Posts by Alex Stuart 110 publicly visible posts • joined 20 Jan 2021
Generally, Big Tech has been running rings around governments for years. Our governments are slow, lumbering things, and by the time they've caught up to BT on naughty thing A, they've already rolled out B and C, and so on. It helps to have essentially infinite money to a) pay an army of lawyers to drag out appeals for as long as they can get away with and b) shrug off any punitive fines. They behave like they're above the law because they essentially are.
I am no big Apple fan because I am a tinkerer, but Meta make them look like angels, they are an incomparably more moral company.
Apple are not faultless, but they make genuinely good products, care about the impression and experience people have about their products, and their security/privacy (to the kind of extent expected these days)
Meta would happily get the entire world hooked on slop for 8 hours a day from childhood if it made them extra billions in ad money. Zuck is either a sociopath or a synth. Teenage depression, addiction, political polarisation, CSAM, animal abuse, extremism, data theft? Who cares, just keep scrolling and click that ad, the lawyers will see the govts in court.
I see your point, and agree that health should be a priority, but you're baking in the assumption that it has to be either or - which may well be true in the UK right now, but is not the case everywhere - see Japan and wider Europe.
I'd rather hammer the point that we fundamentally fail at building things in general, so should fix that, then we can have both infrastructure and healthcare.
70 years ago, we built the world's first commercial nuclear power plant. That we are now paying France gazillions to build another one massively late and over-budget is...a mind-boggling timeline to have ended up in.
Likewise, we invented the steam engine and the world's first commercial railways - this stuff should be easy for us.
IMO buying in at today's prices is very risky. Continued huge growth is priced in, but this might not happen if LLMs turn out to not scale usefully beyond GPT-4 levels. Because then the big capex taps turn off and nVidia goes back to making mere billions as opposed to tens of billions per quarter. Plus, competition in the accelerator space from Google, AMD, Intel and other in-house designs.
On the flipside, if LLMs do continue scaling, or if some other AI technique comes into play that scales...they're going to be selling chips as fast as they can make them until the singularity/utopia/war of the machines and today's prices will look cheap in hindsight.
The calculator usage (at least, for a single calculation) isn't a good example of tool use for exactly the reasons you say.
Replace it with something like 'search the web' and now things are getting more useful e.g. 'tell me the important things to know about dihydrogen monoxide' - the LLM goes off and searches and returns a curated summary to you much faster than you could summarise yourself.
Or 'keep trying to compile code to solve problem X and tell me when you have a working solution', 'keep using this Go app to play against youself and tell me when you think you can beat Sedol', etc
(I'm not an AI fanatic either, at least not regarding LLMs. They have some utility now, but still too prone to hallucination for my liking and I doubt they can reach AGI without some sort of symbolic reasoning system built-in)
I wouldn't mind Settings if it had all the functionality of Control Panel, but this change is almost a *decade* late.
It absolutely begs belief that with the enormous cash MS have been generating for years, they couldn't have put a couple of programmers on a task to build the entire Control Panel functionality in Settings with Win10 UI. Something is seriously wrong with how the product is managed. Possibly related to all of the 'product managers' they wheel out for discussing new features being hired in place of actual coders.
Indeed.
Can someone savvy in this area explain why newer versions of Android couldn't basically *just work* on older ARM-based phones the same way Windows does? I have a 13 year old rig that runs modern Windows fine, it just uses older drivers for some of the hardware.
What is special about Android that an OS image with a bunch of older drivers couldn't just install on an older phone? The install detects the IDs of the various modules - screen size, radio/wifi module, sound processor etc, and just uses the appropriate driver. It seems like that would be minimal work for either Google or the individual handset manufacturers, but I'm probably missing something.
Agreed, though I've had a bad impression of them since school, in the eras of having no physical floppy eject button, a single-button mouse, and being made to stay behind and do lines of 'macs are not crap' by, in hindsight, a pathologically fanboyish CDT teacher for saying (you guessed it) 'macs are crap'.
iTunes coming along and being horrifically bloated compared to Winamp made the impression worse, as did the iPod and needing a Winamp mod to be able to drag and drop MP3 files to it (my girlfriend's, not mine, of course)
Further shenanigans of multiple flavours have done nothing to endear them to me, though I'll never deny they make excellent hardware.
Agree completely. I've always turned this crap off in rental cars when it inevitably starts doing stupid stuff like wobbling the steering wheel because it doesn't understand junctions/sliproads.
Happy with my mostly-analog BMW that only has ABS, which I appreciate is good tech though only ever kicked in when I forced it to see how it worked. No touchscreens, no interference, proper handbrake, no other solutions for problems that don't exist like keyless entry etc.
Auto-brake is as far as I'd go as long as I could trust it to work even better than my own eyesight and reaction time - from reading comments about it, it seems like it's not there yet.
> Google is worth at least a couple billion in revenue for Apple
More than a couple - circa $20bn a year in recent years just for Google search as default. Another way of putting that is Google pays for almost all of Apple's R&D expenses. An incredible deal for Apple, so any concessions would be unsurprising.
> There's always the wonderful 3rd option that we are both non-deterministic and non-free because the illusion of free will arises from the action of some effectively random process along the lines of atomic decay.
That is what I believe is the case - non-deterministic yet also non-free.
Sam Harris (big no-free-will proponent, wrote a book on it) and Daniel Dennett (prominent philosopher/cognitive scientist) had a good debate on this, if you're interested - https://www.youtube.com/watch?v=_J_9DKIAn48
> So, it is for when you are at home but don't want the convenience of using a real controller
It is a real controller - it's almost identical to the DualSense, that is one of its selling points vs using a third-party controller wrapped around a phone screen etc.
It also doesn't need to be on the same network as the PS5, it works remotely too.
> Other than "continuing your game while on the toilet" I can't comprehend the use case for this
Easy;
- Playing PS5 while missus watches TV/plays PS4/Switch etc
- Playing PS5 casually with TV on in background
- Playing PS5 in other rooms
- Playing PS5 remotely i.e. in a hotel, work (I've not tested this yet, but read that given a good connection it works quite well)
So, yes it's limited in use in that it's not a standalone handheld console, but for those uses, it is a nice piece of kit that does the job it says well.
AMD has had the better CPUs for a few years now - they are plenty fast and dominate in power efficiency. Intel's will match or slightly beat the performance, but at the expense of needing a modular nuclear reactor to power them. Intel's 'e-cores' are not necessarily a bad idea, but they have obviously been shoved in to try and make up for the dismal efficiency of the p-cores.
The next gen will be interesting - we'll be able to see if Intel's Gelsinger has managed to right the ship on architecture.
> A human brain consumes around 20W, and even assuming the "training time" is every living moment from birth to a wise old 80 years, is a total consumption of just over 14MWh.
In a sense, the training time of a given brain is more like the history of all of its mammalian ancestors' brains, which via evolution has trained the brain matter to develop and use as yet unknown algorithms in a certain fashion to learn and reprogram itself. Which is....a lot of training.
> Now were at a time where even many governments don't have the required funds any more to make it happen.
I don't think that's the main issue as they can essentially just print the money. The main issue is we appear institutionally unable to build things at scale anymore. Some combination of lack of knowledge, structural government/council issues, corruption, legal issues/NIMBYism or simply the inability to take any long-term actions from a governing system that works in 5-year blocks, and it's crippling for something like nuclear power. It's bad enough for much more simple projects like train lines, or even houses - we cannot even get that right.
> The write up seems to suggest that this vulnerability affects just the vector registers, and I wondered how the vector registers were actually used.
> I'm no expert, but the way that I thought these units were used was mainly for mathematical operations, like array processing.
Me too. But, from the github page for the exploit - "The AVX registers are often used for high performance string processing by system libraries. This means that very high volumes of sensitive data pass through them."
> All the "motorheads" I know who say they love the speed and the acceleration and the wind in their hair, etc. won't touch electric cars or bikes. Despite the fact that they are the fastest moving things around. What they mean is "they want to make a nuisance and let everyone know they drive a flash, expensive car".
It doesn't mean that, that's a non-sequitur. I suspect you simply don't get the appeal. I'd bet the majority of sports car/bike owners are neither nuisances nor care for how expensive the vehicle is. See - popularity and reverence of classic cars like old M3s, R34 GT-R etc.
It's OK - most people see cars as simply A to B transport, don't enjoy driving, and don't appreciate any of the design or engineering of a performance vehicle. Which is why the roads are full of bloated crossovers on stilts with zero steering feel or liveliness to them. No problem, not everyone shares the same interests and hobbies.
> * my kids do not understand the difference between "WiFi" and "the internet
From what I've seen, this seems to apply to most people under 30. Initially I thought they were stupid, then realised they have likely never used an Ethernet cable - to not even speak of an RJ11 - in their life, and that actually I'm just old.
> Is it possible that, taking only biological factors and not social factors, men and women tend to like different things?
It's not just possible, it is absolutely the case, yet inevitably ignored in every single report/article of this type. Specifically, men are more likely to be interested in things, and women more likely to be interested in people.
Even without studies supporting it, this is just self-evident stuff for anyone talking to a bunch of men and women about their interests over the course of their life.
A relatively recent study on this issue - https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0261438 - found that not only do these differences exist, but that in countries with higher levels of female empowerment, the ratios in career choice are in fact *bigger*. It's a total repudation of the idea that if careers aren't 50/50, there's some bias or other negative effect at play.
Every time I've read a CEO's bleating about getting back into the office for productivity etc, there never seems to be any evidence to support it.
Example - number of tickets closed per day, cadence of successful releases, etc.
Surely, especially at the same time as an org being *data-driven* being so trendy, there should be...data?
Meanwhile, my personal productivity from not having the commute and associated faff is strongly supported by many KPIs. Number of parcels not missed, average percent full of washing basket, number of trips to refuel, kilometres ran per week, videogames completed per year....
> because Islam teaches that nationalism and other fake religions are alienation
Ah yes, the other fake religions.
It's remarkable that everyone who thinks this way is so lucky to have been born at the right time and place (in the last circa 200,000 years), to the right parents, that they grew up indoctrinated into the *true* religion - while everyone else has the misfortune of being born into one of innumerable fake religions, 'idolatry' or atheism nonsense.
It's really quite an astonishing stroke of luck, don't you think?
The "plateau" (area under the tail end of the bell-curve more accurately, it's not like we're programmed to self-destruct after X years) on lifespan won't be moving significantly until we fix a few very hard problems - brain degeneration (so dementia etc), cancer and heart/artery disease to name three. Risk of all of them go up with age.
Even then, 'healthspan' won't quite improve in the same way, as a 100 year old free of major disease is, well, still 100 years old and inevitably quite clapped out.
The real, root-cause fix will be to treat the *disease* known as 'aging'. So, looking at gene therapy. Research into super-centenarians has found some gene variants in common, and of course genes are also responsible for the extraordinary longevity/cancer resistance of some species e.g. Greenland shark, naked mole rat. We need to find the right genetic software to give all cells youthful vitality/immortality, but at the same time prevent uncontrolled division AKA cancer. Very difficult task, but no reason to believe impossible. Question is only will we be able to reach that point without destroying the planet in one form or other first? And if we make it that far, the ethical questions of who gets to have their firmware patched to human 2.0...
> My experience is that the better the 3D device, the worse the nausea is. What I think is going on there is that you are successfully fooling the bodies visual inputs, but these are being cross referenced against your bodies Vestibular system in the inner ear.
This is true. But some people, at least, can get used to it to a significant degree.
My first session with Playstation VR, major vestibular disruption and I tapped out after 30 mins or so.
A few sessions later, I was doing hours at a time with no problems at all.
To this day, pulling extreme maneuvres in a fighter jet in VR induces a wave of unpleasant sensations - but then so would doing it for real!
For those who can never find their 'VR legs'...yes, value very much limited to no-lateral-movement experiences.
of the lengths that Facebook et al go to monitor every minute detail of your interaction with their app to make sure you, dear product - oops - customer, don't have a single wasted second of potential 'engagement' or 'value'* kept from you. After all, merely using psychological tricks, sorry, methods, as part of UI design and behaviour to maximise 'engagement'** of dopamine systems might leave some spare 'engagement' on the table.
* definitely not compulsion, addiction, etc.
** definitely not abuse
> Just ban all children from social media.
This is the way. Social media is not only a gateway into all sorts of toxic content and an attention-destroying dopamine pumping device, but it's a global psychological experiment on homo sapiens that we didn't realise we were signing up for, with the consequences on society and our brains still unravelling.
But it can't be done by parents. Has to be baked into the apps - need verifiable ID to register and occasionally re-authenticate.
Won't happen though. We'll just get the token hand-waving and virtue signalling whenever something bad happens and the occasional unenforcable/impractical bill going through.
Sure, but you're missing the point. Labelling a population of millions of people, of whose motivations one cannot know in entirety and will certainly include many non-racist reasons, as racists (or in this case, racist scum), is an act of group-based discrimination in exactly the same manner as racism or other biases.
To amend your analogy, in the same way a Pole doesn't choose to be Polish, a non-racist Brexit voter *didn't choose that other Brexit voters may be racist*
I don't think it's a scam either, but I'm also yet to see a convincing argument for the use case of it vs a centralised relational database.
Your Starbucks coin example, current loyalty apps/accounts will store their users' points on a database, what is the problem with this that needs the solution of defi? Especially PoW vs PoS based which wastes massive amounts of energy vs a database.
> So, by implication, you think it's easier to provide 400 MB/s via 5G for the whole country than it is via fixed lines? If so I have a bridge you might be interested.
I didn't imply anything about how easy it would be. Though I'd assume that erecting a radio mast or two is significantly easier, cheaper and faster than digging up hundreds of streets, of course.
Cable is of course better, but not all of us have that luxury. Nor can we realistically do anything about it other than 'register interest' in Virgin Media while watching the years go by.
My point was merely that, in such areas, 5G would be quite useful indeed.
> Other than the bandwidth monitor, what Apps do you have on your 'phone that can use 400 Mb/s? Can anyone sling ads at you at that rate?
Nothing. My point is not that I need a 5G *phone* connection, but that a 5G mast in my town would enable my home devices to have much more bandwidth, for which I have multiple uses.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
