aerospades

10 mesi fadiscuss

I disagree with the author's thesis about data scarcity. There's an infinite amount of data available in the real world. The real world is how all generally intelligent humans have been trained. Currently, LLMs have just been trained on the derived shadows (as in Plato's allegory of the cave). The grounding to base reality seems like an important missing piece. The other data type missing is the feedback: more than passively training/consuming text (and images/video), being able to push on the chair and have it push back. Once the AI can more directly and recursively train on the real world, my guess is we'll see Sutton's bitter lesson proven out once again.

aerospades

anno scorsodiscuss

> In comparison, the parallax method was far less accurate, locating New Horizons within a sphere with a radius of 60 million kilometres, about half the distance between Earth and the sun.

NASA SEXTANT mission demonstrated pulsar navigation to about 10km of error, and should be valid through interstellar space. This parallax method seems not really in the ballpark? Still pretty cool they are able to teach an old dog new tricks given New Horizons launched in 2006!

aerospades

anno scorsodiscuss

I'm not sure if that's a sarcastic pitch for SLS or shuttle, but both of those have been proven dead ends.

I feel like the overarching Starship design could be adapted to accommodate larger engines if SpaceX found an optimization there. What they need is the culture and financial runway to pivot/adjust enough to eventually get to a better optima. Which SpaceX seems to have.

aerospades

anno scorsodiscuss

The Falcon architecture is objectively the best rocket that has been designed and flown (most launches, most successes, cheapest). Either that is roughly most the optimal rocket architecture that is possible within known physics, materials, and Earth gravity, or there is something better. Those are the only two options.

Starship is SpaceX's bid for the next best thing to obsolete Falcon. It might not succeed, but the alternative is what? SpaceX just sits around, mirco-optimizes Falcon forever? At least they are trying. The potential is there, and the iteration is what smokes out issues in complex systems much better than analysis paralysis.

Let's check back in on this take in 10 years.

aerospades

2 anni fadiscuss

George Hotz calls this "the button" on his last Lex podcast. <https://youtube.com/watch?v=dNrTrx42DGQ&t=5247>

"This is Mark Zuckerberg reading that Paul Graham asking and being like, I’m going to show everyone how alive we are. I’m going to change the name [from Facebook to Meta]. Does Sundar Pichai have the authority to turn off google.com tomorrow? They have the technical power, but do they have the authority? Let’s say Sundar Pichai made this his sole mission. He came into Google tomorrow and said, 'I’m going to shut google.com down.' I don’t think you keep this position too long. Now, does Mark Zuckerberg have that button for facebook.com? I think he does. And this is exactly what I mean and why I bet on him so much more than I bet on Google."

comments