Once again, the environmentalist fringe has set standards so high that they are impossible to meet so that they can berate folks for not meeting them.

Meanwhile, PLA 3D printer output can be trivially mechanically shredded and extruded into new filament several times. It's hard to say that PLA isn't recyclable with a straight face.

+1. In my home town, the pay phone by the high school was used for exactly two things: calling parents to pick kids up after away games and calling in fake bomb threats(*) to get out of tests. I would expect similar behavior from public phones today, sadly, minus the kids calling their parents part.

* When I was a freshman, this is what the seniors told me people had done in previous years. I cannot corroborate the story with any actual evidence. Also notable: this was in the early 90s, before school shootings and bombings were really a thing.

My father was a consultant; and he always told me that there were two very different types of client: Some clients had a decision they needed to make that raised questions they didn't have the expertise to answer, other clients had a decision they had made for which they wanted additional justification. The former wanted actual analysis both of whether the questions they had were the questions they should have and the answers to the questions they should have. The latter absolutely wanted the performance of analysis, clearly shoddy work or an obviously stacked deck(metaphorical or slide) defeated the point and made the cynicism of what they were doing too overt; but they were not hiring you first and foremost to get them an answer they didn't think they could get themselves.

I am significantly less clear on how much benefit the first class of clients is getting from 'AI', allegedly there are some narrow use cases where performance actually lands in the same ballpark as hype; but the second class of clients could absolutely do as well, or better, in terms of adding prestige and second-opinions-were-obtained cred to whatever decision they already wished to arrive at; given the absolute mania for anything you can call 'AI' in management circles at the moment.

If you are basically calling in McKinsey to add gravitas to your layoffs that seems like business they are either going to lose or have to do at pitifully low margins to keep up with the 'AI' guys; I just don't know what percentage of their business is mostly about adding prestige or letting an outsider be the one you can point to when the axe starts coming down vs. actual analysis where asking the right questions and getting the right answers is important; where AI hype could still make landing gigs harder; but the bot will have to deliver or the pendulum will swing back after a bunch of embarrassing failures.

But it costs 100x as much, if not more. Running an LLM can be done on a notebook these days. But training one requires an entire data center of expensive GPUs. Not to mention that the notebook will run a reduced (quantized) version. Go check huggingface how large the full models are.

And also, LLMs are still suffering from a number of issues. For example, on many non-trivial tasks, the LLM is still unable to follow simple instructions. If you use LLMs routinely, you likely found cases where it has zeroed in on one - wrong - answer and no amount of prompting can convince it to give you a different one. It'll even totally ignore very clear and explicit prompts to not give that same answer again.

A human will understand "if you give that answer again, you're fired". An LLM... well you can tell it that it'll get shot between the eyes if it repeats that once more and it'll tell you where to get help if you have suicidal thoughts.

These things are both amazing and amazingly dumb at the same time.

The part that I suspect they genuinely don't like is that the "MicrosoftXTA" CPU vendor code, which corresponds to a Windows ARM device(which I think at this point means 'Qualcom'; possibly a VM on a mac?) is meandering between .08% and .07% and back.

Despite those systems being genuinely well above average in terms of bringing remotely mac-like battery life to Windows; and(despite...optimistic...MSRPs) often appearing on sale at decently attractive price points; it appears that some mixture of apathy, incompatibility, and the total disaster that was the rollout of 'recall' and 'Copilot+ PC' seems to have just cratered those; at least among people who touch steam even casually.

Could be that windows-on-arm is flying off the shelf somewhere else; I don't have MS sales data; but when what was supposed to be the halo product of the win11/Glorious AI product era is under .1%, beating out those well-known Debian gamers by .01 to .02%, they can't be entirely thrilled.

What would be interesting to know(I did some poking; but didn't see CPU information breakdown by architecture or model number; just vendor, clock speed, and core count; and no computersystemproduct/other platform identifier; my apologies for asking a dumb question if I missed something) is what the percentage of linux on steam deck 'like' systems is.

The steam deck itself seems to have held up very well in terms of the semi-custom CPU's priorities, the target resolution, the peripherals included, and the overall polish and user experience; but it is definitely not getting any younger; and there are a bunch of options that ship either with the Z1/Z2 or generic newer AMD laptop APUs, plus MSI's 'Claw' with an Intel(that actually puts in really respectable numbers when the drivers aren't letting it down); but consensus on win11 as a touchscreen OS on devices either without a proper pointing device or with a teeny little one seems to be pretty solidly negative.

That makes me curious about whether gaming handhelds get converted to linux at a significantly different rate than other form factors. I'd assume that 'gaming' laptops are probably about the most hostile hardware flavor; since Nvidia has massive share in discrete laptop GPUs and the 'Optimus' arrangement that allows all the internal display and the video outs to be wired to the iGPU, with dGPU picking up work as needed, is massively driver dependent; desktops are probably the easiest(since you have more control over parts; and you can just shrug off "weird ACPI quirk causes BT chipset to not sleep properly" because you are on the wall and who cares; where that would potentially drain a sleeping laptop's battery pretty quickly; but desktops are also the place where win11 is as inoffensive as it is possible for it to be(still pretty obnoxious; but when you've got a large screen and a real pointing device and keyboard its complete unsuitability for handhelds doesn't matter; even if you hate copilot and the MS upsells).

Yeah, likely true. Also, if you're intentionally looking for them, half the people you think are trans probably aren't.

That's really entirely the fault of laziness by the IRS and/or Congress. We should have laws requiring all of those companies to provide the complete set of information necessary to file your taxes in a computer-digestible form. There's no excuse for having to manually change several *hundred* lines one at a time to tell TurboTax that they are short-term or long-term gains, or whatever the one random piece of information that it needs from my Edward Jones statement every f**king year on a third of the transactions because it is trying to parse a d**n PDF file.

What makes it a nightmare is that even though all of the forms theoretically have compatible fields, they aren't actually standardized in their formatting, layout, which fields are omitted, etc., and that's true even for the easy stuff like 1099-INT, much less nightmares like 1099-B. And they are provided in formats that are intended for human consumption, not software consumption, so they're having to do crazy amounts of interpretation to figure out what the numbers mean and how to correlate them with other things on a page. This is the stuff of nightmares.

Instead, these data formats should be standardized with a mandatory standard format (XML, JSON, etc.) and shared schema. Providing data in that format should be a hard requirement for all financial institutions, and if a financial institution's data is unparseable by standard tools or is wrong in any meaningful way, the company that provided it should be on the hook for the cost of any additional interest and penalties caused by the taxpayer relying on that data blob.

Once you have that sort of strict data portability and interpretability codified into your tax code, tax filing software *should* become easy, because it's just shuttling data from one strict standard format into another strict standard format. This would be very easy for the financial institutions to do, because they already have the data. It's hell on earth for TurboTax to "Intuit" from human-readable PDF files. (See what I did there?)

In case you never took that course, the classical economist David Ricardo figured out that if you were a tenant farmer choosing between two lots of land, the difference in the productivity of the lands makes no difference to you. Thatâ(TM)s because if a piece of land yielded, say, ten thousand dollars more revenue per year, the landlord would simply be able to charge ten thousand more in rent. In essence landlords can demand all these economic advantages their land offers to the tenant.

All these tech companies are fighting to create platforms which you, in essence, rent from them. Why do you want to use these platforms? Because they promise convenience, to save you time. Why do the tech companies want to be in the business of renting platforms deeply embedded in peopleâ(TM)s lives? Because they see the time theyâ(TM)re supposedly saving you as theirs, not yours.

Sure, the technology *could* save you time, thatâ(TM)s what youâ(TM)d want it for, but the technology companies will inevitably enshittify their service to point itâ(TM)s barely worth using, or even beyond that if they can make it hard enough for customers to extract themselves.

I think you're understating the odds a bit. The average person meets 80,000 people in a lifetime. If your numbers are correct and trans women are 0.35% of the population, then on average you will meet 280 in your lifetime, which is a far cry from it being easy to go your whole life without meeting one.

This ignores social aspects, where in some parts of the world, you may go your whole life without being aware that you've met one because they all go out of their way to hide it, or where in some places you might not meet one because they've all left because of persecution, but that's a rather different statement.

Orderlies are on the 'least affected' list.

I am actually kinda surprised at that one. It seems like it would be fairly easy to automate that one, not even sure you need AI.

It is just scheduling and some image processing/lidar/sonar/etc to make sure crafts are clear of, properly positioned in the lock before raising or lowering.

I don't really know, I am not lock operator and I am not belittling what they do but it seems to me it should be simpler automation problem than trying to do say, Self Driving for a car.

And yet Lofgren usually gets tech policy at least half right, while still having a background in immigration law. To be fair, she represents part of Silicon Valley, and thus presumably has great advisors, but the point remains that being a lawyer shouldn't be an excuse, particularly if you're in Congress. I mean, you're right that the lawyer monoculture in Congress is a disaster and leads to policies being frequently irrational from the perspective of common sense when applied to technology, but I think it's more than that.

IMO, the bigger reason is that Congress is old. The average age of the current Congress is 58 years old. For context, the youngest people who had any non-negligible chance of owning a personal computer as a kid are in their late 40s now. The youngest people who had Windows-based or Mac-based computers throughout their school career are in their mid-40s. You didn't get to the point where half of kids had computers in their homes until about 1996 or so. Want to know how many members of Congress had a 50/50 chance of having a computer at home by school age? Figure out how many were born after 1991 (34 years ago). The answer is six.

Going one step further in our analysis, anyone over age 59 would not have even encountered a graphical user interface until they became adults. So for approximately half of Congress, if they know modern computer technology at all, it's because they took the time to learn it on their own AS ADULTS.

This is a staggering statistic, and explains why you will never see Congress be competent on technology issues unless they get lucky and find really good advisors rather than just listening to the lobbyists; based on age alone, you'd expect a statistical majority of Congress to have no idea whether a technology policy idea was good or bad without help. And this is wildly optimistic, given that most kids weren't exposed to computers (beyond playing educational games on an Apple II) until probably the early-to-mid-1990s.

We don't just need non-lawyer members of Congress. We need younger members of Congress. We need congress to be a representative sample of the people they serve, where the median age is 38.7 years, not 58. I mean, we're not going to get all the way there because of age limits (25 for the House, 30 for the Senate), but having only 1.1% of Congress under age 35 represents a massive distortion of the demographics of the country that leads to poor technology policymaking.

SE did not really simplify much. It was just Windows but they intentionally crippled / broke some stuff.

Much of the bloat, none of the value; does not a product make. Meanwhile by '21 Microsoft was all on o365 as the future anyway. If that is vision what is needed is well ChromeOS essentially a web browser with a bundled HAL.

This is the same problem with that version of Win2k8 server where you could chose not to install the desktop experience...Reality is that did not mean much more than setting the shell=cmd[.]exe in the registry in actual practice. Windows just isn't modularized and does not lend it self to stripping down. You need "a lot' of hardware/storage just to get a basic UI up.

The common Linux stack is much easier to cherry pick just what is essential to run chrome and little but have it all still work right. That is where the heterogeneity of the Linux world is a strength. Various system components and user-land software a like does not get to make so many assumptions about what else will and won't be there.

You said 'a long time ago' so you could be forgiven for the lack of NAC etc, but no dhcp/helper and snooping?

Even with access to wire closet, no DSL installer should have been able to get off the ground in the first place, not in the last 25 years anyway.