sobellian

6 ms·

91.
by sobellian 6mo ago
The 'confession' is a CYA. Honestly the whole story doesn't really make sense - what's a "routine task in our staging environment" that needs a full-blown LLM? That sounds ridiculous to me. The takeaway is we c
92.
by sobellian 6mo ago
More like "I'm going to run every possible marathon route on the Earth's road network."
93.
by sobellian 6mo ago
The thing about the Russo-Ukrainian war is that it is a failure for both sides. The primary lesson from this war is, how do we avoid ending up like those poor guys? If the US Army fights a war with anyone, let alone China, on the doctrine t
94.
by sobellian 6mo ago
Would it be a vote of no confidence in Full Self Flying?
95.
by sobellian 6mo ago
Machine eps provides the maximum rounding error for a single op. Let's say I write: let y = 2.0; let x = sqrt(y); Now is `x` actually the square root of 2? Of course not - because the digit expansion of sqrt(2) doesn'
96.
by sobellian 6mo ago
Back in my college days, embedded control was required for both ECSE and mech. E.
97.
by sobellian 6mo ago
IMO AI will make plain the divergence between "good design" and what people actually want. You're absolutely right that from an artistic perspective, it will produce the heat death of UI. I just struggle to think if teams bui
98.
by sobellian 6mo ago
This, plus the alchemical nature of these tools, seems to have made users pretty paranoid (I admit I am also guilty of paranoia). Maybe there's room for a Standard AI - we may change the prices based on market conditions, but we always
99.
by sobellian 6mo ago
Their issues seem to extend well beyond inference into services like auth.
100.
by sobellian 6mo ago
Execution is very important. Startups almost never start with the "right" idea.
101.
by sobellian 6mo ago
"This thing might destroy humanity - we need to build it ASAP" does not really make sense. But it enthrall[s/ed] many smart researchers who would normally demand specific, testable claims and logical responses to those claims
102.
by sobellian 7mo ago
I didn't make any claims contradicted by literature. The only thing I cited as bedrock fact, NFL, is a mathematical theorem. I'm not sure why Nim shouldn't be relevant, it's an exercise in logic. > “AlphaZero excels a
103.
by sobellian 7mo ago
Hey man, it sounds like you're getting frustrated. I'm not ignoring anything; let's have a reasonable discussion without calling each other ignorant. I don't dispute the value of these tools nor that they're improvi
104.
by sobellian 7mo ago
The space of programs is incomprehensibly massive. Searching for a program that does what you need is a particularly difficult search problem. In the general case you can't solve search, there's no free lunch. Even scaling laws mu
105.
by sobellian 7mo ago
Scaling laws vs combinatorial explosion, who wins? In personal experience claude does exceedingly well on mundane code (do a migration, add a field, wire up this UI) and quite poorly on code that has likely never been written (even if it is
106.
by sobellian 7mo ago
If the payload is a credential stealer then they can use that to escalate into basically anything right?
107.
by sobellian 7mo ago
IIRC there's a video by 3b1b that talks about that, and it is important that gaussians are closed under convolution.
108.
by sobellian 7mo ago
I think the article in fact reaches the exact opposite conclusion it should. I'm not really sure how useful it is to talk about sharing and commons and morals when the point raised was about what is possible. The prescription includes
109.
by sobellian 1y ago
SpaceX uses flight proven boosters. The rockets aren't quite as gigantic nor as single-shot as the Saturn V. Also, they launch satellites into LEO for commercial reasons. It's quite a different beast from lobbing LEMs at the moon
110.
by sobellian 1y ago
I'd say material science since the only non-ablative material we can use is too brittle compared to a normal fuselage. I really hope they succeed but it's a pretty fundamental problem to have unanswered this deep into the program
111.
by sobellian 1y ago
The space program stalled because pouring national wealth into gigantic single-use rockets was unsustainable. They tried with Shuttle but the material science wasn't there yet (heck it might not be even now, it doesn't seem that t
112.
by sobellian 1y ago
Having done this, and acknowledging the fuzzy definition of "rich" etc - going through YC after graduating is a great career move. You have to do everything yourself so you learn a lot. About business, about a specific industry, a
113.
by sobellian 1y ago
There is so much work for skilled tradesmen that they would rather see more automation so they can take more jobs. Even many unions, e.g. carpenters' unions, think this way. I mean no large group is a monolith so I'm sure one can
114.
by sobellian 2y ago
> It's very difficult to find some way of defining rather precisely something we can do that we can say a computer will never be able to do. There are some things that people make up that say that, "While it's doing it, wi
115.
by sobellian 2y ago
He didn't topple the system but it wasn't for lack of trying. Pence had to refuse Trump's repeated requests to fix the election (a precedent that would have guaranteed single-party rule). You are relying now, as you were then
116.
by sobellian 2y ago
Okay but we were talking about the unemployment statistic in this thread. Does it add any information? It likely does not.
117.
by sobellian 2y ago
The proposed measure is highly correlated with U-3, so as time-series they should basically tell the same story. If the assertion is "U-3 doesn't predict this phenomenon but this other measure does" it's likely to be wro
118.
by sobellian 2y ago
AFAICT the article was written by the guy that germinated the concept in the first place. You can see the paper at https://cdn.prod.website-files.com/63ba0d84fe573c7513595d6e/... tl;dr It is very highly correlated to U
119.
by sobellian 2y ago
In practice it would be very difficult to predict RDRAND outputs. Even so I believe the truly paranoid can use RDSEED to skip the PRNG step. Not qualified at all to talk about how they de-bias the measurements.
120.
by sobellian 2y ago
> Like, life in USA became more or less good only several generations ago, after the country became giant economical winner of WW2. This judgement of course depends upon the standards of the observer and where in the US you look. Before

More ›