Posts by Howard Sway 1626 publicly visible posts • joined 12 Oct 2020
So simply making something work is now seen as "improving the user experience" is it? That implies that having something that flat out doesn't work is an acceptable user experience, albeit one that can be "improved". Microsoft's historical attitude of happily selling unreliability seems to have migrated from the desktop to the cloud along with their software.
Won't happen because Musk is incapable of charm. He is however very good at being offensive.
Good luck to any new executive that dares to tell him the truth - that the only way the advertisers are coming back is if he shuts his arrogant mouth.
Looks like Azure is getting more and more complex, combined with a desire to change it frequently with agile "sprints".
Frequent change combined with increasing complexity is only ever going to go one way : more disasters and decreasing reliability. And it sounds like their response to this problem is to keep applying sticking plasters. A more fundamental rethink of approach is going to be necessary at some stage to prevent it falling apart.
"Congratulations to Devon & Cornwall Police for completely eliminating crime in their area for the last 3 months. A team will soon be visiting them to learn how we can apply their approach across the whole country, and ensure that the whole country is completely crime-free by the end of this year."
This is erroneously assuming that there is now a worthwhile "Twitter experience" possible without encountering trolls. Trolling is now its chief purpose, so just block the entire site.
Even better, offer a paid $10 a month "green tick" service to individual twitter users, where you check they're NOT trolls and add an exception to allow access to their account. Musk would like that.
So there are now 25,000 different types of coin listed by these crypto pushers. Presumably when someone this week thinks "you know what, I'm very late to all this, and I've heard about quite a few scams and people losing everything, but I think I'll invest every last cent I have in one" they then go and "do their own research" in time honoured internet fashion, i.e. go down a rabbit hole of crypto-pushing sites and get pushed one way or another into picking one to waste their money on. It'll be either "pick a popular one, it's safe" or "take a risk on a smaller one : you could get rich beyond your dreams". Both options are of course bullshit.
If the $8m that got wasted on "Jimbo" is taken as a sort of baseline of how much gets blown on even the most obviously shit coin scam, that means at least $200 billion must be tied up all in the small fry, waiting to be stolen or just evaporate away into nothing. What a fucking stupid waste.
Most sensible investors try to understand them before deciding to invest in them. Piling money in without thinking and then saying "turns out this was worthless crap" once you've lost it is about as wise as thinking that there is such a thing as "lucky numbers".
Incidentally, there are 18,181,818 new shares available in my new AI crypto-metaverse company if you're interested, only $20 each.
Wow, I would never admit to assuming something like that. Especially as "Ben" should have been able to form an opinion of the MD's technical nous during his time there. The fact that the boss of this tiny company didn't already have an account on the system, and couldn't create one himself is enough of a clue as to the level of his Unix skills.
Plus, if you have to do something like this, at least stay and watch the boss login once you've created the account, so they can't later accuse you of giving them a password that "doesn't work".
So you put in $10 million to buy some crypto company that looks completely fraudulent due to massive broken "promises", and state that you intend to take out $35 million a year in "fees". And some regulator looks at this and decides it's all above board and approves it, because they presume that this business model can generate enough to cover the full deal, and the money will keep rolling in after news of its dodginess has emerged.
How is this not a licence to swiftly extract another massive share of the money from the fools who invested in yet another worthless coin conjured from thin air before it all completely collapses?
They've just shown some of it on the news. What an embarrassment, hearing Musk giving it all the "is it still working?" stuff was priceless.
It may well have been a clusterfuck because he'd sacked so many staff, but I'm guessing that a lot more are going to be fired after this.
Great idea : developers never need to fix problems during off-hours do they? And even if your client is losing millions in revenue, it's too much of a risk to contemplate being bilked even more for cloud processor time............
If this "helpful feature" doesn't make you see that cloud-based developer environments are a terrible rip off, frankly nothing will. Anybody incapable of setting up their own development environment (or a common, standard one for a company) isn't worthy of being a developer anyway.
I'm sorry, who has started callng them UAPs instead of UFOs? Anyone you know? Hell, even the spellchecker checking this comment in my browser is happy with UFO but not UAP. Ever seen a headline saying "man claims he was abducted by UAP"?
It's not even linguistically necessary : the distinction between "flying" and "aerial" is very pernickety, as is that between "object" and "phenomenon". The people pushing for this linguistic change about unexplained sky things are not my cup of tea, indeed they deserve to be softly hit by a flying saucer.
Something like this was tried in Britain in the late 1960s, a campaign called "I'm Backing Britain" trying to get people to work longer for free. Needless to say it didn't last long.
It even had a truly terrible theme song, sung by Bruce Forsyth, which you can listen to here
Earn 10 bucks a month. That'll be the level he soon arrives at.
As a science fiction "author", he's surprisingly not yet seen that some kind of random generator of tropes, character types and plots could generate the input to feed into the LLM in huge volumes too, producing hundreds of these crappy "books" per hour, which could ultimately make him several dollars richer, and the world a little bit worse.
Textbook example of an organisation being lured onto the rocks of money-wasting pointlessness by foolishly succumbing to the sirens of a stupidly overhyped tech fashion that's a solution in search of a problem.
For an industry that's meant to be analytical and have a keen insight into productive uses of new ideas, it has an overabundance of people who do little more than act like youngsters chasing after the latest hip new thing, encouraged of course by too many influential "thought leaders" keen to jump on any opportunity to profit from the silliness. If only all this noise could be ignored, and more attention paid to things that have proven themselves in practice, we'd all be better off in the end.
Yet more money poured into meaningless temporary little PR exercises about launching strategies for opportunities for future unicorns ....... much easier than doing something that actually produces something, even if all they actually get for the money are a few positive headlines because of the illusion of activity it generates.
Of course they have, so much easier for a Tory to convince themselves that an increasingly authoritarian state is a good thing if someone's making bucketloads of money out of it as well. Even better if you get to put your own nose in the trough. If you're one of the 85% of innocent unfortunates who get wrongly identified just suck it up. At least we got 15% of people who were actually guilty. Probably.
I know things don't work this way, but the image I just had in my mind of Musk turning up one morning in his Tesla and being confronted by big piles of Twitter's office furniture and PCs outside the building and his entry card no longer working is quite delicious.
I doubt they'll wait for AI to improve - as it's BT, they'll just impose it on their customers anyway no matter how crap and useless it is. After all, it's the "new big thing", and using it despite its' complete unsuitability for interacting with real people about real problems will make them look excitingly modern and cutting edge to the more credulous market analysts who influence their share price.
My first thought is that these rules will just give these "assets" a veneer of respectability as an investment, when it would be better to just point out that they're nothing but speculation, intrinsically worthless, regularly subject to scams and a really bad idea to put your money into.
Unfortunately this does nothing to solve the criminality and money laundering issues, but if enough people stop foolishly putting their money into them they will collapse as a store of "value" anyway, which ultimately would be the only way they stop being used for these purposes.
The solution offered seems to make it very clear why the error is being triggered. Mainly that Henry Fraud has been reduced to looking for underhand ways to keep the cash rolling in to keep his auto empire afloat.
And he promises, absolutely promises, that he'll never in any circumstances interfere in her job in any way or overrule any of her decisions even if he disagrees with them.............................. Sure.
If she attempts to disinfect the sewer at all, I hope she realises just what sort of awful reaction Elon's incel fanbois are going to have. And finds out in advance whether Musk is going to react by saying "Sorry, that's just free speech, can't do anything".
They need to explain exactly what they mean by "product" here. Do they mean commercial, compiled applications? Is source code a product? It can be if you pay for it, but it doesn't actually do anything until you compile it and release it. A more useful way of describing what needs to be regulated would be "end product" or "commercial product", indicating it has been sold for profit. Frankly I've bought way too much shoddy crap commercial software, to object to suppliers at least having to prove that it meets some kind of quality threshold. But I don't see open source as a product, it's just available code that you can turn into a product if you want to, so why would it be within the scope of this legislation?
Did they have a meeting where they specifically discussed how they could make it much easier for cryptominers and other machine hijackers to misuse other people's computers? Or has nobody yet clicked that allowing arbitrary low level web assembly code downloaded automatically from a webpage to access hardware is a really really bad idea?
Silly me, I thought that during the last 30 years of making IT systems, I was doing my best to remove the drudgery from people's work by automating it as far as possible. But that required understanding and, yes, creativity to do well.
AI will remove the drudgery of work for people who mostly produce output of questionable value, by removing their jobs from them, and producing even more output of questionable value. Creativity is the last thing it will show, as it produces the same bland goo for everyone that uses it. It might occasionally spark off a useful thought in someone somewhere, but in general saying "you can turn your brain off now, the computer will do it all for you" is a bad idea when original ideas are much more valuable than output from these still rather new tools.
Presumably the non-leaders will be encouraged to follow the example they set. So they can change to more inclusive talk about management such as :
My boss is as useless as a fairtrade chocolate teapot.
My boss couldn't organise a tasting at an organic fruit smoothie cooperative.
My boss is as non-verbal as a bag of nail-placing implements.
The reason we want this data that companies collect anonymized is not because we fear that someone else might get their hands on that data and do something awful with it.
It's because we suspect that the companies who collect it will do something awful with it.
Some people have described him as being the same thing on the internet, because of what his company allows.
Anyway, brag not accepted without proof. If only he owned some kind of website where video evidence could have been posted, the whole world could have had a good gawp at his alleged skills.
I disagree with this. LLMs are big business right now. They are commercial programs into which copyrighted works have been knowingly fed, in the full knowledge that they will form some part of the output. One surefire way to lose copyright cases is if it can be proved that you intended to knowingly breach the copyright, which seems easy to prove when that copyrighted work appears verbatim in your LLM output, and you have to answer in court whether that copyrighted work was inputted into the model.
At that point, a defence of "I don't know, we used tons of copyrighted stuff to train it" isn't going to work.
Remember that his big plan is to turn Twitter into the "everything app" that you trust all your money and online activity to. Looks like he needs to hire a lot more more rigorous and experienced engineers throughout the company if it's not going to end up as the "everything about you and your finances has just been leaked online app".
He'd better do it quickly, otherwise there'll be a big pile of stories like this when he launches it and tries to assure people it's "safe".
Right, so you encrypted your users passwords so that they couldn't be misused after a data breach. But it never occurred to you to encrypt their names, addresses, telephone numbers, email, because you didn't realise that these are also commonly used for phishing attacks when they get stolen.
20 years I've been informing companies of this problem, and imploring them to encrypt a lot more of the personal data they hold. And still I encounter the shrug of the shoulders and get told that they don't think it's that important and will make more work for them. Still, this story is one more example to add to my file of reasons why they're wrong.
In other words, we've screwed everything up, so now in an act of desperation we're going to allow a complete free for all, in the hope of bringing in lots more activity. And the problem that this will see lots of malfeasance resulting in investors left penniless? That's just an unfortunate risk that's inevitable as we try and win yet another race to the bottom.
It may not be high quality any more, but it will be cheap : it's the Poundland strategy steadily being applied to the whole economy, a really stupid road to choose to head down.
Amazing that none of them saw what it was going to be, and how rich it would have made them. What's not often remarked upon is how long it really took to take hold as the dominant internet service. Most people first saw it in those days as an additional add on service grudgingly offered by AOL / Compuserve, on top of their proprietary protocols, or getting their first taste of Netscape from coverdisks on computing magazines. And the number of sites was so small that the main way of finding stuff was through AltaVista who tried to classify the whole web into separate categories.
As usual, the funky early days were a lot more fun before corporate mega-greed and market domination squeezed out the entertaining experimentation to the margins. But it's a credit to the openness of the underlying tech that this will never be totally squished..
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
