Posts by SonofRojBlake 229 publicly visible posts • joined 24 Oct 2008
Incredible Hulk debuted in 1978.
Just to put that into perspective - The Incredible Hulk TV show debuted closer in time to the release of the original King Kong than to today.
And while Lou Ferrigno is not "wimpy" compared to any actual human, I think the point of comparison was either (a) the comic or (b) the more recent CGI version(s), which in either case make any mere human, regardless of physique, look wimpy.
" It's urged them to remain calm if a robot approaches them and not deliberately hinder movements or damage the machines"
I'll bet it has. I don't want to make any racist assumptions about the demeanour of South Koreans, so I'm going to assume that at least some of them are temperamentally indistinguishable from your typical Toxteth scally. Those robots are going to need some very good insurance... but not necessarily for crashes.
"I'm a little bit cautious about the idea of forcing companies that are building AI to enter into bespoke agreements with individual rights holders or an order to pay for content that doesn't have economic value for them"
You're a little bit cautious about having to deal with rights holders? I've got a solution for that - stop using their output. Simples.
You're "a little bit cautious" about an order to pay for something that doesn't have economic value? Well that's perfectly reasonable. If you think something doesn't have economic value, then you have no need of it for your profit-making AI business... right? Because if YOU NEED IT to train your profit-making AI, then by definition it has economic value. I can see why you'd be "a bit cautious" about being made to pay for it, because that'll impact your profits, obviously.
It's clear that the economic model used to justify the creation of LLMs took into account the colossal cost of the compute required, but assumed that the training data would be free. That's like assuming that your costs for a gold mine will be things like shovels, picks, dynamite etc., but not taking into account anyone who might have been living on the land before you turned up and started digging.... and assuming you can just make them, ahem, go away.
I wish I'd known that 20 years ago before I tried impressing my girlfriend by fitting a lock to her bedroom door (don't ask) by drilling into the frame... all the lights went out, and it was then I saw the wire literally GLUED to the doorframe. I get the reasoning now, but it just didn't occur to me. Now I assume there's a wire everywhere I drill.
"the guidelines call for research of several areas, such as operating in harsh environments, creating sensors that detect smell, and high power density hydraulic servo actuators breakthroughs"
Two observations:
1. isn't that research already happening?
2. is "calling for it" going to make it produce results any quicker?
Six years ago the bones of my shin were poking out through two BIG holes in my skin. After some morphine and a quick free helicopter ride, they gave me some ketamine and pulled the bones back inside the skin.
The ketamine trip was the worst thing that happened to me that day.
Wasn't in the room for the leg-pulling, though, so there's that.
"nothings that B&W"
Au contraire Blackadder. On a upper tier COMAH site, you'll find that a lot of things very much are that black and white. It's not about how critical THIS safety system you deliberately defeated was THIS time that you defeated it. It's the fact that you're the kind of person who will defeat ANY safety system EVER without orders signed in triplicate - the kind of person who needs to be removed with extreme prejudice from any facility featuring a major accident hazard, lest your judgement fail and fill the housing estate/hospital/school next door with chlorine or something.
And to state the bloody obvious, putting ear defenders on during an alarm test is not "defeating a safety system" since (a) it didn't stop anyone else hearing the alarm and (b) it didn't actually stop YOU from hearing the alarm, just dulled it a bit and stopped it causing actual damage to your hearing.
I have had to watch a lot of CSIB videos in the course of my work, and the most terrifying thing about them is the coda where they describe how, having investigated the incident described, they identified the cause and disseminated the information to other plants operating the same process.... who ignored the advice because it would cost them money to implement, and the basically identical incident was repeated x days/weeks/months later, with y injuries and z fatalities.
Not an IT bod - chemical engineer here. But you might be surprised to learn how often large chemical factories - some of them handling things that are frighteningly expensive, toxic, explosive or all three - "solve" nuisance alarms by looping out the instrument that's causing the alarm to go off. It happens less and less nowadays, in fairness... but it hasn't stopped.
>Any installations on the Moon could be negatively affected by this.
And if I live to see "any installations on the Moon", I'll be surprised. Space:1999 turned out not to be a documentary.
> I would worry about a big crater
Why?
> debris being flung widely over the Moon's surface
So?
> possible moonquakes
Again - so what? It's the MOON. It's not like there's anything there. Not even sea, and birds, and fish, and twenty thousand tonnes of crude oil, and a fire, and the part of the boat the front fell off. It's outside the environment.
> some of the debris being flung into the space between the Earth and Moon.
There is a LOT of space between the Earth and the Moon...
Again, if you're not getting the idea of scale, just play Elite Dangerous long enough to get an access permit for the Sol system. When you've done that, fly around in that system without engaging supercruise (the in-system FTL drive). You'll be limited to "just" way, way faster than anything we can build right now, and the game will be unplayable because the distances... no, the VOLUMES are, well, astronomically huge.
Crash an asteroid into the Moon, and yes, it'll splash some rocks into the sky. And maybe, possible, some of them MIGHT burn up in the Earth's atmosphere years or millenia later. But seriously - what are the odds, vs. all that lovely cadmium/niobium/whatever RIGHT NOW. Yay capitalism, and so on.
Lobbing an asteroid at the Moon would be a LOT easier than trying to insert it into Earth orbit. deltaV is almost irrelevant - you're not trying for a soft landing.
And the idea that you might miss the Moon and hit the earth instead just shows you've never played Elite Dangerous. Missing the Moon and hitting the Earth instead by accident would be like throwing a piece of cotton over Niagara Falls and accidentally having it thread a needle someone else was holding a quarter of a mile away from the one you were aiming for.
" it's *way* easier to transport the machinery out to the asteroid, mine and refine the materials, then transport back only the stuff you want"
It's way easier than that to just transport a thruster and some fuel out to the asteroid to nudge its orbit so it crosses the path of the Moon in a year or three. That's all it needs to do - it doesn't need to slow down at all - in fact it's arguably preferable if the speed on impact is huge - the innards of the asteroid would then be spread out over the surface of the Moon in a more easily accessible form than would be the case if it "landed" more or less intact. Then "all" you need to do is get your mining machinery to the Moon.
All of this is predicated on the assumption that nobody gives a shit what hits the Moon or where. It'd be a terrible shame if the impact site turned out to be right where the Eagle landed.
I wondered when I would first see it in the wild - late last week a Guardian columnists referred to someone "posting on X", without immediately following it with the words "formerly known as Twitter". I figured that wouldn't happen until, at the very least, links to X didn't get you to Twitter.com (they still do here in mid-October 2023).
It's good... but go here: https://www.theregister.com/2023/09/18/opinion_column/
They deliver TEN paragraphs of story about BMW charging for use of fitted seat warmers and worrying that if drivers let that happen they may start doing the same for air conditioning, then deliver this sentence: "It's more than the 20 bucks a month that jars. As everyone with an ounce of forethought readily realizes:..."
I don't want to spoil the punchline - just go and read it, and if it doesn't make you want to stand up and respectfully applaud, we can't be friends.
I can access those things via my Roku box yes... but I can't skip the ads like I can on a recorded version of the show. This is particularly egregious when in an ad break for Taskmaster I get to see the same advert for a Samsung phone SIX TIMES IN A ROW IN THE SAME AD BREAK. Sometimes recorded is just better. Also, once I've got it recorded, I've GOT IT. It won't suddenly evaporate at the whim of the broadcaster if someone who appears in it has some allegations of dodgy behaviour 15 years ago pop up (just to pick a topical example).
Sky+ subscribers are paying for access to channels, not the ability to record them.
If you want the ability to record the free channels, but have no interest in the Sky exclusive ones, there's a box for that. Literally just one box, two sizes of hard drive, but that's it. Tis good though - all the features of Sky+ (pause and rewind live TV, record two channels at once, record a whole series automatically and so on) for a single one off payment. "Freesat recorder", get yours at Argos, any time you like.
What keeps people using the Sky+ box is, I think, a combination of ignorance (not realising there's an alternaitve) and apathy (it's "only" £40/month or whatever). As soon as I found out I didn't need a Sky box to do all those things, mine was GONE. The only channel I missed was Sky Atlantic, and that was only worth watching for a fairly short period when they were commissioning original comedy. (If you haven't seen "This Is Jinsy"... give it a go.)
"If AI can write you a story as good as the authors can, why pay the authors?"
The outrage here is that the machines are no longer coming for the jobs of the working class who toil and sweat and use their hands. Now they're coming for the comfortable middle class who've got (to quote Pratchett) an indoor job with no heavy lifting. And I think the aforementioned working class aren't going to be brimming with sympathy for the keyboard jockeys who see their livelihood going the way the coal mines went in the 80s.
They're not coming for the GOOD ones - not yet. So far the only ones they can actually replace are the derivative hacks... but most authors, even the good ones, start out as somewhat derivative until they find their voice. Pratchett's "Strata" was a transparent parody of Niven's "Ringworld", and clearly a sort of practice run at a Discworld. And even the biggest Pratchett fan will admit it's not as good as most of what followed (I happen to love it for what it is.)
But I think if someone were able to synthesise a new Culture novel (not a parody, not a reboot, an actual new Culture novel)... I think I'd want it. I'd dearly like IMB back, but if a LLM (with help, presumably, from someone with the right prompts) could make more work that is aesthetically equal to what already exists... why wouldn't you want it? Just out of principle?
"Ogden was the startup's first tech hire and after a few years on the job found himself leading a team comprising eight developers, a pair of sysadmins, an IT manager, and the customer support team"
I feel like this doesn't add up. Mainly because I remember those businesses selling ringtones, wallpapers etc. to people to tech-unsavvy to simply generate them themselves - and that business model seemed only to be a viable thing for about two years at the absolute outside.
Am I just remembering wrong? Were people really PAYING for low-resolution background images for their phones for "a few years"?
I assume my TV has been building an ad profile for me... very badly.
I stream content from Channel 4. It's interspersed with adverts I can't skip (although I can mute them...). I have noticed two very specific things about these adverts:
1. their supposed targeting is absolutely baffling. Example: I am bombarded with ads for Coutts, the Top Person's bank, a place I'm given to understand requires customers to have a current account balance over three million quid before they'll even be considered worth having on the books. Why this company is advertising to my broke ass is beyond me. In fact, why this company is advertising AT ALL is beyond me. I don't think I've ever in my life seen a TV ad for Lamborghinis or Patek Phillipe watches - I've always assumed that people in the market for such things will find out about them by other means. So why are bankers to royalty slumming it in the ad breaks in Taskmaster?
2. The ads repeat. I don't mean that I'll see the same ad several times in the same night. I don't even mean I'll see the same ad several times in the same show (three ad breaks in Taskmaster, and I'd not comment or notice, probably, if the same ad was in each break). No, I mean I'll see the same ad repeated several times in the SAME AD BREAK. The record, so far, was an ad shot like a horror movie with a whole bunch of young, pretty people pushing Samsung's latest pholdy phone. When the ad break started, I look down from the TV to my (not-a-Samsung) phone, but didn't bother to mute. I looked back up when, at the end of the ad, the exact same ad started again. "Huh", I thought, and looked back down again, assuming there was some kind of cute "spot the difference" gimmick going on that I had no interest in. Then back up, when the ad started for a third time. And a fourth. And a fifth. There was no gimmick - Ch4 were just showing me the same ad five times in a row.
There were six advertisements in that ad break - and five of them were the same ad, over and over and over and over and over again. I can't imagine that's what the people placing the ad actually want, is it?
You would indeed need to replace it periodically. But I need to replace the water in the water dispenser in my fridge periodically. It's not that onerous.
If I were designing a liquid cooling system for e.g. something that had to be somewhere sensitive, then there would be a number of things I'd do. I'd limit the inventory of cooling fluid to minimise the consequences if it leaked. I might use something like glycol to give me a lower freezing point. I'd run it through tiny labyrinthine pipes and circulate it to maximise transfer, and when the hot side got back to the reservoir, I'd actively cool it with some kind of refrigerant setup, because with limited inventory you'd need to shed the heat to the air quickly before reusing it. I've more or less described a domestic fridge, there.
BUT...
If I were designing a cooling setup for something I'm just using casually at home, my model would not be the fridge, it would be the sink full of ice water I use to cool beer cans.
For device cooling, step one is make the device heat sink long enough (150mm?) and heat conductive enough that the device and sit on one side of a reservoir lid and dip through it into the cooling medium, which is just tapwater, with optional ice. With an essentially unlimited inventory (a five litre tank wouldn't be terribly inconvenient to have on your desk) there's a HUGE capacity for absorbing heat there. Little tap on it to drain it off when it's hot, funnel at the opposite corner for topping it up from a jug, largish lidded opening for adding ice-cubes. You might top it up once a day? Convection alone would probably be enough but if not a tiny stirrer just to get the water moving would do. Little cheap instrument sitting on top to alert you if the temp gets too high. Can't see the whole setup costing more than £50 retail. That said, can't see many domestic customers for it either - it's slightly more faff than just switching your PC on and going, and slightly more faff = no interest, I suspect.
I lack the expertise to know if anyone cares enough about cooling Pi's and similar to make something like this interesting enough to produce. Anyone?
I can understand keeping your Pi away from water, but... does nobody cool their PC with liquid? Is it really that hard?
I'm picturing a Pi (or NUC) that's a 4" x 4" square box... but poking out of the top of it is an aluminium rod of a standardised diameter. This is the heat sink. All you need to do is dip it in cold water. A glass would do, a nicely designed tray with a good strong lid with a hole for the heat sink to poke through would be better. Chuck in a few ice cubes and you're getting way better heat transfer than you'd get from ambient air. Is this done?
(as in, is this done simply and cheaply in rigs costing less than thousands? I know top end gaming rigs get liquid cooled, I meant something simpler.)
Mid 90s. Early in my career as a chemical engineer, so my memory of details is hazy. Had a job to do monitoring the organic chemical emissions to air of some plant or other. Had a whizzy bit of kit to log the emissions: a flame ionisation detector, which would output an electrical signal, and a datalogger in the shape of a Psion Organiser 2 - the big chunky calculator one, not the cute microlaptop Organiser 3. There was, IIRC, some kind of interface thingy on the top, and we were supplied some cableage to connect the two devices. Again, IIRC one of the plugs was an old 9 pin D plug like an Atari joystick. And the same connector was used for a power supply...
Long story short, it was physically possibly to connect that up incorrectly. Which of course at some point I did. And at that point the magic blue smoke started coming out of the Organiser. So I didn't do any monitoring that day. The organiser and the cables went back to the supplier, and were returned, free of charge I think, with some of the pins snapped off one of the plugs and some of the holes glued up on the socket... making plugging the wrong thing in there physically impossible. An elegant solution, and just a shame they hadn't foreseen the possibility before handing the thing to a wet-behind-the-ears numpty like myself.
Politicians are interested in evidence with regard to their policies.
Unfortunately, their interest is not : "Does the evidence show that the policy will work?"
Their interest is : "Does the evidence show that bringing in this policy will mean people will vote for me?"
And in fairness, that's a rational choice for them.
It's been answered already, but I'll chuck in my 2p-worth - years ago I flew paragliders. There were a few guys who'd turn up to the same hill(s) to fly RC gliders. One night after all the wind had dropped and we'd all walked back to the car park, one of them pulled one of these newfangled "drone" thingies out of his boot and wazzed it up and down and around for a bit. It was impressive. He'd built it himself out of 3d printed parts and bits he'd bought off the net. I think an Arduino was involved. I asked him how hard it was to fly. His reply was succinct: "I'm not flying it. I'm just telling it where to go. It flies itself." He demonstrated by putting the controller on the ground. The drone hovered dutifully where he'd left it. We carried on chatting for a few minutes, him ignoring the thing, me giving it more and more frequent nervous glances. After a few more minutes, it flew back over to the car park and landed next to his car. He'd never touched the controller. Apparently he'd programmed it to fly back to its launch point when the battery was getting low. That would have been ten years ago.
They became drones when you could take them out of the box and IMMEDIATELY, with no training, make them hover and fly to anywhere you want with no training or practice. It's like the difference between driving a car and having chauffeur.
Understatement alert:
"each one costing around $3 million – well below the price of crewed fighter jets like the F-35 or F-22".
Well, yes, accurate, but in the same way it's accurate to say that the height of my original Kenner Yoda action figure (about 6cm) is "well below" my own height (six foot in my socks). For the price of an F-22 you could buy nearly FIFTY of these things.
They spoke to NPR this week about it, but when did the scanning happen? Because if it happened any time recently, these people are scabs, aren't they?
Also, echoing the comment of "why bother?" - surely by definition a background extra doesn't need to have an actual existing person's face, any more than any computer game character has to. Just slap a generic face on a generic body and move on.
Iain Banks (bafflingly no M in this one) had a good idea in "Transition": tourism. Specifically this: it is ludicrously unlikely that a planet has life. It is more unlikely that it has complex life. It is more unlikely that it has intelligent life. We tick all those boxes. But we tick another box: our planet, possibly uniquely in the entire galaxy, has a sun and moon that, temporarily but right now, subtend precisely the same angle when viewed from the surface. Thus, total solar eclipses are possible. This very well may be the only planet in the entire Milky Way where this is so. Thus: if you want to find visiting aliens, don't look in remote woodlands, don't look in secretive airbases... look on chartered yachts floating in the path of total eclipses, yachts whose charterers's cars have suspiciously darkened windows...
If all the conductors in a desktop/laptop were able to be replaced with SUPERconductors, what would that mean?
Is the current in use low enough that the current density isn't an issue? Would it run stone cold and not need cooling? Would it run faster? Would it run longer on a battery charge? Not an electrical engineer, but want to know...
"An AI can no more unlearn a component part of its 'education' than a human being can"
The difference is that if a human learns something as part of their education that they're not supposed to know - that it's ILLEGAL for them to know - then there are certain ethical problems inherent in simply deleting that instance of Homo Sapiens and starting again from scratch with a new one. There is absolutely no ethical reason not to simply destroy the offending AI and retrain it on a compliant learning dataset... and to have to do that every time something non-compliant is found in that dataset. Oh, it's time consuming and expensive? Tough shit Mr. LLM operator, perhaps you'd prefer a job in a coffee shop.
It's not really like that though is it? LLM operators could comply with the law in more than one way. Sure, they could simply shut up shop entirely, but the law doesn't require them to do that at all. It does require them to ensure that training data doesn't include [$certain data]. That's far from impossible, it's just time consuming and expensive, to which their response is "we can't possibly do that, it would affect our business", to which the correct response is "tough shit".
""The Right to be Forgotten may very well be a well-intentioned regulatory protection, and many would argue that it is an important right to be protected. However, there is a clear disconnect here between law and technical reality.""
"Technical reality" is a very slanted way of describing a tool someone(s) designed and built.
This is NOT a case of "well the law says the land can't go more than 500 feet above the sea, but look, there's a mountain". It's a case of "the law says a building can't go more than 500 feet above this street, but we, y'know, built one anyway. Deal with it."
The arrogance is staggering.
I can see why they might want to hold on to the rights. Nobody predicted, in e.g. 1982, that within 30 years there'd be (a) widely available emulation (b) widely available reproduction controllers and (c) mobile as a platform for games, which would lend itself to the short, ephemeral experiences common in early 80s arcade games. So games that might have laid fallow for decades are suddenly marketable again. And who's to say when their time will come?
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
