Risky buy, too. Their T150000 are famous and infamous for Discount Brand Build quality.
Fair. I'm assuming a USB device going into a computer has a power supply has enough current limiting to prevent wires or PCBs catching on fire, but that was an actively bad assumption back in 2005.
Okay there's a large antenna in there, but that just raises the question of why the antenna is larger than the antenna in a wifi dongle? And why add a 2m cable for a wireless dongle?
They badly overengineered it. Both these things significantly improved reliability, because all other things being even a larger antenna will have much better signal characteristics than a smaller one of the same general design, and being able to place a receiver on the front of a computer rather than the back had a pretty massive difference. Some of those decisions weren't even crazy for the time -- the 360 released in 2005, where a lot of people still owned big CRT and plasma TVs designed for play from 6+ ft and built into furniture, and even for desktop computers CRT hadn't completely gone the way of the dodo yet, and especially major vendors will still big fans of making PCs (even gaming PCs) cases big thick piles of steel
There's a lot of more subtle goofiness like this : it uses a custom wireless protocol that was a lot less funky than bluetooth of the time in order to reduce latency from retransmits, for example.
The fuse existing is mostly a problem downstream of how the USB standard evolved. Originally, there was a hard 500mA@5V limit per USB port in the standard, but this was held more in principle than the breach; even by 2000 you could find cheap USB chargers that would put out 2A@5V ish, gfl. Wireless dongles were only supposed to use around 400mA, but even slightly sketchy source (eg, powered usb hubs) could push enough voltage to get at the fuzzy edge, and even if the dongle could tolerate that wider range, the wires (and some sources) wouldn't be able to supply it permanently without damage. As time went on and those skuzzy chargers became more common, it was just accepted (and saved money) to use fuses a lot less or with a much higher tolerance than the official rating, under the presumption that devices which could be damaged by sourcing more power would have safeties against it.
Why it failed is unfortunately probably more boring. While there are some potential weird cases (Five Below-brand USB hubs, badly implemented cell phone chargers, putting it on top of a wireless pwoer charger), chances are pretty good it's just time and entropy. Fuses, especially older SMD fuses, are both temperature sensitive and relatively fragile devices, since they work by breaking. Over time, a 500 mA fuse will become a 470 mA and then 450 mA, until eventually the intended current passing through the device will bust it. That's worse on wireless devices, since the antenna is basically an inefficient hot plate, and worse still with big PCB antenna like that particular dongle made, and worse still on devices like this that were pretty close to the edge of their power envelope to start with. The new fuse, especially if you bought it recently, is almost certainly going to be much more reliable, even under the same conditions. If you want to be extra-safe, I'd unplug it when you've got long periods where it's not in use, but it's probably going to be good for another 10 years.
Last one was April 2024. I was talking with Ymeskhout about trying for one in late March of that year, but even when we fired up the Discord call it didn't really go anywhere successful, and then his last post on the motte was last April, and he deleted his x twitter account somewhere around the same time frame. Still active on substack, so doing... well, I won't say okay, but still typing.
For retro or retro-like games, there's some advantages -- trying to play something like Megaman X or Legend of Mana on the standard Nintendo layout means either fighting with an analog stick that's either too eager to have you go diagonally or nonresponsive to small movements (or both!), or offsetting your left hand to get to the dpad and feeling a bit like a crab a couple hours later. There's even a few specialized retro handheld consoles that will let you switch what layouts are present for this specific use case (and to support six-button controllers like the Genesis on the right hand side).
But dunno if there's that much on the Switch market that's actually focused toward those games.
A 'blade'-style automotive fuse housing costs 50-75 cents at scale, and a glass vial-style fuse holder costs 30-50 cents, and the fuses themselves cost money on top of that. A single 0606 500ma fuse is cheaper at unit one digikey prices; at scale, they're basically free. They're probably only included at all because they're part of the USB standard -- overdraw could and does kill older motherboard USB ports, and rarely even entire motherboard USB controllers! -- not that it stops designers in other cases.
Ideal answer would involve a PTZ self-resetting fuse (though they're not great options for USB devices people are likely to leave plugged in indefinitely, and usually end up 5-10c), or a through-hole conventional fuse that could be fixed by anyone with a soldering iron and patience (though mixing SMD and through-hole parts gets stupid expensive, esp for boards with SMD on both sides). But it's not as nonsensical as it looks at first glance.
There doesn't seem to be anything special about this form of getting money as opposed to any other form of getting money (except that it's bad and the left did it, so it's a chance to get in a partisan dig).
It is, potentially, a massive amount of money; it can, potentially, be specifically targeted and legally obligated to be used for a specific partisan activity; it also leaves a massive ideologically-unappealing penalty that will often be directly acting as a reminder while waving signs on the lawn of the bad actors in question.
Uhhh, so how does that help? Is that what was demonstrated to work in the past? Did prior Democratic administrations actually fix something about the banks or whoever they sued when they got money from them? If not, then ???
The Democratic administrations did, in fact, get the banks (and many tech companies) willing to bend over backwards out of fear of costly not!fines which would sent to activist groups that hated them and would have the backing to bring other costly lawsuits. I wouldn't call it fixing, since I don't have the same goals as the but the banks drastically revamp their behaviors for more than a decade, even through the first Trump admin, both on who they allowed to have accounts and who they didn't.
There's reasons that might not work for the Republican Party -- judges tend to treat colleges better and Republicans worse, having an adequate supply of favorable news coverage seems like it was important, the Red Tribe does not have as many of the relevant dedicated administrative agents required, and there's just a second actor disadvantage. But it's not an Underpants Gnome proposal.
It doesn't reduce the ability of the federal government to act against universities, if that's what you're asking. But that ship has sailed; no one has any proposal with any chance of working to do that. If we want university administrations to be less likely to actively discriminate, and to not promote hilariously fraudulent partisan activities under the auspices and honors of 'research', I'd love an answer that wasn't the government's carrot or stick. But there's zero idea on how to do that.
Your own proposal of requiring administrators to affirm things isn't even coherent within that framework, but it's also a joke given that these orgs were long supposed to already be affirming it, and were more likely to get in trouble for fucking with an antivirus setting than for putting out Whites Need Not Apply signs.
This (bad, partisan) way of getting money may be doable and hard to undo, but it seems to not even have a passing familiarity with solving any of the actual problems we set out to solve.
I think it does. There are several extant lawsuits focusing on unlawful DoE discrimination against disfavoured minorities, university discrimination against disfavoured minorities, of widespread fraudy behaviors by colleges and their research components, and that's before the widespread tolerance or outright advocacy of political discrimination or violence. Many of these orgs running those lawsuits have a lot of focus on these problems; many of these lawsuits are focused on the very specific issues that impact the ability of academic institutions to perform in their claimed roles.
And those are just the lawsuits already in pipe. A lot of the other stuff doesn't have lawsuits floating around simply because any lawyer worth their salt knows without a friendly federal admin it'd be a vanity suit.
Again, I'm not convinced this will work! But again, it's also far from Underpants Gnomes.
The trick's that the same chips used to produce a model are also usable to run the model for someone else, and a lot of the technologies used to improve training has downstream benefits on inference or implementation improvements. Every AI vendor has its own complement to turn into a commodity.
If they are selling inference for more than what it costs them in chip deprecation and electricity, that is only because they have a moat in the form of good models. If they ever decide to stop burning through money to make more powerful models, they will quickly find that without that moat they will only be able to charge the same as any rent-a-chip company.
User interface, agentic behaviors, and (down the road) deep research tooling matters, a lot. It's possible to set this stuff up even as a single dedicated user, but there's reason that approximately zero people have home deer-flow setups, and there's vast economies of scale once you do configure them
We can just read your comment and see that that's all that you had. How is that supposed to work? Give me an example, an idea, a process, an anything
The specific behavior is called a cy pres settlement; where the recipient of a settlement is not available, or where their personal damages represent only a small portion of all people harmed, a judge may authorize a large 'donation' to a third party as part of a settlement.
The easiest case is where the federal government is acting as a 'friendly' defendant. Rojas v. FAA? The FAA can suddenly have a change of hard, and decide that in addition to giving a million bucks to the harmed parties and their lawyers, they can also want to give a hundred million dollars to a I Hate Affirmative Action group.
There's limits to this approach; while cy pres settlements are very hard to challenge, it can happen, and some settlements in general end up worth no more than the value of the toilet paper they were written on after an administration changes. A naive person would argue that recent court cases have shown the willingness of Biden-friendly judges to put the kibosh on those efforts; a remotely aware one would recognize that those principles don't cut both ways.
But it's still a powerful tool, and one that's very hard to undo. Meanwhile, thanks to the very slow pace of any attempt to bring a court case to full and final judgement and the increasing tolerance of standing gamesmanship, it's near impossible to actual complete a judgement by putting a law on the books, or force an unfriendly administration to do anything.
((Though not impossible. There's another very dangerous option, and that's intentionally arguing cases as poorly as possible or with such 'incompetence' as to be sure that the courts will not 'agree with your claimed position'. As I continue to be fond of pointing out, Guiliani could absolutely use a job where making false claims, butt-dialing privileged information, and making incoherent arguments is tolerated, and the feds love two out of the three.
You would think, given the impact of res judicata, that this would be extremely harmful, and you'd be right! Too bad fewer and fewer people care.))
Given theNybbler's environment, the first thought would be credit card skimmers. They're not common, but urban areas finding 'modifications' to gas stations happens enough I've seen it in person. Moving to NFC/smart card transactions rather than magstripe helps, but we haven't fully migrated over, in no small part because the US chip cards kinda suck, so we still have magstripe and a lot of card issuers having to deny sketchy transactions. Most people who do it get caught pretty quick, but only after doing a lot of damage.
Less common is the old analog loophole: someone writing down (or taking a picture of) a card number, name, and CVV code.
Most online vendors also need zip, too, but if you're stealing card numbers in a diner, you can make some educated guesses about zip codes pretty easily.
I've run a few 4-byte quantized 70B models on a small home gaming machine pretty easily (Intel i3-13100, nVidia 3060, 48GB RAM). It's a little slow -- non-MoE models can go into a couple tokens-per-second, and MoE seldom go higher than 10 tps -- but there are some set-and-forget use cases where the difference isn't a big deal, and you're just a couple GPU generations away from it going faster.
Both ollama and lmstudio work pretty easy 'out-of-the-box'. You can dive down the deep end if you want, and start moving to vllm or others, but it's far from necessary for most use cases.
Scaling up without waiting can get expensive, though. Used server GPUs aren't ludicrously expensive and buy you more RAM (and thus more context/bigger models), but they're slower than current-gen (or even two-gens-old) gaming cards. Trying to break past 24GB VRAM gets into the kilobucks range, and while nVidia says that they're dropping a card that will change that in a few months, it'll probably be seconds before it get scalped. For LLMs, processing power is lower priority than total memory bandwidth, so you can get away with some goofy options like the Ryzen Max series and run 128 GB ""VRAM"" with a CPU, but setup is more annoying and throughput suffers a lot, and it's still not cheap.
There's also just a massive amount of equipment, material, hardware, and even facilities designed around imperial units, sometimes practically irreplaceable. Switching over to metric, even solely for new projects, isn't just or even mostly a matter of getting people to use new units on drawings.
One of the nonimmigrant visa tiny exceptions is :
admitted to the United States for lawful hunting or sporting purposes or is in possession of a hunting license or permit lawfully issued in the United States.
It's... much easier to argue for recognized sports or permitted hunting than for machine gun tourism, though I'm not aware of any prosecutions in either case.
Not aware of the law here: what's the legal status of illegal immigrants possessing firearms?
Federally speaking, there's a specific statute prohibiting possession (or sale to) to illegal aliens, or to legal aliens on non-immigration visas (with a tiny number of exceptions not relevant here):
It shall be unlawful for any person... who, being an alien... is illegally or unlawfully in the United States... to ship or transport in interstate or foreign commerce, or possess in or affecting commerce, any firearm or ammunition; or to receive any firearm or ammunition which has been shipped or transported in interstate or foreign commerce.
The ATF has taken an unusually even-handed approach to this matter and does not consider the bare possession charge to apply to nonimmigrant aliens (though they can only purchase lawfully from private sellers), but it defines those who have overstayed their visa as specifically not in nonimmigrant status.
Theoretically speaking, this only applies to firearms or ammunition which has been shipped or transported in interstate/foreign commerce, but I wouldn't be my dog's life on the ATF making a distinction.
In practice, prosecutions are rare.
Constitutionally speaking, there's been some recent cases about how much the illegal immigrant must know that they are illegal, (caveat: I can't find if he was retried; the man was almost certainly guilty under the new standard of proof, but that doesn't always mean much). The prohibition itself hasn't made it to SCOTUS, but it's been pretty universally upheld by appeals courts. Some states prohibit possession by even federal-permitted lawful aliens (or even non-citizen US nationals), and those are on sketchier constitutional ground in my opinion, but they've also been difficult to challenge for procedural reasons.
He's already a State Senator, but fair. On the other hand, we also know exactly what sort of demands get thrown out -- and sometimes half-hearted apologies still offered -- when random no-name chucklefucks that are too far to the right for their own base and electorally irrelevant get.
Most of the better-publicized examples tend to be culture-warry and tied to emergency response stuff that's hard to measure directly, but there's a lot of stuff in this class, too. For a well-documented-if-poorly-known one, I'd point to checkrides.
Pilots are required to pass a checkride for their pilot's certificate and for a variety of add-ons after that point. These exams are lengthy processes that can only be provided by FAA examiners directly, or by FAA-approved examiners called Designated Pilot Examiners (DPEs). FAA Examiners offer the service for free*, but have become increasingly unavailable over the last twenty years; in the modern era, >95% of exams are operated by DPEs, amounting to tens of thousands of exams per year. Because of the exam's complexity, it's very rare for a DPE to do more than one exam per day, there are a wide variety of practical constraints due to weather and other environmental conditions, and there are less than a thousand DPEs in the entire US. That was in an awkward but plausible equilibrium for most of the 2010s, but post-COVID, there was both a glut of new pilots and a lot of DPEs who had drastically reduced availability (it's very difficult to make a full-time job, so you get a mix of retirees and weekend warriors), along with other constraints getting baked into the system that made it hard for remaining DPEs to maintain the same velocity as before.
As a result, if your flight school did not have a staff DPE (technically against the rules, but largely tolerated), it could take months and cost over a thousand dollars to run the test for your initial pilot certificate, and if you failed -- or even if you had to cancel because of weather! -- you'd have to pay it a second time later. Most students also had a maximum time between graduation from their flight school's internal tests and when they even attempt a checkride, so other delays could lead to even more costs. This was a very well-known problem in pilot communities to the point I'd heard about it by March 2022. By 2024, a law passed with a specific requirement to start an office specifically monitoring the problem and by 2024 Congress had sent the FAA a further letter asking what the fuck was going on. Complete mess, entirely an infrastructure and coordination problem, zero culture war politics...
And a lot of internal political problems. Flight Standards District Offices (FSDOs) manage DPEs in their geographic area, and while there's been a long waiting list for DPE applicants, FSDOs don't like actually certifying them or managing a large number, both because of the recurrent inspection overhead and for more interpersonal reasons. (I dunno if the DAR/DER stuff is any less bad, but I've heard stories.) And once you became a DPE, the gig was extremely renumerative while their shortage existed, and coincidentally the people who did get to become DPEs inevitably were or became well-known by the FSDO. Fixing this was, inevitably, going to ruffle feathers.
But allowing a snowjob of a biannual report to float through with a general Solving Inefficiencies was easy. Guess what we got? The first biannual report revealed that the FAA, six months in, still wasn't trying to collect data on how much DPEs were charging. Almost zero information about why FSDOs had so low a pass rate for DPE applicants, or why wait lists to apply as a DPE were so long. They did switch around a lot of individual DPE in and out in the local area (sometimes without even telling ex-DPEs why!), as if they only problem was the physical offices of those DPEs, and for a good year it actually got worse in our area. Modernized the search tool, and it's almost impressive how bad it is. Absolute epitome of following the streetlamp effect off a cliff.
Tbf, it's still early game for the current admin; I don't have great hopes.
It might work for the ICE attacks, where we have a clear policy, perpetrators in custody, and alignment with other groups. I’ll say that Democrats could and probably should do that.
I think Democrats should do that. I don't think they can, and I'm pretty willing to bet that they won't.
Trivially, there's not been an outpouring of support for ICE being attacked by a literal roving mob. Bringing in the national guard to fix the problem resulted in a minor constitutional standoff rather than an embarassed looking-the-other-way. Newsom, widely suspected as the frontrunner for the next Dem presidential primary season, is in the middle of fighting ICE on several fronts, a good number of them ranging from mildly to hilariously unlawful, nearly all of them bad ideas, along with his more general accelerationist behaviors. A judge is on trial for concealing an illegal immigrant, and the state governor opposes it. The new Texas candidate for AG is on news today talking about how ICE invited this attack.
There's been no willingness to budge on sanctuary cities, even in the most egregious cases that everyone with the slightest clue knows is going to blow up in pro-immigration faces. There's no triangulation, no Sister Souljah from a Dem going onto ABC and smacking someone for saying gestapo. On this very forum, we have dishonest partisans who can't go further than promoting the Lankford bill, even this week, without ever confronting the serious flaws in their claims.
Not so much for the assassins. No surviving perpetrators.
Of the 'successful' ones, we have Kirk's killer sitting in a jail today. Utah's asking for the death penalty; tell me if you can find a Dem partisan who wants the murderer to fry and doesn't call the shooter a groyper.
One attempted assassin was on trial literally yesterday, tried (poorly) to kill himself with a pen after being found guilty. Another went trans and has sentencing coming up soon. Supposedly well-respected people aren't sure if the Zizian attacks 'count' as left-wing (later deciding no!). How has the coverage on the left side of that aisle looked, to you?
And, yes, if we include Luigi fandom, he just got murder one dropped down to murder two for his trial, and even if you think that's a necessary conclusion from NY's esoteric statutes, we have wide evidence of what happens in other legal cases where prosecutors or the Democratic party don't agree with a specific statutory interpretation, and this ain't it.
No comparable groups to discredit,
Several anti-immigration and anti-abortion groups were discredited for merely having similar-sounding names.. I don't think that's healthy, but we have Options, here. People in the 80s and 90s who were too happy to bring down the hammer on organizations that merely echoed the language or defended killers or violent protesters on the right, and we now have a surfeit of test cases. We'll see how it looks in a week!
no social networks to ban,
Funny to mention that!. ARFCOM had its registrar boot them based on rhetoric that merely looked like that of violent protesters; YouTube booted gun owners for things that they merely thought might ever look illegal. We'll see how this looks in a week.
no pet issues to excise from the party planks.
I mean, 90s Republicans didn't abandon anti-abortion or anti-immigration or pro-guns positions entirely. They just made massive and often-painful compromises until they could rebuild political capital. Do you think that's going to happen, here?
And widening them enough to punish all of antifa, the Democratic Party, or “the left”…that’s going too far. I understand that it feels natural for an average Republican to make that equivalence, but I believe it’s wrong.
I'd love to agree with you. The trouble is that I can absolutely agree with you, and also have nothing incompatible with :
It seems to me that not only does the left have a very serious violence problem, but that there is no one on the left capable of engaging with that problem in anything approaching a constructive way. Simply put, the American left has invested too much and too broadly into creating this problem to ever seriously attempt to resolve it. There is no way for them to disengage from the one-two punch of "The right are all Nazis/Nazis should be gotten rid of by any means necessary"; too much of what they have built over the last decade is predicated on this syllogism for their movement to survive even attempting to walk it back. The vast majority on the left cannot even bring themselves to admit the nature of the problem. But at the same time, at least some of them do seem to recognize that this is getting out of hand in a way that may not be survivable.
Futzing around with force myography for gesture recognition. I'm not convinced it's better (or even 'good enough') compared to electromyography, but if I can get anything workable, it seems likely to be a lot more comfortable for prolonged wear.
There is currently no Hi Point firearm that it is lawful to sell new in California. Hasn't been since January 2024.
That's an unpleasant problem, but it's also a fully generalized one.
This is, of course, why no one ever compared Mitt Romney or George Bush to Nazis.
What constitutes a “serious attempt to resolve” this situation? Does it involve public disavowals by the leadership? Cancelling any streamer stupid enough to say something edgy? The DNC taking responsibility for a terrorist act like it’s al-Qaeda? Maybe some time in the stockades, or a few televised executions? What would it take for you to feel like “the left” was making a good-faith effort?
... we kinda have examples, here.
To go to the Charcoal Briquettes Rant, the aftermath of the Oklahoma City Bombing had a wide variety of 'compromises' lean heavily to the progressive, specific legislation to make that form of murder or its advocacy more difficult, and while the execution of the bomber (under Bush!) wasn't televised, it's one of the very few times that the Left and Right agreed on just straight up killing a dude. The aftermath of anti-abortion murderers resulted in near-total ostracization of anyone promoting violence as a means to end or reduce abortion, long serenades from both political and spiritual leaders, the 'compromise' of the FACE Act, and several executions (at least one under a different Bush). Even minimally-violent disruptive protests that are wildly tolerated in other contexts are given a long standoff from mainstream Republicans and antiabortion advocates, today. Only two out of three people in the Flores murderers got a life sentence, but it also lead to widespread delegitimization of even unrelated border groups, in some cases treating them like terrorist organizations.
I'm not going to demand anything that strong, but I made a long list of things that would have surprised me had they happened, and of them, the closest to a 'success' story also has someone saying "But I will not reflect on our shared humanity, nor will I mourn his passing."
Your argument ab adsurduem has been table stakes.
- Prev
- Next
Some domains they struggle with wordplay for tokenization reasons, especially for matters like rhyming, counting syllables, so on. Dunno if this is one of them.
This may be an artifact of the LLMs you're using being trained to avoid the idea space you're working around, although like SubstantialFrivolity I'll admit I don't know the 'right' answer, either. Nevoria, which is trained on a lot of sexual chat, gave ""A Peter Puffer" and "A Rod Rocket", which doesn't seem awful even if I'd have gone with something like "Fred Fluffer", but I get the feeling that there's some specific domain knowledge that I'm missing.
More options
Context Copy link