Read again, I'm talking about price per MW, not scale. Yes, tokamaks can be big, roughly by an order of magnitude power density for same volume. And also over more than an order of magnitude costlier.

FWIW, the limit for fission plants isn't "inefficiency" of fission, but scaling beyond 10GW, especially coolant wise (at some point you have more of evaporation towers than of a power plant).

CFS/ARC is a bit of unholy union here. The ARC people are extremely cautious but willing because theres simply no public money in the US for this, so they'll do anything. Meanwhile, CFS people are super overselling it. The idea is basically, get few hundred of millions of VC rube money to fund SPARC, turn it on and ask "will it quench?". The consensus is along the lines "most likely, but we'll probably figure out whether cuprate superconductors are at least worthwhile for this sort of thing".

ReBCOs are known for 30 years. Multiple accelerators toyed with it, and ultimately went back to good ole NbTi. Turns out preventing quench at 1.9K isn't *that* hard, your magnet being made of textured material causing a quench at "high temperature" of 20K is.

Ditto for field strength, conventional superconductors can go well over 20T, but it's rarely used in practice (construction limits with dense fields).

Or economics. It may turn out fission is just far cheaper including the cost of waste disposal. Currently, ITER promises "only" 50% more expensive reactors per MW than super-safe modern LWRs. In theory, because nobody has managed to run tokamak for longer than 6 minutes, so people are especially vary of such cost promises that are made on napkin, with no relation to actual costs of commercial reactors because nobody is building one.

Still nifty tech with bunch of application elsewhere (accelerators..).

And those that do just hang out in VRChat (modern day SL, more or less, sporting 20-30k concurrent users). The whole problem with Zuck's vision of "metaverse" is its sterility, lack of content, and being awfuly late to the party in a saturated niche market.

> The world has always been owned by rich who had an early start at something, even well before the industrial resolution to say nothing of robots.
Technology now creates eternally entrenched incunbents which is for the most part unprecedented. Oh right, Google is irrelevant because Bing exists, lol.

> The ideal that all information should be free and equal and that the human race only works for a common collective good is a fantasy best left to sci-fi novels.
The know-how eventually does spread, on account of such scarcity being artificial. It's all about how long that takes, and whether monopolies can be even competed with at that point.

> Yeah it would be nice if we started with an open source dataset, but that isn't a minimum requirement, it's just an optional extra.
It's a massive political issue in ML research already. I'd not be so dismissive about it.

I'm all for robots. What the usual neoliberal spiel of "bigger pie, you'll get bigger slice" loves to omit is who gets to own the robots. Datasets like this are not opensource by any stretch. The earlier you start mining it privately, the more of a headstart you have.

"It's totally for your safety, you guys, totally not a dataset to train a model for robotic hands to replace you with, plz trust us".

This is just one of the first instances of many.

When it comes to social network valuations, you're always buying the audience. It's $100/head for twitter - definitely overpriced. But not entirely out there. For instance, facebook is $80/head.

> buy your way into respectability
Blue checkmarks had been anything but respectable for a long while now. Why should "I'm openly authenticating myself to the public" be hidden behind some byzantine process exactly? Seems like internal bureaucracy overgrowth that serves no purpose.

At best, you're paying for a CS:GO skin here. The bottom line is that subscription does in fact authenticate you saying that you are who you claim you are (provided it's not anonymous crypto payments) - at least to twitter (who may choose to refuse service to typosquatters subsequently). Isn't that the whole purported point here?

> leave in droves outside of maybe an account managed by their publicist?

Leave to where? Parler? I agree the blue checkmark had been useful as a bait. Now's the time for switch.

> doubling down on on things that won't work.
Nonsense like hyperloop are more about personal brand building. The whole "say something ridiculous to rouse crowds of terminally online people who feel the need to correct the record" is twitter marketing 101 - just poke useful idiots to give you the spotlight, free of charge. Unfortunately leaking even into Slashdot and HN discussions.

It's even in the name itself. The brand of tesla 100 years ago too was a lot of stuff *way* ahead of competition. And also a lot of it just point blank nonsense because you can't just bend limits of physics, no matter how much you want to.

>Because the kids only speak Russian?
That's the least of it. For US kids, war and sanctions are in the way, obviously.

It really does. If I were to criticize something market wise, it's this new brand of insider trading one can do with a single tweet. For instance, saying that your own stock is massively overvalued (which it is, by all reasonable market measures).

This is why telling kids "you can't be a cosmonaut" is a bad idea.