Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
vessenes
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
16 ms
·
421.
▲
by
vessenes
7mo ago
Most reasonable analysis here! But you forgot: “retail loves it and buys it, providing capital sufficient to stabilize.” I also like SpaceX - one thing many of the kids around here seem to forget is that elon has managed extremely dire capi
422.
▲
by
vessenes
7mo ago
I like this idea on expert streaming. I've been poking around fairly thoroughly at the same idea - can we fix a set of experts? when can we fix them? How long is the top-k selection "good" for in terms of number of forward pa
423.
▲
by
vessenes
7mo ago
Cheapest just isn't a very useful metric. Can I suggest a Pareto-curve type representation? Cost / request vs ELO would be useful and you have all the data.
424.
▲
by
vessenes
7mo ago
> The suggestion engine (korb suggestion threshold) is re-implemented in Lean 4 with five mathematically proven properties: suggestions have positive frequency, are sorted descending, come from ordered and available products, exclud
425.
▲
by
vessenes
7mo ago
Here to say -- great name. It's not just a reference to our modern times, it's a sign of brilliance. (I wrote this myself with no clanker support)
426.
▲
by
vessenes
7mo ago
Friend, this is just not true. Thousands of years of large sophisticated governments and civilizations in what is now Mexico and farther south. Its reductive to the point of being completely wrong to think of the Americas in this way. Teoti
427.
▲
by
vessenes
7mo ago
Oooof. This is going to take a lawyer. A team of lawyers. Everything on that account has no doubt been LOCKED down for CSAM-grade review. I'm just imagining how you'd get a judge to help you here. You'd need to have found an
428.
▲
by
vessenes
7mo ago
I haven't seriously looked at Bitcoin's PQ plan for a couple of years, so I might be (I am almost certainly) out of date, but my recollection is that there's a "pre working attack" phase required, in which everyone
429.
▲
by
vessenes
7mo ago
> Is this market manipulation? No > why do they think that the US government would care about securing cryptocurrencies? Our largest institutions manage tens of billions of dollars in cryptocurrency and the US government has designate
430.
▲
by
vessenes
7mo ago
there are some different options depending on settings - apple will encrypt to an internally (apple held) key that your iCloud login will unlock under most circumstances. This can be turned off by consumers, and I would expect by IT departm
431.
▲
by
vessenes
7mo ago
Thanks -- btw the Pricing link on the site pulls up a form, not that page.
432.
▲
by
vessenes
7mo ago
cool. What's pricing like?
433.
▲
by
vessenes
7mo ago
They export something else.
434.
▲
by
vessenes
7mo ago
That is the point of cheapening the dollar, BTW. The local wages can stay 'high' dollar denominated, but the euro-denominated value of those wages drops. It was for some time the strategy of the Chinese central bank; you can keep
435.
▲
by
vessenes
7mo ago
Yes, a log chart would be better. That said, apples cannot be compared in this case; probably very few of us would choose to go back to 1914. A Tesla model Y would cost $1,680 in 1901 dollars, but would have been worth millions of those sam
436.
▲
by
vessenes
7mo ago
It was designed at Apple, and is sold and marketed and branded by Apple. What do you consider it? Apple infrastructure?
437.
▲
by
vessenes
7mo ago
Hypercard, at 50th, is underrated. We're all here because of Hypercard.
438.
▲
by
vessenes
7mo ago
obligatory - that first famous private server was done because someone wanted a blackberry like Obama had, and was told no by NSA. Man that BB keyboard was good.
439.
▲
by
vessenes
7mo ago
I prefer claude for generation / creativity, codex for bull-headed, accurate complaining and audit. Very rarely claude just doesn't "get it" and it makes sense to have codex direct edit. But generally I think it's h
440.
▲
by
vessenes
7mo ago
Not sure why you'd want this over an apple setup. M4 max is 545GB/s of memory bandwidth - $2k for an entire Mac Studio with 48GB of RAM vs 32 for the B70.
441.
▲
by
vessenes
7mo ago
Wrong question. I suggest: 1) Do models generalize? 2) If they do, and they generalize from this, is that a win? Chollet was one of the first “they do not generalize” evangelists. I’d be curious to hear what he thinks now, because a) most d
442.
▲
by
vessenes
7mo ago
I’m not a Chollet booster. Well, I might be a little bit of one in that I admire his persistence. I really like these puzzles. There’s a lot to them both in design and scoring — models trained to do well on these are going to be genuinely m
443.
▲
by
vessenes
7mo ago
Most important comment here!! I would love to buy like a 6kw battery induction kettle for the states. As far as I can tell, they don’t exist.
444.
▲
by
vessenes
7mo ago
Back at you! Writing's always work, especially with cool graphs. 1 - Agree this is true for real-earning-loss countries, recently. (as in last n decades) 2 - I agree this will keep shrinking. I moderately disagree with the 'compar
445.
▲
by
vessenes
7mo ago
Thanks -- interesting. I like the idea of ablating layers. I guess you could get a differentiable stack that has a layer skip and layer copy/loop and a total memory use loss function; that would let someone ship either a big (usually a
446.
▲
by
vessenes
7mo ago
David, Thanks for this research. I remember being stunned when Goliath showed up and .. worked; this feels like under explored research right now. I've been thinking about implications of this for local generation -- what's really
447.
▲
by
vessenes
7mo ago
Interesting points, made in florid style. I'd make the point a little differently, though -- in the next n years, you want to be on the side of capital. Labor is in for a really tough time. Missing from Daniel's analysis is total
448.
▲
by
vessenes
7mo ago
It seems almost impossible. I spent the weekend comparing nanoclaw to openclaw - nanoclaw is a slightly more secure version - containerized filesystem basically - and very popular. It's a) harder to setup, b) less functional out of
449.
▲
by
vessenes
7mo ago
They will. Either new architectures will come out that give us greater efficiency, or we will hit a point where the main thing we can do is shove more training time onto these weights to get more per byte. Similar thing is already happening
450.
▲
by
vessenes
7mo ago
Yes, yes it is. And it's amaaaazing. We're going to have lots of sharp edges getting stuff like this secured, but it is not going to go away. Too useful.
More ›