Totally understood. My point being: in this case, to enjoy [benefit of legal differences], fork should be different enough from parent project, that such legal differences apply. And thus implies work on that fork to make it so.

No pain (from forking) = no gain (from having forked project be different in some useful way).

A naive "calories in = calories out" is not entirely wrong, except it ignores some things:

• Calories put in mouth != calories absorbed by body (with in this context, body = excluding your guts). Ratio of put-in-mouth vs. absorbed by body is a complex matter.
• The bacteria in your gut provide benefits in other ways. They may modify the chemical makeup of calories absorbed by your body, produce some nutrients / vitamins that your body can't make itself, stimulate production of hormones to provide a "full" feeling, 'tickle' your immune system, and other functions we may or may not know about.
• They may even affect how your body processes nutrients already absorbed.

The dumbed-down view ignores these other processes, their complexity & effects. While in reality, those effects are significant.

In short: your guts' health is tightly coupled to your health. Staying healthy includes keeping your gut bacteria happy. Ignore at your peril.

...when hardly any actual forking is done? Where forking = development of said fork.

Ignoring the (non) desirability of having forks in the first place: any value (?) of a fork derives from being meaningful different from its parent project, right? No difference = no point in having a fork. Very different -> fork may hold value in some way.

Read: to provide value, a fork should pull away from its parent project. If eg. the only goal is to remove binary blobs, then "take parent project's source, remove blobs, make result build-able, remove code that's left with no functionality" would be the way to go. Even if remainder would essentially be cherry-picked bits & pieces from parent project, with nothing added.

If that would attract developers, have at it. Over time, patches could (and would) be exchanged between the projects, like features / patches that cross-pollinate between *BSD projects, for example.

If that would not attract developers, then again: what the point of having a fork at all?

FWIW: Libreboot developers should just pull their hands of any fork. Do your own thing, let GNU do their thing. If the fork proofs viable, exchange patches as desired.

Taken to an extreme:

All this happens on-the-fly. Like how MP'3 are decoded as you play them.

Start with a small archive of storyline-generating templates. Add parameters to describe general idea, and presto! There's your storyline (generated as you watch, possibly even incorporating various types of feedback on-the-go).

Same thing for virtual actors: small set of templates, add parameters -> virtual actor / actress, skinned, clothed & supplied with personality as appropiate for the job.

Same thing for architectural elements, landscapes, vehicles etc. As you know: procedural world generation has already been done (long ago, depending on definition).

Friend comes over, and all of this is on-the-fly tweaked to work around sensitivities this friend may have (slim/fat, gay/straight, race, political views, etc etc).

Caching of result may be done in case you'd care for a re-watch, but in practice this is mostly pointless.

Now, assuming sufficiently advanced AI, and sufficient compute, who of you think this would (at least in theory) be impossible? If so: why?

Come to think of it: add enough real-time interactivity, and distinction between "movie" and "game" would quickly blur.

I use dead tree maps, you insensitive clod!

Don't know how, but somehow they still work. Kept working right through several Android releases, too.

^^^ This is logical:

If /. would not post at least some doom-predicting articles now & then, its readership would fizzle, and we would not be here to discuss anything.

If only 1 of these articles were both accurate and urgent, we would have found our doom, and thus not be here to discuss it.
If we were busy to avoid impending doom, we would not have time to be on /. discussing it. If the impending doom were in a somewhat distant future, we might just spend our last days discussing it on /. Which is what we're doing.

Conclusion: all of this logical, and us being here means things haven't (yet) gotten so bad that we're unable to discuss it on /. Doom hasn't (yet) been fatal, but may be impending or already happening, though.

Hopefully some of these funds will be directed towards RISC-V based, EU-grown, mass produced silicon.

RISC-V is the hot new CPU kid in town, and it's nice to see its ecosystem picking up steam lately. But if you go looking for actual products, it is mostly IP cores designed in US, actual silicon made in China. Other places doing university grade FPGA based projects, prototype silicon or similar, not mass production.

In terms of being less dependent on other countries for computing needs, RISC-V is a near perfect 'vehicle' for the job. Even India is running with it. Europe? I'm not seeing it. Consumer goods like toys, tablets, automotive, RP2040 style chips for hobbyists or educational purposes, etc, etc - the list of potential applications is endless. Expertise no doubt is present here. But actual silicon on the market, available for purchase right now? Practically non-existant (or so it seems). EU has some catching up to do here.

That's not good. That will only serve to make the software stack running on it more complex, less efficient & less responsive, for no net end-user gain.

Oh yeah, it may feel like an upgrade now. Just give it a few years, until this newfangled tech is considered a bare minimum.

You're funny, this sounds like a What If? from xkcd... Let's walk through this, shall we?

a) Just building a million satellites would probably involve a massive amount of CO2 output, negating any possible gains.

b) Never mind the CO2 output caused by building the required # of rockets (or obtaining the raw materials for such a massive project), and launching those rockets.

c) You can't 'set' a satellite at fixed point in space. It orbits the planet, choose orbit. Always sunny = orbit in 'equator' plane perpendicular to the sun = no shade on the earth. Other orbits = orbit partly on sunny side + partly in earth's shade. And always moving. Yeah I know about Lagrange points - doesn't help here.

d) Do the math: 1000 x 1000m = 1 km^2. So a million satellites 1...2 m^2 each works out to, ehm, 1...2 km^2. Read: 'nothing'. For example the Sahara desert alone is many thousands of km^2.

Options to improve: some mass driver, nuclear powered rockets, space elevator, building these satellites in space, using materials gathered from space as well, etc etc. Read: unproven / non-existing technology. And no we don't have time - it's running out, we need solutions that help NOW.

And we didn't get into things like space junk. Starlink has ~4000 satellites launched, aiming for 10k+, and astronomers are already complaining (rightfully, imho). You proposed a million satellites? Oookkkaaayyy...

TLDR; takeaway: do stuff on the ground, that's easier, cheaper & more effective. Like (if practical), paint the roof of your house white. Yes, that does help. :-) Plant some trees, install solar panels, commute on bicycle vs. using big honking SUV, reduce your meat consumption, etc, etc.

The other part: ruthless engineering.

I recall some video/slides about batteries. The problem: internal resistance. Tesla's engineers looked at this, and found that one factor was the length of the path charge carriers took from connector to active site (well that, and the usual electrochemical nastiness of course).

Most cylindrical batteries are made of strips of material, which are cut, stacked, rolled up, and shoved into a can. Electrons have to run the length of those strips in order to reach their 'work site'. Tesla's fix: add some 'flaps' that bridge the layers (electrodes), so that electrons only need to cross such a bridge, and run down the height of the cylinder. Much shorter path. Effectively turning long strip into side-connected stack, ceramic multilayer capacitor style. I don't remember the details, but that was the basic idea.

Result: batteries that heat less on charge / discharge, allowing them to be made bigger (surface vs. volume!), reduced cost of assembling a pack, and simplified cooling.

Or a case where some car manufacturers bought a Tesla, had their engineers take it apart, only to be amazed by how much fewer parts there were compared to combustion-based car, and how well the engineering was done. Leading them to conclude they were years behind Tesla no matter what resources they'd throw at it.

Rockets: "if only we could re-use a rocket stage once (with relatively little effort), we'd essentially halve the cost of that part". Okay, how about reuse 2, 5 or 10 times? And see where SpaceX is today.

Add to that a 'benevolent dictator' who goes all-in despite all the naysayers, and you're off to the races.

I agree with you, and yet I can't blame the public. As another poster noted, the public's trust in their government(s) have been much eroded as of late. And in the case of nuclear waste, all of the following applies:

• You can't see, hear, or feel it.
• It's nigh impossible to estimate effects of any intake.
• You can't personally check what happens at point of release (pipe into the ocean).
• And yes you can measure things, but dude in the street doesn't have gear like Geiger counters or mass spectrometers lying around.

The Koreans hoarding sea salt isn't even irrational imho: it's only used in small amounts (so a long-term supply doesn't take much space), it's cheap, doesn't spoil, while even being an essential nutrient to some degree. It's like hoarding toilet paper, but easier. By the time your stash runs out, the wastewater-release-and-dilution will be done. What more you want?

See salt trade in these countries will probably crash after this spike though. ;-)

[Privacy respecting] and [user repairable] are independent attributes, imho.

So there will be a market for phones that are easily repaired / upgraded by users, but still come with the usual Google stuff. Just like there's a market for regular phones running alternatives like LineageOS. And as you said, there's different ways to include Google stuff.

What is it with this pervasive "old software = bad" mindset?

Imagine you have this antique DOS based app, that happens to do what it's used for, 100% fine. And has been doing so for decades. Running on a mini, silent industrial PC, Raspberry Pi style box or similar, super stable & at lightning speed. Why on earth would you 'upgrade' that?

Sure, if something new does it better by all relevant metrics. Transistors have obsoleted relay based computers. Sure, if requirements change, old system can't handle that but new system can. Sure, if system is internet connected & doesn't get security patches any more. Sure, if you're a developer and the latest tools & gadgets are your bread and butter.

But lacking that: upgrading = pain, cost, risk, lost productivity. If it ain't broke, don't fix it.

Much discussion about the implications for designing new CPU's etc, but how about figuring out how existing black boxes work? This research shows great potential for that.

I'm thinking something like: hook up some black box IC to probes which register all its inputs & outputs, 'exercise' the equipment it's in by running a variety of tasks, and after enough rinse & repeat of that, have the AI spit out a rough model of the probed IC's function. Or start with die photo's of a decapped IC, and have an AI go over the tedious work of figuring out what each element is, how gates combine into registers, ALU, decode, control circuits etc, and map that into a high level model of the IC's function.

If/when possible, I'd expect such a model to be very rough & perhaps having some strange design features. But still, a mostly working model could be a lot easier to debug (by humans) & iron out corner cases, than starting from scratch with only a vague concept of what you're looking at.

Applications: obscure CPU's, DSP's, audio / video / chipset IC's whose designs have long been lost (or locked up in corporate vaults), parts in more-or-less vintage game consoles, engine control units, you name it. Yes, knowing the exact inner workings of vintage equipment is usually less & less relevant as time passes. But that's not always the case. And some things are relevant from historic perspective.

Obviously this applies to reverse engineering software as well. But let's be honest - disassembling a binary & figuring out what it does, is a lot easier than figuring out the inner working of a (complex) IC. The hardware probing equipment is the easy part there (while not being easy - depending on how well equipped your 'lab' is).

Sorry... according to various news reports, that's 1300, not 1500m. Mr. Rush's bad. ;-)