Nobody in the world is going to build nuclear power for economic reasons at the moment, for the same reasons nobody is going to build coal power or even combined-cycle gas power plants.

New wind and solar, backed by gas peakers and hydro, are WAY WAY WAY CHEAPER THAN THE ALTERNATIVE NEW-BUILD OPTIONS. Coal plants are shutting down because they can't compete with new-build renewables. Don't believe me - read this, in that radical lefty rag Forbes.

It's not completely beyond the realms of possibility that these costs may change at some point, but the nuclear industry has had several decades to get its collective act together and has failed.

And the rest of the world is a pretty big place. There's going to be a market in other wealthy countries with rural and remote populations - while Europe is more densely populated than the United States, there are some pretty remote places in parts of, say, Scandinavia. Australia screwed up its national broadband project so there are a lot of Australians even in suburban areas who can't get proper broadband (though Starlink would be competing with 5G in that case). And that's even before you get to rich people in middle-income and poor countries.

Have a look at this data on death by violence in preindustrial societies. It suggests that in many such societies, upwards of 10% (and in many cases close to 30%) of people died violent deaths.

Sounds pretty much like "chaos" to me.

Oh, bollocks. If you're really in the 0-day intelligence business, you'd keep buying exploits (though possibly at a reduced rate) even if they're abundant.

Occam's Razor suggests that Zerodium are a front for the companies that build hacking tools for American and allied intelligence agencies.

The genome of the virus has been analyzed and there is no evidence from that analysis to support the proposition that it was an engineered virus.

Yes, that's one study, but I have not seen any reports of a counterargument from the scientific community.

One piece of good news about coronavirus is that children, including infants, seem to be much less severely affected than adults and especially the elderly.

Credential stuffing attacks typically work like this (simplified, and assuming the absence of countermeasures to make this process harder):

• 1. Hack some poorly secured credentials database with hashed passwords
• 2. Use a chunky GPU to compute the hashes of a few squillion combinations of passwords
• 3. Look for matches. Now you've got $LARGE_AMOUNT of username and password combinations.
• 4. Log in using the same username/password combination on other sites, as many people reuse passwords.
• 5. Profit!

This is a preprint, not a peer-reviewed study, and there's considerable doubt whether their data actually supports the conclusion.

Andrew Gelman, a stats professor at Columbia, has looked at the statistics behind the study and concluded

...Again, the point is not that the paper’s substantive conclusion—of the positive effects of air filters on cognitive performance—is wrong, but rather that the analysis presented doesn’t provide any real evidence for that claim. What we see is basically no difference, that becomes a large and possibly statistically significant difference after lots of different somewhat arbitrary knobs are twisted in the analysis.

As somebody who taught at a respectable but not quite as crazily resourced university as Harvard, it's hard to care what they have to say about anything relating to teaching CS.

Their approach to teaching is a magic show in lectures, then throw a book and an SDK at students and pick marks out of their butt based on how cool the projects they build are. That works if your students are workaholic autodidacts with massive self-confidence and you can grade-inflate (and your students all know that a run-of-the-mill Harvard degree transcript is a near-automatic meal ticket anyway). It also helps to be able to provide them samples of every cool gadget under the sun courtesy of your industry sponsors.

I'm an Aussie who drinks beer. Haven't seen a can of Fosters Lager since the 1980s.

More seriously, an Australian standard drink contains about 10 grams of alcohol. In most Australian states, the standard serve of beer in a pub is 285ml (roughly 9.6 oz, so smaller than an American beer glass). 285ml of a typical lager (4.8% ABV) will have about 10.7 grams of alcohol. So, to a rough but good enough approximation, 1 standard drink = 1 glass of beer ("pot" if you're in Victoria, "middy" in NSW).

If you're an American, 10 standard drinks a week would be the equivalent of roughly 5.5 16-ounce pint glasses.

Call the Greta Thunberg helpline. They have professionals ready to help people just like you!

I fully understand that in professional environments, the cost of the hardware is a small factor compared to the cost of employees and software. But, still, if I was architecting an application that could use this much compute, I'd be looking to shift as much of the compute as possible to shared servers or the cloud so that expensive compute wasn't sitting idle much of the time.

What's the use case that requires this many cores and this much GPU locally? Is it an artifact of old-school software licensing?

It took the Japanese and the Koreans a couple of decades to design and build cars to their own design that were competitive with existing automakers. No Chinese automaker is quite there yet. Tesla took over a decade to deliver a mass-market vehicle and has taken enormous cash losses in the process; maybe those massive bets will pay off but if the company wasn't run by Elon Musk it would have run out of cash a long time ago.

As such, it's hard to imagine that a new company, with a new supply chain, will be able to deliver saleable cars in the American market by 2021, and cut costs and improve the product fast enough to be profitable before they run out of money.

The traditional way to pull this off (excepting Tesla) is to get your home government to keep out foreign competition until your cars are competitive with global firms, then start exporting.

Newsflash: two things can be bad at the same time.

I'm old enough to remember how the release of hit games drove a lot of people to buy computers capable of playing them in their full glory.

If Alyx is a sufficiently compelling game, and given that Half-Life is a franchise with a lot of resonance with older gamers with the resources to buy a grunty gaming PC if they want, couldn't it drive the sales of a lot of VR gear?