Breck Yunits on Warpcast

Chris Dixon pfp

“The United States’ advantage is decentralized and open innovation. Some people argue that we must close our models to prevent China from gaining access to them, but my view is that this will not work and will only disadvantage the US and its allies. Our adversaries are great at espionage, stealing models that fit on a thumb drive is relatively easy, and most tech companies are far from operating in a way that would make this more difficult. It seems most likely that a world of only closed models results in a small number of big companies plus our geopolitical adversaries having access to leading models, while startups, universities, and small businesses miss out on opportunities.” Mark Zuckerberg - “Open Source AI Is the Path Forward” https://go.fb.me/ceql4u

24 replies

80 recasts

1453 reactions

Breck Yunits pfp

The math is simple: E=TA! Any society that continues to suffer from the "Intellectual Property" delusion will go extinct. https://breckyunits.com/eta.html

5 replies

1 recast

18 reactions

Cobe Liu pfp

If E (time for intellectual evolution) = T (time to test) A (size of assembly pool)! Wouldn’t a larger assembly pool increase time for intellectual evolution? You seem to be arguing that a larger assembly pool is good which I agree with but the equation doesn’t reflect that. It seems to be a little more complex. Maybe evolution of ideas does not have to occur after exploring every permutation of idea combination but can occur at any step in process and is more likely to occur if assembly pool is larger in scope and scale. So eqn looks more like: E=T/P(A) Where P(A) is the probability of finding coming up with beneficial idea and increases with A. Could be something like log(A). So E=T/(log(A)) Less elegant but lmk what you think.

1 reply

0 recast

3 reactions

Breck Yunits pfp

You are absolutely right. I'm not clear enough that as you add items to the assembly pool, you also increase the compute capacity proportionally. @nor also brought that to my attention. I still need to clarify a bit. Thank you for the feedback. https://github.com/breck7/breckyunits.com/commit/6c974491784c015c791bc0fcd4558c6764abad31

1 reply

0 recast

1 reaction

Breck Yunits pfp

ah, here’s my mistake. originally we called it EA!T, then flipped the last 2 for the better acronym, forgot about the divisor! will fix

1 reply

0 recast

2 reactions

Connor McCormick ☀️ pfp

Connor McCormick ☀️

ping me when you have the updated version, I was thinking about this when you messaged about it and I can see the argument for how if you have perfectly parallel trials then increased assembly size does reduce organism duration though maybe find a way to parameterize it by capacity for parallel trials

1 reply

0 recast

2 reactions

Connor McCormick ☀️ pfp

Connor McCormick ☀️

there's an interesting way in which this would explain why humans will be the dominant species for a shorter period than dinosaurs: intelligence is a faster and more parallel way to run trials, and so we can burn through the available assembly space faster, thereby arriving faster at a form more fit than us

1 reply

0 recast

2 reactions

Breck Yunits pfp

> thereby arriving faster at a form more fit than us wow, hadn't thought of this. thanks-- you gave me the thing to thing about on today's walk.

1 reply

0 recast

1 reaction

Cobe Liu pfp

Sick. My only food for thought is the num of parallel trials that can be run is limited by the number of minds/intelligences working on the problem not necessarily the number of of circulating ideas in the assembly pool. These two numbers will often correlate but aren’t the same. So may have to introduce variable representing number of intelligences working on the project and could possibly make A a function of this number and time. Let me know if this makes sense or if I’m missing something.

2 replies

0 recast

2 reactions

Breck Yunits pfp

Yeah I need to learn more Unified Field theory before I can make more progress here I think

1 reply

0 recast

1 reaction

Cobe Liu pfp

lol only on Warpcast do ppl go back and forth for a theoretical eqn on idea evolution

1 reply

0 recast

2 reactions