Not sure to whom you are responding, but if it's to me, I thought my explanation was clear enough; that you generally didn't have to provide ID, unless you'd registered without proper ID or the last 4 digits of your SSN, in which case you would have to provide ID only the very first time you voted. Your simplification of this to "doesn't require voter ID except that it does" is a bizarre and incorrect non-sequitur.

The general election audit process in most democracies is to ensure that voter ID is presented in all cases at the polls for each voter, and I had honestly thought all here would be up to speed with that as a concept, and I outlined exactly that in my post with my own experience. Clearly you are not, and fair enough.

California not only does not require this, it actively prevents local governments from requiring it. Perhaps this is a very good thing; I do not know, as I cannot evaluate the terrible disenfranchisement effects of a malign racist legacy in California in particular and the US in general. I am told it is so bad that normal safeguards for other democracies should be overridden and this may well be true. In any event it is for Americans to decide.

So if you are genuinely too short for this ride and not simply being silly, here is a pathway where ID will never be required under current law (not just a pathway where ID to vote at a given poll will never be required)
1) Voter registers and provides last 4 digits of an SSN that match their name.
2) Someone later turns up, asserts they are the voter, and votes or a mail-in ballot is delivered to an address associated with the voter and is subsequently returned.

[2] Repeats for the next 100 years until the voter dies at age 118. Presentation of ID, in CA under present laws, is never required in order to vote.

Given the staggering rate of identity theft in the US for much lower stakes this seems, perhaps naively to me, to be a strange idea. But so be it.

This may well be excellent; in which case, let's look at enabling this for driving, buying firearms, travel. entering parks, and so on.

Not only does California generally not require ID to vote[1], they have passed a law, SB1174 2024, explicitly forbidding local governments from passing requirements to check IDs, and from enforcing such requirements if they already exist.

I grasp the need to avoid disenfranchising people (which the US has a sorry history of), but from where I live in Canada, it does seem slightly curious that governments can require ID to drive, to buy alcohol, tobacco or firearms, to travel (especially by plane), to enter government buildings, to visit national parks, but not to vote.

The last federal election, even though the elections officer knew me (and greeted me by name), and both scrutineers recognized me, I still had to show ID. It made good sense not to have exceptions to the rules.

If voter fraud is as vanishingly rare as most in the media assert it is, California is probably making the right choice in terms of avoiding disenfranchisement. That said, perhaps we could also see some loosening of ID requirements for other things.

[1] Apparently if you don't present valid ID (or the correct last 4 digits of your SSN) when you register to vote, you need to present it the first time (and only then) you vote if I read California laws correctly. IANAL.

If you're running them locally, and they do claim to run something locally, sure. But with an anemic 5 GPU cores and a completely undisclosed RAM capacity (I'm guessing 6-8GB?) you're not going to run much of a model on that. I can't see more than 8-12 GB of LLMs and visual generation AIs combined, and that's being wildly generous. Probably more like 4-6GB for local storage.

So the suggestion that 128GB is needed because 'Those AI models need all the space they can get' seems a bit silly or unthinking on The Verge's part. RAM, sure. RAM bandwidth, sure. If they had an iPad with ten times the RAM and ten times the memory bandwidth, and ten times the GPUs, then, sure, that'd need a lot of space for model storage. And I'd be buying it even if I had to keep it plugged into the wall.

But 128GB is just catching up to what Samsung was shipping with the small Galaxy Tab S7 over 4 years ago. (True, Apple's vertically integrated stack is great, and CPU/RAM requirements for the same performance can be lower for an Apple product, but that generally doesn't translate to storage or even to LLMs)

I guess it's nice that Slashdot is running PR pieces now?

Well, there is ROCm, which, if AMD can meet certain price performance points, could become interesting for prosumers and hobbyists. (Yeah, I know, I know).

This capability of easily resizing RAM dedicated to a "GPU" (Let's recognize that in this application it's more an NPU/AI) engine raises an interesting prospect. Doing some rough estimates below (and in come cases mentally converting from C$ back to USD$).

The cost-sensitive poorly addressed AI window that seems open at the moment is the ~48GB-192+GB space. You want to run a 70B[illion] parameter model, say some Meta-Llama variant with decent-sized token window, and quantization no lower than ~6-8 bits. Right now you're looking at maybe $8K? for a mac studio ultra (192GB max, and no CUDA, and wildly walled garden with not great performance but low power consumption). You're looking at 15k per 48GB Ada card from NVIDIA, so probably 60K++ plus your computer. You're looking at loading up with a pair of 256-thread EPYCs and high speed RAM and running it all in CPU (256-512 GB np), that could be close to $30-40k+. Or you pay maybe $5+k and accept a stupendous quantization hit and get a 4090 machine, 7 or 8k if you want two. With the exception of the mac, none of these are great for power consumption, especially the 2000W? needed for your Ada AI workstation.

Some of those will give you "good enough" performance for certain applications that don't require anything like a near-real-time response. Only the ones $30k and up will likely give you anything like decent performance, and even then, I'd guess that the multiple Ada card option would be by far the best.

There's a big window here in the $5-10K range for embedded NPU/AI cores that are very high-performing with "fast-enough" RAM. Dual socket at ~2x460GB/s (6 channels?) doesn't seem awful. Being able to adjust what RAM is dedicated to the GPU/NPU/AI cores on the fly could get rather useful.

Offering open-source AI compute engines that can scale from handheld levels to laptop to console to well beyond RTX 6000 levels in model complexity (though probably not immediately speed) could put AMD in an interesting space. Would it be a winner? Depends I guess on academic researchers and hobbyists, both groups typically on a budget and willing to be inventive.

Horses for courses. I have over 3000 physical books, but most of what I buy now is digital. I used to lug up to a dozen books when I travelled for more than a few days. Now I carry them with me. Similarly, search and rapid access (most of the time; barring events such as this) is better.

More interestingly (and critically) a bibliophile relative had a bad set of brain bleeds. He lost the use of pretty much half his body. Thankfully his cognitive capabilities have almost completely recovered, and he's gained very limited walking ability, but even turning pages and holding a book is close to beyond him physically, and forget about easily accessing bookcases filled with physical volumes. E-readers are a godsend for him.

I get all the negatives. Soulless. Narrow number of gatekeepers (mostly Amazon and Ratuken/Kobo). Easy to lock you out of your books, whether through malice or incompetence. You don't really own the books. You can't legally easily lend them. They can be stealth-edited. May well harm authors in the long run due to massively reduced competition. Lack of longevity. We can discover centuries old volumes, even millennia-old, from the past. Good luck with Kindle Unlimited in... a century? a few decades?

But simply snarking about the superiority of physical books (and IMHO they are mostly a superior and more beautiful technology) is to ignore a non-trivial range of use cases where they are, sadly, inferior.

How exactly are they crybabies? This is an absolutely expected outcome where property rules are really unclear. Even the "wild west" during the goldrush had a much more solid legal framework of mining and property claims than we currently have in outer space. And there, you not infrequently saw all sorts of secrecy with respect to where people were mining, and filing claims until the last possible legal moment.

We don't have to (and shouldn't) turn space into an anarchocapitalist free-for-all (though that would probably be preferable to a totalitarian regime that worked us all to death as indentured servants in asteroid mines) but we need to have a better legal regime that benefits those, including smaller nations, that discover something (and can exploit it in a reasonable timeframe).

"with my VIC-20". Yeah me too.

[For more, see Brian Bagnall's "Commodore: The Early Years" or "The Story of Commodore: A Company on the Edge"].

Interestingly, Commodore significantly outsold Apple for quite a while in that early 1980's era. But Apple had the Steve Jobs Reality Distortion Field/great marketing working for them, and Commodore had some of the best hardware engineering and worst management imaginable.

The really funny thing is that Commodore owned the 6502, which most of its competitors used, so it had a pretty good idea about both MOS' [the company not the process] and MOS-licensee volumes. And they still did a pretty bad job of trumpeting their own successes in North America especially.

That said, I had a VIC, and no question the Apple ][+ was what I wanted for a time (never said I too wasn't susceptible to Jobs!). But $300 vs $2000, I could see why my parents weren't going to go down that road.

I would speculate that there was a lot more software for the PET/VIC, but that the article could well be correct that there was more commercial software for the Apple 2. Wikipedia airily suggests (with no great sourcing) 300 cartridges for the VIC and 500 [commercial] cassettes. "Free" programs you could type in on the Vic were absolutely legion and available even in the grocery store impulse purchase magazine rack.

Interesting times.

Thank you, I was unaware of these efforts. +1 informative if I could.

The spirit of this idea is bang on. Let's stay away from the IRS "seizing" more than it needs to, and let's avoid having the IRS getting into the streaming business. But absolutely, the material should become public domain (or copyright ceded to the Crown/Library of Congress etc. as applicable by country).

Sadly, depending on the country, preserving these things may well be piracy, but you're right, it definitely should not be.

Very fair point made by both of you -- icthus and Rhipf -- and obviously nothing from HP that fit my specs was available for me to buy (supply chain issues, hence my nod to Brother's distribution channels) or I'd have done that comparison. (Ditty Lexmark).

Depressing though. Did we really hit peak mechanical build quality in the 1990s? Some of that can be explained by the Japanese bubble economy -- consider the million mile 1st gen Lexus cars, and the 25-year-old Honda Accords (in Ontario, rust capital of North America!). But surely not all.

(NB- yes, newer cars are better in many respects -- better active and passive safety features, self diagnosis, mileage, much better automatic/CVT transmissions. But longevity and general mechanical excellence?)

Sad not to be readily able to buy a printer that can reliably print non-crooked labels in 2022.

Brother printers have great features for the money and are relatively clean to install and use. Excellent point on Linux. They also seem to have good distribution in North America at least. But "great" build quality?

I was astonished at how flimsy my new (2022) Brother printer was compared to my ~20 year old HP (from back when HPs were good). Poorly fitting parts, a lot of slop in mechanism movement. Unsurprisingly, labels all print crooked.

If you need a quick cheap laser printer with impressive features for SOHO use, Brother is great. But I think you get what you pay for, and if you think they have "great" build quality, I shudder at what competitors you must have suffered through using in the past.

I agree with you on the necessity of Jobs (though without Woz there would have been no Apple I) for Apple to exist. But on Commodore I think you're quite mistaken. Remember that MOS ran into a buzzsaw of litigation from Motorola and wound up being bought out by Commodore in 1976. While Jobs offered the Apple I to Commodore (wouldn't that have been an odd reality), Chuck Peddle of MOS/Commodore persuaded Jack Tramiel to instead go for what became the Commodore Pet in 1976.

Certainly absent a Jobs/Apple, the Pet, the VIC, and the C64 would all have existed. Commodore would have owned the North American educational market far longer without the existence of the Apple II. Commodore Europe would have been even stronger (recall Commodore Germany outsold IBM in the PC compatible market). Competitive pressures from TI and Atari would likely have forced Commodore's hand on something very much like the Amiga, as happened in our time.

Likely one of Tandy, Commodore, TI, or Atari would have survived and prospered in the computing space (Better UX, Not a PC) that Apple wound up dominating. From a technical perspective, with the first multimedia machine, Commodore and Amiga would have been a pretty good bet. Failing Commodore, I'd probably have speculated Atari with GEM.

Brian Bagnall's "Commodore: A Company on the Edge" is a worthwhile read for those interested in pondering what might have been.

Can you please tell me what you mean by "subsidizing the fossil fuel industries"?

Serious question, as this is something I've oft heard as a sound bite, but never heard defined. (FWIW I think even in a stipulated absence of near term catastrophic anthropogenic global warming, continuing to burn fossil fuels seems extremely dangerous and wasteful as an open-ended dangerous experiment with our atmosphere).

If you mean reduced tax rates, that's a kind of a sketchy bit of reasoning on your part since it assumes all profit and wealth belong to the government.

If you mean actual cash flowing from the government in excess of tax revenues received (say both over a rolling 10 year period), then yeah that'd clearly be a subsidy.

What are the subsidies -- in say the US or Canada -- that you would like eliminated? (My best guess would be so-called "exploration" tax credits, but even then are they actually subsidies by any rational definition? If so, fine, let's eliminate them).

Thanks
-Holmwood

It's wonderful that you aren't bothered by WTN (wind turbine noise) (n=1). The reality appears to be more complex. In a sound Health Canada study (search Health Canada Wind Turbine Noise and Health Study) they concluded that there was no direct correlation between lower ( 45dBA of noise) levels of WTN and objectively measured sleep and health, but that WTN did directly proportionately increase self-reported annoyance throughout the community, and that had objectively measurable statistically significant effects on health (e.g. higher blood pressure, hair cortisol, etc.)

Note that this study was unable to examine effects for higher levels of noise due to low numbers. Critics suggest, not unreasonably, that this is a flaw.

Some people, particularly perhaps those on the autism spectrum, are considerably more troubled by noise than others. Given that measurements were consistent throughout communities in a reasonably large dataset, this suggests that the problem is likely real, even if largely psychological.

Interestingly, reduced effects (though not eliminated) were reported for those who received compensation (e.g. rent) for the wind farm's presence.

And this suggests a simple solution: for changes in the local environment (in this case, power generation such as nuclear, hydro (spillways), or wind), offer compensation that corresponds to the diminished value of surrounding properties. Doesn't matter if the windphobes or nuclearphobes are irrational; there is real diminished value that can be measured. So compensate owners (and renters!) and let them make their own choices. This would usually be trivial compared to litigation fees in US (or even Canadian) courts.

First, yes, does this tech scale? No strong indication of that in TFA, and the analysis of a small ICE engine's requirements upthread are a useful start. Possibly most useful for spacecraft.

Meh, I'll bite.

The idea that we aren't at a scale to affect nature is passing strange. Anti-colonialists can ask after the passenger pigeon which once darkened the skies of North America. Anti-communists can ask after the Aral Sea, one of the worst environmental disasters of modern history thanks to Soviet arrogance, mismanagement, and environmental predation. And anti-capitalists/anti-industrialists can ask after a great many things. All of these were accidental and oft incidental at the time.

Now if your argument is either that we're still a Kardashev Type 0 civilization, or that we can't significantly hold off the next (overdue) ice age with our current socio-technical capabilities, sure and very probably respectively. But that's not what you stated. It was that "human actions are simply not at the scale to affect nature".

Heck you're a giant pessimist! Leaving entirely aside the 2bn-odd Jews/Christians/Muslims who believe in stewardship of -- and therefore the ability to affect -- the earth (Bereishit - Genesis, and Quran 7:56), countless atheist (and not a few religious) SF fans have rationally believed in terraforming as a possibility for well over half a century. They may all be wrong, but atm, I'd bet on terraforming as being feasible. Or Elon Musk is going to be very sad.

-Holmwood

No!

The NVIDIA shield is not some subsidized Chromecast or SmartTV or FireTV. You pay $200-$300 for a good one of these devices.

These are people who have deliberately paid a premium for unsubsidized hardware that doesn't need to be connected to a BigTech account for non-BigTech direct functions. And until now it never featured this kind of space-occupying advertising.

At the very least many users not unreasonably believe there was an implicit contract that they would not be the product.

And the ads are quite negative in functionality, taking up ~40% of users' screen space as users try and select something.

This is a surrender that makes us all products.