Analemma_

What evidence would count? Obviously any dangerous misalignments are going to come from the frontier labs first, because by definition they're the farthest ahead. If nothing they say can ever count as evidence for misalignment it's hard to see how anything ever could.

Just because you can, doesn’t mean you should. Approximately nobody thought Google’s current emoji family needed a total overhaul, stop breaking our pattern recognition for no reason other than your designers are bored and don’t have enough real work to do.

The person you sold it to after borrowing it, who paid for it at the elevated price.

- AI investment is basically the only thing keeping the American economy treading water at the moment, and kicking the chair out from under that industry benefits China tremendously.

I'm definitely not saying that's the only factor, but I think it's naive to assume it isn't at all a factor.

The LessWrong people hate OpenAI, they all back Anthropic.

I cannot think of a less trustworthy source of information than Emil Michael on the All-In podcast. Why on Earth would you take his statements at face value? I grant that nobody in this situation is totally trustworthy, but if it's his word against Dario's I'm believing Dario 10 times out of 10, and it's astonishing and dismaying that anyone might do otherwise.

Keep in mind that Emil Michael's job at Uber was to slander and destroy their political enemies. What makes you think he hasn't continued exactly as before?

What does the relative level of government spending versus taxation have to do with whether businesses will self-regulate? You're just spewing non sequiturs here.

The fact that "the name on the ticket must match your photo ID, but you can resell at any time before the event" would instantly solve scalping problems and yet nobody does it is your clue that nobody actually wants scalping problems to be solved. Scalpers are the sin eaters for Ticketmaster and the events themselves, taking the hate so the performers don't have to.

Oracle Cloud sometimes feels like an elaborate prank that I'm not in on. I know people and companies on AWS (obviously), Azure, Google Cloud, Hetzner, CloudFlare's various PaaS offerings, etc., but I can't name a single thing running on Oracle Cloud. Somebody out there is clearly using but I'll be damned if I know who it is.

DNS servers never see subpaths you request, only the domain itself, so that wouldn’t help with a hidden path. But there are lots of other ways to get it: caches/CDNs can leak paths, Chrome presumably sends Google a bunch of request details, and so on.

It’s a different story if it’s a subdomain though, OP wasn’t clear.

I don't think the parent is being tongue-in-cheek, and even if he is, plenty of people have expressed that sentiment seriously. Robin Hanson famously thinks addiction is impossible (or rather, that it's a true expression of revealed preference), which he ideologically needs to do because to admit otherwise creates problems for his defense of unrestrained capitalism.

This is emblematic of a certain kind of foolish libertarian thinking where capitalism is treated as an end rather than a means. To normal people, the entire selling point of capitalism is that it is pitched as the best way to achieve a happy and comfortable life for the greatest number. If it's working against that end, what good is it? Toss it out and replace it.

Structured data as the communication medium between shell applications is so obviously a good thing, once you've used it, the Unix way of chopping up text with flags and formatting utilities seems barbaric. PowerShell got it half-right but blundered by talking in binary blobs of .NET objects instead of structured text; it was an understandable decision given how Microsoft worked at the time, but held it back from greatness.

But it's not clear how to get from here to there; Nushell doesn't seem to be catching on. I sometimes fantasize about adding two more standard streams, stdstructin and stdstructout, to go with stdin/stdout/stderr, which by informal or formal standard would write JSON (or whatever). Then coreutils and other programs could adopt them gradually and fall back to stdin/stdout whenever encountering a program not using them, and we could have gradual migration while keeping backward compatibility.

You are waaaaaay too optimistic about RAM prices. This is not a one-quarter crisis and you need to settle in for the long haul: the era of "hardware always comes down in price" is over, definitely for the next 3 years and possibly the next 5-8. The new DRAM fabs aren't even coming online until 2028 and there's not much reason to think they'll make much of a dent in the supply-demand mismatch even when they do.

There's a guy in Germany making homages to the cosmonaut watches under the Strela name (maybe he bought the rights? I don't know how IP works re: former Soviet brands) which have Chinese Seagull movements in them.

They look really cool, although for the price he's asking (around 350€), I'd almost rather they use quartz movements despite the hit to historical accuracy, I don't think a 350€ mechanical chronograph can really be trusted.

LPEs do need to be fixed, but for most people it's not a threat model they need to worry about.

No, the people responding to you are correct: even the crummiest mechanical watches should not drift by more than 10-15 seconds/day. COSC certification, which most nice mechanical watches have, requires no more than 2 minutes/month (4 seconds/day) of drift, and most do even better than that, 1-2 seconds/day is normal.

If your Seiko is really drifting by minutes per day, something's badly wrong with it and you should get it serviced.

I don't know if watchmaking is one of them, but there are a bunch of traditional crafts which are actually approaching a danger point because there aren't enough up-and-coming acolytes in the discipline to replace them, even though the craft still enjoys enough popular support to have a thriving economy.

Anecdotally, I see enough mechanical watches on wrists and in duty-free shops that I imagine there's enough of a pipeline there for at least one school. Much like vinyl records it doesn't appear to actually be going away even if it's superfluous.

This kind of shoddy, surface-level "incentives alignment" analysis can just as easily be used to prove the opposite: with the "pay once, then pay for upgrades" model, sellers are incentivized to never actually fix all the bugs so that you are always forced to buy the next version.

Construction companies can still turn a profit building and selling them. The idea is that houses should be depreciating assets, like cars. If houses are appreciating in value it means you're not building enough of them.

“Building wealth through real estate” should not be possible, and it’s half the reason we’re in this mess: people treating their homes as a store of wealth and blocking new construction to protect the value of their investment has locked young people out of the housing market. Hence all the gambling. What you should be hoping is that YIMBY policies carry the day in the end and “building wealth through real estate” is eliminated as a thing.

Remember to flag it, I've been doing this consistently with AI slopicles. Eventually we can build a social consensus around this and stop having to see them.

I'd say that about several other CEOs too if their fanboys wouldn't jump down my throat about it. Zuckerberg is just universally unpopular enough that you can get away with it

In its defense, it actually asked me whether I wanted to spend my 500 bytes on more resolution for the main continents or more landmasses, and I opted for the former. You can blame me for that, not Claude :)

Fable did that, I pretty much just told it to find the best Fourier sum which can draw a map of the world in less than 500 bytes of R. It downloaded geo data and made the curves to fit that on its own.

Well, here's what Fable came up with in 499 bytes of R in about half an hour: https://pastebin.com/sBsiGD9t, result: https://imgur.com/a/W3eDdIC. Probably with sitting down and tweaking you could do even better, but I think this is a decent first start.

Presumably you could precompute some parametric function (probably a Fourier sum) which draws a reasonably close map of the world, and get that into 500 bytes with a math-focused programming environment (R, Julia, etc.)? I might try throwing Fable at this and seeing what I can get.

In the 2010s Seattle briefly legalized SRO-like "pod apartments" of 300-400 sq ft, and several were built. A friend of mine rented one for several years and it was fantastic. But nothing makes NIMBYs throw a shitfit like the word "SRO" and they were eventually made illegal again.

It might also be that you just don’t want or need an ortholinear layout, plenty of people don’t. If you like the split design and programmability of the Moonlander but not the layout and associated relearning curve, I can recommend the UHK as a good choice.

Heh, it's not crazy if you're here in the Bay: I know multiple people who more-or-less disappeared for days when Fable came out because they were running their benchmarks, and only emerged blinking into the sunlight when the USG banned it. That's just how things are here now, most people are normal but there are some serious LLM dope addicts out and about.