Your comparison between phones and VR/AR is reasonable but a bit different as when windows phones were discontinued, Microsoft had pretty much lost the phone os race. Also the windows phones sucked, I’ve used them…

IMO microsoft gave vr/ar a fair chance. They might have been early, but if we are eg. a full decade off must buy VR, then it might not be worth waiting.

To OP: this is a much clearer & better explanation for what I was trying to say.

This is an imperfect analogy, but think of updating between Windows 10 and 11 versus installing updates on windows 10 or win 11.

I have no experience with Fedora, but AFAIK at least in Ubuntu/Debian land, updates are installed from OS version specific package repositories. When the version of the OS is no longer supported, those repositories might not receive updates anymore.

EDIT: this is the main reason I have a rolling release distro on most of my personal machines. The package repos have the newest packages without having to update my OS major version every now and then.

I found steamdb.info. According to them Godot seems to be growing steadily.

My bad. You don’t need a rocket to launch stuff.

I mean you could stay in space, if you were to reach escape velocity. Heat might be an issue though.

Tweaking the nuke to achieve a specific orbit might prove to be difficult. Also it’s not like Blue Origin has nukes right?

This was well put and a good summary of the situation!

In a less resilient democracy attempts of interference in the election process might not cause the same uproar it has in the US.

This also works the other way. The prosecution of Trump seems to be handled with care to ensure that the charges are justifiable. In non democratic countries a political opponent would first go to jail and then the prosecutors would try to invent some kind of corruption charge.

Congress members and senators still showed up to work, and the decisions they took still mattered, even if some of the Republicans were constantly violating precedents and norms. The judicial system still kept churning and mostly following the laws and precedents, even if Trump appointed a lot of unqualified partisan judges.

From an outside perspective this is a good demonstration that while your system is somewhat flawed, it’s still resilient. By flawed I mean mainly the two party system and stuff like judges being appointed by politicians. However if your system didn’t have some builtin failsafes, it would have been much more vulnerable to influence from unwanted sources.

Even if most trump voters wanted to turn the US into a proper aristocracy, (some right wingers actually do*), the process would have been much more complicated in comparison to countries that have become dictatorships in the past decades.

*I’m referring to a somewhat new trend, where influential people are claiming that the US is suffering from a dumb population, and that experts should be given more power.

That’s a job for the parents though isn’t it? And for early teenagers people seem to forget what positive influence the internet could have on their lives. Eg. many IT workers started fiddling around with stuff when they were quite young.

Obviously that has to be reflected in the price of the product. Presumably even more so with storage.

Also there might be a use case, where cost is paramount and the drive would experience very limited writes.

I’ve got a personal anecdote that’s not entirely the same, but I’ve bought a bunch of flash chips from china to use with retro games. Those are often salvaged, but they are also cheap and available to buy. It doesn’t matter if the chips can’t take too many write cycles, if you only flash them a couple of times.

When I did play foss games, I played battle for wesnoth, teeworlds, minetest, super tux csrt and openarena. Lsst one might be dead due to being a mainly multiplayer game.

Should the delays and subsequent costs overruns then be simply attributed to increased regulatory complexity or corporate greed?

I’m looking at the list of reactors in France, most of the builds during the last millennium were completed in more or less 10 years. Then there was a gap, and the new one is taking way longer than previous ones.

Same thing has happened in many other countries. Including finland, where at first we got 4 reactors in 6-10 years, and then after a gap of 25 years the next reactor was a clusterfuck that took almost 20years to build.

Both of these reactors are of the same design, and the issues are at least partially attributed to the company having forgot how to manage such large projects due to the years long gap in construction.

Competent nuclear engineers and technicians have retired without being able to pass on their know-how, and cutting-edge nuclear-related industries have disappeared or been converted.

This same fear has been enough to fund SLS and Ariane programs. Basically to avoid the loss of a capability in case it’s needed later on. For some reason it doesn’t seem to apply to nuclear. And now people are complaining that building new reactors is expensive, arguably at least partially due to the supply chains no longer existing in the same scale as before.

I like his approach of stating that these are big claims, but cannot be really evaluated in any meaningful way, as there is no useful public data to look at.

In my opinion this is the main crux of the whole topic. One party (the US government) has a boatload of various quality sensor data that could be studied, but it’s classified. As a result nobody can believe anyone’s statements as they aren’t supported by anything.

Most likely you won’t even notice some of the changes. Reasonably believable cars can already be added to films in post, so no reason why humans couldn’t be. This might not be driven only by AI, but instead on more general tech developments in vfx and such.

Thats the point of having energy abundance. When electricity costs are low enough, it wouldn’t matter, if the source of synthetic fuel was not the most energy efficient one.

But the naysayers will argue that your problem is not novel and a solution can be trivially deduced from the training data. Right?

I really dislike the simplified word predictor explanation that is given for how LLM’s work. It makes it seem like the thing is a lookup table, while ignoring the nuances of what makes it work so well.

The same guy x:d this. Apparently a chinese university has replicated at least the diamagnetism claimed in the paper.

https://twitter.com/Andercot/status/1686286684424691712

Shouldn’t a different algorithm that adds a some sort of separate logic check be able to help tremendously?

Maybe, but it might not be that simple. The issue is that one would have to design that logic in a manner that can be verified by a human. At that point the logic would be quite specific to a single task and not generally useful at all. At that point the benefit of the AI is almost nil.