You get mush: the average of distinct individually viable perspectives is unlikely to be v...
...oes on, with feedback and curation throughout. The continuous curation keeps it higher quality. The append-only nature of chats is really limiting. A conversation that goes off the rails you have to restart.
...his is also why feature length animation has a slight tendency to be on average higher quality than live action. It's so hard to produce it requires more planning and workshopping before it gets to production. With LLMs making the actual coding...
.... There was something the user could do to meet the system where it was and get higher quality output.
...ls to specific use cases, or a carefully calibrated UX flow to help squeeze out higher quality React apps out of models. But then a new model comes out that was directly trained on doing that use case well and it leapfrogs all of your linear im...
In low-friction environments, higher quality things have compoundingly better outcomes.[akv] The seed crystal exists because the first few people who looked at it found it useful. The boundary g...
...s that scale test-time compute (e.g. the approach that O1 and others use to get higher quality reasoning) as a savvy technique to wring more baseline knowledge out of a system. LLMs never get bored, and never run out of ideas; if you give them ...
...og DNA to fill in any ambiguities. But as the model improves, the frog DNA gets higher quality, and the outcome gets better. The quality of the output depends on the quality of the frog DNA which is dependent on the overall quality and ability ...
...w a single model provider, they know it. And although other models arguably are higher quality now, they aren't an order of magnitude better. OpenAI has a significant lead in subscriptions, meaning users will stay out of inertia – why try the o...
...e new math AI paper, shows that the more formalizable the system and rules, the higher quality the training, the better the inferences.
...inct! Being higher absolute quality will tend to make it more obvious that it's higher quality, but not always.
Things people do with their discretionary effort tend to be higher quality per unit effort. When you are compelled to do something, you need to clear the bar of the person who has compelled you to do it. Satisficing mindset....
... answer it, the answer will be of their own discretionary effort. So it will be higher quality than if it were compelled.
...a very small number of people use the secondary use case, that trickle makes it higher quality, which pulls in more people: a compounding loop. Some secondary use cases have such a strong network effect that over time they grow to eclipse the p...
...s much time as you want. So a benefit of this approach is that you can also get higher quality output. Much less agility and no power of the swarm, but like a steamroller reliably producing quality of a given level.