jkf
1d ago
Even if the roads are winding?
I mean you have to slow down if there's a sharp corner -- then you speed up again!
jkf
1d ago
One feature of rural living (that can be shared with decaying urban areas) is roads in generally bad condition forcing you to drive more carefully and slowly.
People in rural areas have a lot more ground to cover, and fewer police hanging about -- we usually drive ~80mph everywhere paved regardless of road conditions.
Typical speeds on decent loose gravel would be 45-50 -- slower than that and you no longer float over the washboard.
jkf
1d ago
German beers and various dark types were a mainstay of first-wave craft brewers in the late 80s -- they were occasionally too malty (especially the dark ones) but some persist to this day!
IDK where you are going that meads are in style, and in any case the ones I've had are very much not beer -- that leaves Trappist and Sour which (assuming you mean sparkling-monk-style due to EU regulations) are pretty far from mainstream IME?
Whereas whenever I go to a restaurant since the dawn of the millenial, I typically have twenty disgusting IPAs with colourful labels to choose from, and can maybe find one each of decent lager/ale/stout. Sometimes the brewers are even sneaking extra hops into those though, so ordering something you've never heard of is not safe.
I assume somebody is buying this stuff, and my money is on the dudes (still) wearing the ironic facial hair.
jkf
1d ago
Lots of things can be learned by working towards resolving Navier-Stokes
My guess would be that since not much can be learned from a million or so lines of Lean, the results are not exciting in that way -- IANAM, but somewhat adjacent I guess, and my impression is that the result is not really the point in most cases. The proof itself is very important, and if the LLM is acting as some sort of Lean fuzzer then it's not that interesting. OFC we don't know what's going on without seeing the proofs, but I can't imagine they'd be holding off on some novel elegant proof of the Riemann Hypothesis or whatever.
jkf
1d ago
Depends on the photographer I suppose?
Sally Mann would probably want a lot more than the lady down the street who goes to Paint and Wine nights...
jkf
1d ago
I haven't heard a big company come out against it
There's a tonne of them that explicitly ban anything remotely SOTA for security reasons -- they do let you dick around with some terrible private-cloud instance of Copilot or whatnot so the C-suite doesn't worry about being "left behind", but it's a pretty strong repudiation of the usefulness of the thing. (from cybersecurity's point of view)
jkf
2d ago
If you hook them up to a machine that does random stuff (and you have an organization that apparently can't manage a simple internet firewall), I'm not so sure...
jkf
2d ago
He outed himself here months ago -- which is for sure against the rules, but Scott comments that he noted it and just did a shrug emoji I guess?
jkf
3d ago
Worth adding that the fact they still haven't rigged the things to say "I don't know" reliably exacerbates the problem -- an entity that knows a lot but not literally everything is still very useful, but if it's not possible to distinguish between what it does and doesn't know it's really not great.
jkf
4d ago
Where there is a "real Canada" among Anglophones that is distinctly Canadian, it's in the run-down Maritimes, the rural prairie, places which are sheltered from America's impact
I don't think it's so much America's impact as that TPTB have been actively trying to kill it for about 30 years -- Vancouver had a Canadian identity as lately as the 90s, and while it was still pretty self-centred I'd even give it to Toronto in the 80s.
It's something I've noticed a lot in the Trump discourse here lately -- if you substitute the "America" in all the unhinged things people are saying about Trump with "Ottawa", you will have accurately described the situation in vast swathes of the nation since about Y2K. Like, uh "We will not be reduced to a mere vassal paying tribute!" kind of rings differently in Alberta, y'know?
jkf
4d ago
balancing that would be a bear.
Speaking of bears, I'm kinda thinking that the time is ripe for an Oregon Trail reboot -- one where you might get lost and end up in Texas goes well with this!
jkf
4d ago
Actually joining the EU in any meaningful way would absolutely lead directly to a breakup (or at least some sort of splintering) of the nation -- the US might like this of course!
While all the boomers will of course say in context of the current environment that they love Europe, (the comparison to the US is implicit) so far as I can tell the ruling class of the EU is basically Trudeau on steroids -- and that approach is enough to splinter off at least the West in the medium term.
Probably the non-GTA parts of Ontario too -- Quebec would probably be on board, and the Maritimes will vote for whomever sends them the best gibs (although both are capable of becoming riled over seemingly inconsequential things, so who knows) -- I'm not sure what the country would look like ten years hence, but it would not be the same.
Ironically Carney is probably looking for some sort of non-deal that will allow him to claim victory, similar to what Trump seems to be doing in the case of Greenland -- so I agree that anything substantive is unlikely to happen; he runs the risk of summoning that which he can't put down though.
jkf
4d ago
America has never cared a pennysworth for the "sovereign rights" of other states -- America is about individual rights.
And in the event, this would be framed as using their imperial might in defense of the individual rights of the good people of Greenland/Canada against the Yellow Peril -- even if in reality broader State Department priorities like "we can't have China on our doorstep" were more at play. (see: Ukraine)
jkf
4d ago
Only if they really have a choice -- similarly with Canada, if China started actually doing anything serious in Greenland, a bunch of Marines would be showing up tout de suite.
jkf
4d ago
One wonders what the implications for the other Arctic-holding State to the US' immediate north is, as this agreement inherently (and formally) threatens all Canadian trade capacity with non-US countries.
I'm not seeing it -- can you expand?
It does put us at risk of having to fight a liquor war on two fronts though -- not quite a land-war-in-Asia level blunder, but I'm not sure I like the vibe...
jkf
5d ago
complete lack of info coming down from the top other than "nothing is happening, everything is fine, proceed BAU".
ruh-roh.jpg
jkf
5d ago
But do they care quite so much?
Because if they don't have Taiwan, they are bottled up in any big war involving the US, and this is a serious strategic problem! (and strategic problems in potential wars are taken seriously there, I guess?)
I've heard the same about Tibet -- it's military importance (in a hypothetical big war involving India) is much more, um, important than any deep cultural need to oppress Tibetans -- makes sense to me.
jkf
6d ago
C# is also kind of bad, yes -- to the extent that I had a point beyond lazy shitposting, it was that the LLMs seem to have picked up on a particular... style of OOP which is not only overly verbose, but also misses the basic idea of the thing. All of the factories are more of a symptom, people do the same thing even when trying to be object oriented in Python. Rigid adherence to something that they read somewhere while not really understanding the reasons, I guess? The "Clean Code" series is likely the source of a lot of it. And yes I have seen it in the wild (from humans) all the fuckin time -- you can actually coax an LLM not to be too brutal with it more easily than most human perps, but so far as I can tell it's absolutely their default mode. (due to training data prevalence I should think)
Which sure, fine -- but shouldn't the default architecture for any tools that implement LLM call and response be pretty fuckin agnostic towards what it's plugged into? Like, we are sending some questions and getting back answers; "the answers keep getting better!" seems like an awfully lazy excuse for not making the tools also get better!
jkf
6d ago
Chinese interest in Taiwan goes back to the Chinese Civil War and is completely unrelated to hopes of taking the semiconductor industry intact. In fact at this rate they might do it themselves to slow down American AGI scaling. They'd prefer no war at all, though.
I thought the main motivation these days (and for the past quite a while) was preventing their navy from being bottled up? (and shipping blockaded I suppose)
Not that the "Taiwan is a fake country" stuff is itself fake exactly, but I'm not sure they'd care quite so much if it were a fake country in a non-strategically important location?
jkf
6d ago
It's too busy making ObjectFactoryFactoryClasses to deal with such things, I suppose...
jkf
7d ago
Part of the reason why this hasn't been built is that the underlying models are improving too fast and too unpredictably to make building it a good business.
Do we not have, uh like, modular software design? Encapsulation? Maybe instead of hiring weirdo rationalists to philosophize about sentience and safety these companies can pick up a couple of 90s-era C++ developers?
jkf
8d ago
I'd believe anything from "months to repair" to "we've already restored most of the capacity".
If you can build a pipeline, you can also fix one -- "however long it would have taken to build however many feet of pipeline got blown up + time overhead for general setup" would be a good first guess. Of course if a pumping station gets hit or something it will take a little longer to get everything in place, but I'd think "days not weeks" when shipping fees on 6-7 MBpD are on the line...
Don't get me started on the Boomers...
More options
Context Copy link