svara

5 ms·

61.
by svara 7mo ago
I think you're misreading me. My point isn't that you can't in principle state the optimization problem, but that it's much easier in some domains than in others, that this tracks with how AI has been progressing, and th
62.
by svara 7mo ago
The capabilities of AI are determined by the cost function it's trained on. That's a self-evident thing to say, but it's worth repeating, because there's this odd implicit notion sometimes that you train on some cost fun
63.
by svara 7mo ago
The vibe coding maximalist position can be stated in information theory terms: That there exists a decoder that can decode the space of useful programs from a much smaller prompt space. The compression ratio is the vibe coding gain. I think
64.
by svara 7mo ago
The funny thing about this is that even if the output is bad, it's actually good.
65.
by svara 7mo ago
Could you say more on how the tasks where it works vs. doesn't work differ? Just the fact that it's both small and greenfield in the one case and presumably neither in the other?
66.
Ask HN: How is AI-assisted coding going for you professionally?
434 points by svara 7mo ago|616 comments
67.
by svara 7mo ago
Hey, thanks, that was quite interesting! I'd be curious to hear your thoughts on how the "fixer", who sounds rather ineffective as an executive, came into this position, in what sounds like overall a rather effective organiza
68.
by svara 7mo ago
How large a demand for cars does the Chinese government have do you think?
69.
by svara 7mo ago
> I really don't see a solid economic future for Germany when enough other countries implement more progressive economic policies. People do change their minds when the pain becomes too intense to ignore, but that is what it takes.
70.
by svara 7mo ago
Yes, throughput is determined by the bottleneck and above a certain organization size, the bottleneck often is coordination costs.
71.
by svara 7mo ago
Yes, it's the lump of labor fallacy. Doesn't exclude the possibility of short term distribution, though.
72.
by svara 8mo ago
Can you give some examples of this? Maybe have something online? I would love to learn more about how to do proof driven AI assisted development.
73.
by svara 8mo ago
My understanding is that all recent gains are from post training and no one (publicly) knows how much scaling pretraining will still help at this point. Happy to learn more about this if anyone has more information.
74.
by svara 8mo ago
Do you keep a collection of these? What for?
75.
by svara 8mo ago
I think that mostly depends on how good a writer you are. A lot of people aren't, and the AI legitimately writes better. As in, the prose is easier to understand, free of obvious errors or ambiguities. But then, the writing is also nev
76.
by svara 8mo ago
Yes, it's been odd to observe the parallels with the web3 craze. You asked people what their project was for and you'd get a response that made sense to no one outside of that bubble, and if you pressed on people would get mad. Th
77.
by svara 8mo ago
Opus 4.6: Walk! At 50 meters, you'll get there in under a minute on foot. Driving such a short distance wastes fuel, and you'd spend more time starting the car and parking than actually traveling. Plus, you'll need to be at t
78.
by svara 8mo ago
You're right, and I enjoy using coding agents too. I've built some things with them I wouldn't have otherwise. However, it's been a full quarter now since November 2025. Based on facts on the ground, i.e. the rate and qu
79.
by svara 8mo ago
> So getting AI to do bits, really means getting AI to do the really easy bits. As someone who gets quickly bored with repetitive work, this is big though.
80.
by svara 8mo ago
> The wording here is fascinating, mainly because they're effectively acting as arbiters of "vibes". A very common tension in law everywhere. In the US you now have a 'major questions doctrine'. What the hell is
81.
by svara 8mo ago
One of the things about this story that don't sit right with me is how Scott and others in the GitHub comments seem to assign agency to the bot and engage with it. It's a bot! The person running it is responsible. They did that, n
82.
by svara 8mo ago
If past patterns are anything to go by, the complexity moves up to a different level of abstraction. Don't take this as a concrete prediction - I don't know what will happen - but rather an example of the type of thing that might
83.
by svara 8mo ago
More likely you would just train for emitting svg for some description of a scene and create training data from raster images.
84.
by svara 8mo ago
Software engineers (US based particularly) were more than happy about software eating the economy when it meant they'd make 10x the yearly salary of someone doing almost any other job; now that AI is eating software it's the end o
85.
by svara 8mo ago
It's really unclear unfortunately. The correlative effect is quite clear, i.e people who have high omega 3 levels (eat a lot of fish) have health benefits. But in random controlled trials Omega 3 supplements have not had convincing eff
86.
by svara 8mo ago
I mean, introducing a technique from one field in another is innovative. You don't get to claim you invented it, but a lot of progress happens by finding connections between things that are individually well known.
87.
by svara 8mo ago
I hear you, and that's where my mind goes first on this issue too. But with social media, many of the people most into it, when asked, will say they wish it didn't exist. A lot of kids feel they have to be on it, but wish it didn&
88.
by svara 8mo ago
The writing on this website is giving strong web3 vibes to me / doesn't smell right. The only reason I'm not dismissing it out of hand is basically because you said this team was worth taking a look at. I'm not looking f
89.
by svara 8mo ago
Everything's on a spectrum, but there's a point where you're so far along on the spectrum that it makes sense to call it something else. See, "quantity has a quality of its own". Sometimes you have to leave the theo
90.
by svara 8mo ago
> (...) you get these two camps 'for and against' whereas the best way forward is to insist on data rather than anecdotes. I think that's an issue with online discussions. It barely happens to me in the real world, but it&

More ›