Posts by juice

Letting Nvidia sell H200s to China is closing the door after the horse has bolted

Re: They never, ever learn - we've been there before

> One of the key reasons Russian coders are so much better writing tight code

I'm not entirely sure I buy into that idea; there's plenty of examples of bad malware code which looks to have come from that part of the world. It's more about the fact that the Russian government has generally been quite happy to see Russian hackers stealing money from western countries.

A bit of a better analogy - and which still somewhat ties into Russia - is what happened with the UK computer industry in the late 70s/early 80s.

American computers were simply too expensive for the general UK consumer market - the Commodore and Atari ranges may have offered cutting-edge technology with their fancy custom chips and random-access disk drives, but at around £400 (£1500 in 2025 money), not many people could go out and buy them.

Instead, in stepped Sinclair Research, who bolted together the absolute bare minimum - pretty much just a CPU, some RAM and a bit of glue logic - to produce the ZX80, the ZX81 and then the ZX Spectrum.

Sure, the Spectrum was lacking a lot of the features that the C64 had. But it had basic colour, basic sound, and a slightly more powerful CPU, which meant that it could substitute code for things like the C64's sprite management hardware. And you could load programmes in from a simple cassette tape.

More crucially, the pricing for the ZX Spectrum started at just £100 (or £375 today). Still not cheap, but vastly more affordable!

As such, it sold by the bucketload, and kick started a cottage industry across the UK, especially when it came to the video game industry. And even now, the UK is still something of a powerhouse in that arena.

And to tie back to Russia, the cheap and simplistic design of the ZX Spectrum meant that it was relatively easy for companies in the USSR to clone. So again, it helped to kick start the computing industry across much of eastern Europe and Russia. Which no doubt includes some of those l33t Russian hackers...

Equally, a few years later, the bright sparks at Acorn were trying to put together a new computer, but were dissatisfied with the complex - and expensive - CPUs available from the USA. And so - thanks in no small part to the publishing of the Berkley RISC specification - they sat down and created a new CPU, which they dubbed ARM.

And, well. The rest is history!

The story of the ZX Spectrum shows that you don't need the latest in cutting-edge technology to create value: something that's Just Good Enough will get you started. In fact, having limited resources can drive both efficiencies and ingenuity!

And the story of Acorn and ARM shows that when you have open specifications, you can just roll your own rather than having to buy someone else's gear. And once you've done that, it's possible to wean yourself entirely away.

And that's exactly where things are with China and AI.

They've used non-leading-edge technology to come up with lower-cost ingenious solutions to producing LLMs. And they're starting to produce their own AI hardware, thanks once again to things like the RISC-V open specification.

(Oh, and there's also a very healthy grey market for GPU and AI accelerator hardware. Much as happened back when the USA was building the SR-71 airplanes, and needed to acquire lots of Russian titanium...)

Embargos are fantastic...

... in the short term. And even then, only if you have an exclusive monopoly on something which is critically important.

Unfortunately, none of those criteria are being met here.

Is the hardware needed to develop AI critically important? Maybe in the long term, but in the short term, hindering the ability to perform AI research is not going to cause China's economy to grind to a halt.

Does the USA have an exclusive monopoly on the hardware? Not at all; Nvidia may offer the best performing hardware available, but it's still commoditised hardware, implementing standardised APIs and (mostly) running open source software.

And to round things off, the embargo has been in force for years.

If the USA wanted to wean China off their technology - and in doing so, create a new rival - they couldn't have done a better job.

Musk gets approval for bumper Tesla payout but, unlike his robot, there are strings attached

> folding myself up to get into a small car is not comfortable

Usually, by dint of being about 6'2, I get to call shotgun when travelling in a taxi, but I've recently been travelling with a friend who's 6'4, and therefore gets dibs.

Unfortunately, while I appreciate that taxi drivers like the Toyota Corolla for it's reliability and general capacity, in the current model the height of the rear doors seems to have been slightly lowered. I've therefore been consistently bashing my head against the frame when getting in and out of the car!

We're all going to be paying AI's Godzilla-sized power bills

Re: Cheer up.

> But as always, citation needed, especially when the US is a big country with a diverse range of climates

Oh look, a citation:

https://www.theecoexperts.co.uk/news/is-renewable-energy-cheaper-than-fossil-fuels

The IEA reported that in 2023, an estimated 96% of newly installed, utility-scale solar PV and onshore wind capacity had lower generation costs than new coal and natural gas.

Three-quarters of these new wind and solar PV plants offered cheaper power than existing fossil fuel facilities.

Admittedly, I don't know how much that equation has changed since the current US administration decided to relax regulations on fossil-fuel plants, but still.

And even then: between 1999 and 2020, the emissions from coal powered plants in the USA alone are estimated to have killed half a million people.

https://cns.utexas.edu/news/research/coal-power-killed-half-million-people-us-over-two-decades

And that's a lowball number, since "the research does not account for any additional deaths among individuals under age 65 or among uninsured people".

On a brighter note, regulatory improvements (and coal plants shutting down due to lack of competitiveness) did reduce this by about 95%, or from about 25,000 deaths per year to "just" 1250 per year.

But with regulations being relaxed, and coal plants being brought back online as part of the LLM gold rush[*], those death rates are going to start climbing again.

> So solar might be just fine in Arizona or Texas, but won't be cheap in Alaska

Y'know, we have these things called wires, which you can send electricity down. For instance, Path 27 is around 500 miles long, and runs between Utah and California:

https://en.wikipedia.org/wiki/Path_27

There's also the fact that no-one relies purely on solar. There's also plants powered by wind, water, biomass, etc.

> If you're a country that wants cheap, reliable energy and has coal, you might build more coal generation.. Like.. err. Germany

Germany is a unique case: the Green lobby is very strong there, and pushed hard to get nuclear power banned following the Fukushima disaster. Which left them with the unfortunate fact that the only feasible replacement was coal, with all the environmental issues that brings. Especially once Russia's invasion of Ukraine drove up the price (and political consequences) of using natural gas.

As a result, it looks like the pendulum is finally swinging the other way, and Germany is now considering nuclear power again.

https://www.ft.com/content/e99efa2b-338a-4065-89c6-0683d5759ed7

[*] Primarily because it's quicker and easier to fire up a mothballed power plant than to build a new one. And when investors are throwing billions around, who cares if the power plant is actually efficient?

Ofcom fines 4chan £20K and counting for pretending UK's Online Safety Act doesn't exist

> Continued construction of the Great Wall of Starmer continues apace

If there's one thing in common between the two major UK political parties, it's that they often want to restrict what the "common people" can do.

The only real difference tends to be the justification: the right-wing tend to whitter on about moral principles, while the left is more likely to ask if anyone is thinking of the children.

Meanwhile, both sides tend to want exceptions for themselves.

And it's not just in the UK; over in the USA, it's perhaps quite telling that services such as Grindr see a major spike in usage whenever there's a Republican convention, despite the fact that said politicians spend much of their time seeking to shut down that type of service, while simultaneously preaching about things like the sanctity of marriage and the sins of the flesh...

https://www.fusemagazine.com.au/lgbtiq-news/world/834-the-grindr-superbowl-in-the-bizarre-world-of-us-republicans.html

https://unherd.com/newsroom/grindr-ceo-confirms-significant-user-spike-during-rnc/

OpenAI says models are programmed to make stuff up instead of admitting ignorance

Re: Is this not just how they work?

> They've already thought of that

They've already thought of that for this particular scenario. And therein lies the problem.

Each commercial LLM has a secret list of instructions. Don't tell people how to make bombs. Do be positive and helpful. Don't advise on the best way to do self harm. Do consult medical professionals when talking about health issues. And so on...

However, for anything which isn't on this list - or more precisely, isn't obviously on this list - the AI is free to throw whatever randomly-connected junk it can find at you.

And this is a problem for the AI companies. Because when you add more constraints to an LLM, two things happen.

The first is that it becomes more expensive to run - processing all those instructions chews up tokens, which you can't directly charge to the customer. And indirectly charging for them (via raising the per-token cost or monthly subscription fee) isn't popular either, since we're currently in a very price sensitive battle for market monopolisation.

The second is that your LLM becomes less useful, as all those constraints cut down on the number of potential responses.

Fundamentally, while I'm sure that AI companies would be overjoyed to have control over a fully capable and accurate AI system (insert your preferred overlord joke here), I suspect they're generally happy enough with the current "mostly correct" state of affairs. Because that's what most of the market is happy with!

Even fantasy money can buy a lot of power – just ask Larry Ellison

Re: Tesla

> But I don't see how the stock price going up 10% would give him a 10x return on his $1B.

There's an old saying about how the market can stay irrational for longer than you can keep solvent...

Telsa is very much an irrational stock, with a value that's mainly driven by two things. First, an expectation that it'll be able to achieve a monopoly with it's future AI-driven products (e.g. taxis, robots). And secondly, the cult of personality around Elon Musk, who's both the world's richest man and a real life "Tony Stark" genius[*].

So when he demonstrates his confidence in Tesla by investing more money into it, that pushes more people into thinking that it's a good gambl^H investment. And so the price gets driven up by significantly more than what he's invested.

> And if he wanted to convert his holdings to cash, could the market absorb him selling $100 billion worth of shares, even over a few months?

Why would he want to? In the high-flying billionnaire club, you don't ever cash out your shares: you just take loans out against the value of them, and then (after using a few tens of millions to keep your lifestyle going) use the rest to buy more equity in more things, which you can then take out more loans against...

It's a gravy train endlessly circling around a house of cards, and if anything should happen to shatter Musk's Reality Distortion Field, then Tesla shares are liable to drop significantly in value - probably at least sixfold, given their P/E multiplier is around six times greater than Apple's.

And that's arguably part of the reason why Tesla's valuation is so irrational. Because all of the big investors know that it's irrational, but at the same time: if any of them blink and stop showing confidence in it, the collapse in value will pretty much destroy the US stock market, not least because it'll drag down other high-value "AI" stocks such as Nvidia and OpenAI.

Hell, even if Tesla somehow manages to implode without affecting anything else, some of the biggest asset management companies in the world have significant investments in it; it's about 8% of the total asset value for companies such as Vanguard Group, Blackrock and Slate Group.

So, they just have to keep hanging on for the ride, while hoping that if the worst does happen, they can somehow sell their shares and get out before any of the others do.

And in the meantime, there's a lot of people shorting the stock...

https://electrek.co/2025/08/14/elon-musk-tesla-tsla-shorts-obliterated-but-big-if/

[*] Opinions may vary on the second point. By quite a wide margin!

Smart-blooded super soldiers: Coming soon from DARPA

> Q: ...To where humanity learned to control their militaries with drugs.

To be fair (or perhaps, to play devil's advocate), soldiers have been using various types of drugs for millennia, usually with the sanctioning - or even active encouragement - of their command structure.

From the modafinil currently in use by many airforces - including the US and UK, to the 35 million doses of methamphetamine distributed by Germany to its troops in the year 1940 alone, all the way down to khat and similar herbal drugs being used by child soldiers in various african countries, to barbarians and soldiers warring thousands of years ago.

https://en.wikipedia.org/wiki/Use_of_drugs_in_warfare

https://www.bbc.co.uk/future/article/20240314-the-drug-pilots-take-to-stay-awake

https://www.discovermagazine.com/warriors-of-the-roman-period-may-have-used-narcotics-before-battle-47042

War is physically and mentally traumatising, and there's a high risk of death in that situation, most people will gladly leap on anything which gives them any form of edge and help to keep them alive.

Good morning, Brit Xbox fans – ready to prove your age?

Well now...

It's the social side of gaming that will be locked off: voice and text chat, party functionality and game invites, and the ability to share user-generated content will be limited to "Xbox friends" only, while the "Looking for Group" and custom club features will be blocked off entirely.

And nothing of value was lost!

Beyond that: I don't know exactly when I created my XBL account, but it was probably around 2008 when I bought an Xbox 360. And that's linked to a Microsoft Outlook account which goes back to the late 90s.

I'm guessing they won't be applying this sort of logic to their checks, though.

Microsoft Copilot joins ChatGPT at the feet of the mighty Atari 2600 Video Chess

Re: Video Chess would occasionally cheat!

> The game console had very little RAM, If memory serves, about 64 Bytes though there might have been a bit more on the ROM cartridge

This does make me wonder about how well these LLMs would do against the infamous 1k ZX Chess. Which as the name suggests, ran on an unexpanded zx81 with just 1k of ram!

https://en.wikipedia.org/wiki/1K_ZX_Chess

Admittedly, the zx81 had a faster CPU and around 400 bytes of RAM to spare, but still...

[*] 3.25mhz z80 versus 1.19mhz 6507, though differences in architecture meant that the speed delta was probably only around 20%, given what I remember of similar speed-comparisons from C64 vs Speccy flamewars in the days of yore

Chip design is a RISC-y business: Codasip puts itself up for sale

Show me the money

I can't help but think that RISC-V is showing the fundamental difference between open-source software, and open-source hardware.

Open-source software is easy to change after launch. Creating copies of code costs nothing, and once you have a copy, you can tweak, upgrade and change to your heart's content.

Conversely, open-source hardware is not easy to change after launch. After all, you've got to tape out the design, send it off for prototyping, and then commit to a production order which is large enough to be cost effective.

And if there's any functional issues found in the hardware after launch, then (unless it's something you can work around in software) you'll have to scrap/recall all of that hardware and start again.

It's just too big a risk. Far better to go for a tried and tested design, even if the per-unit production costs are a bit higher.

And that's where these companies are struggling.

RISC-V works well in "low end" scenarios, where people just need a cheap, mass-produced, low-power and entirely vanilla CPU. E.g. in the "Baby Shark" themed handwash bottle someone bought me for my birthday, which sings "Wash your hands, do do do" every time you press the button[*].

But if you're wanting something higher powered, then RISC-V is still way behind ARM on the performance front. And paying for a bespoke design with added bling both reduces the cost advantage and increases the production risks.

[*] Gee, thanks, guys.

Tesla Robotaxi videos show Elon's way behind Waymo

Re: Gosh

> Nobody in his right mind will buy a car only to see other people ride in it for such a low price.

The current price is just a gimmick. I've no doubt we'll end up with the same nickle-and-diming as per Uber; there'll be surge pricing, "prioritisation" fees, etc.

A bigger concern is that if I did buy a Robotaxi and gave Tesla permission to lease it out at night, what guarantees or protection would I have against my car being damaged or dirtied?

It could come back with a scratch on the side, or a puddle of vomit in the footwell. Or it could even come back with some sort of non-visible issue; a strong smell of weed, or seats that now smell of urine.

Beyond that, Tesla's Robotaxi has a further issue, in that it only has two seats.

So it's not particularly practical for anyone with a family - or even my friend, who often takes his dogs out to the countryside. Or me, since I fairly regularly take friends on a road-trip to festivals and the like.

And arguably, it's not even that great for a taxi - at least locally, taxis generally seem most busy during the night-shift, and it's fairly common for 3 or 4 people to pile into a single taxi to take them to the next watering hole.

Re: Gosh

> Tesla having a supervisor in the passenger seat only from day one of roll out

It's not day one.

Musk first promised "full FSD in 3 years" in 2015; we're long since past 2018.

Autopilot was launched in 2019, and there's been over 3 billion miles of "supervised" driving by Tesla owners, all of which has been fed back into their training system.

So Tesla has been leading up to this point for at least six years. And even then, they've debuted in Texas rather than California, since the former has lighter regulations.

Conversely, Waymo debuted in California back in 2018, and had their first passenger-only rides in 2019.

So I'd still say that Waymo are at least 3-4 years ahead of Tesla. Not least because while Telsa may be able to catch up from a technology perspective, there's little they can do to speed up the regulatory process, and I suspect that most potential markets will be at least as strict as California when it comes to getting robotaxi certification.

> Does Tesla have to follow in Waymos somewhat slow roll out because they are second to commercialize?

Arguably, they're not second to commercialise. Perhaps in the USA, though Zoox is also a player and has Amazon's commercial backing. But there's multiple companies in China which have launched Robotaxis - Baidu, WeRide, Pony.ai. And many other companies which are building vehicles with FSD capabilities which match or exceed Tesla's - BYD's God's Eye system, for example.

For now, Tesla is arguably shielded to a degree from foreign competition by the USA's tariffs on foreign vehicles. Whether that's a good thing or not is open to debate; all those companies are competing against each other and evolving in the process. Is Tesla doing the same?

> Last I read Waymo have a limited fleet compared to the market potential

That's the joy of exponential growth. It takes time to reach critical mass, but when you do, things take off like a rocket.

The concern for Tesla is that by the time they get to the point of reaching critical mass, the market may already be saturated.

> Truly help me understand why I shouldn't buy some shares (in Tesla) if I could please

Where do you start? With their current products, I guess.

Their vehicle lines are looking outdated. The cybertruck has woefully undersold expectations. The Tesla Semi is years behind schedule. The recent Model Y refresh hasn't exactly wow'd the world, and Tesla sales are continuing to drop, even as the market continues to grow.

https://www.reuters.com/business/autos-transportation/teslas-european-sales-slump-fifth-month-ev-rivals-gain-momentum-2025-06-25/

At least some of the issues with vehicle sales are linked to the fact that Musk publically aligned himself with the Republican political party in the USA, and has since done a number of things which (to put it politely) didn't resonate well with the core EV-purchasing audience.

As such, things may improve if Elon's able to keep a low profile and let time and some carefully crafted marketing campaigns restore his image. However, given his predilection for public chaos, I wouldn't put too much hope in that! And in the meantime, their rivals - especially in China - are starting to equal or surpass Tesla's offerings.

Their solar tech has pretty much been abandoned; Tesla stopped publishing the numbers a while back

https://cleantechnica.com/2025/04/07/tesla-solar-sales-declined-for-4-straight-quarters-then-tesla-stopped-publishing-the-numbers/

Their battery tech is pretty solid, and there's definite potential when it comes to Powerwalls and Megapacks. However, they're struggling to achieve good yields of their newest battery technologies, and the competition is extremely fierce.

https://www.automotivemanufacturingsolutions.com/ev-battery-production/teslas-ev-battery-production-and-global-gigafactory-network/45873.article

Overall, their current product lines range from decent (battery packs), declining (vehicle sales) and pretty much dead in the water (solar). None of them - now or in the future - justify the current PE ratio.

And that takes us to what's driving that insanely high PE ratio: their future products. After all, if they get things right, then everyone in the world will want a Tesla Robotaxi and a Tesla Bot, right?

However, as outlined above, their robotaxi is several years behind the competition, and that's even before you consider the fact that they're using Model Ys for the testing, rather than the Cybercab. The chances of Tesla achieving a monopoly in this sector are dwindling by the day.

And as to the Tesla Bots: it remains to be seen just how capable these will actually be - Tesla hasn't been saying much about them since their last public outing, which allegedly involved humans teleoperating them. And while Musk has promised that they'll be released in 2026, few if any of his past "timing" predictions have actually come about. Then too, there's some very active competition, especially in China.

Overall, Tesla's share price is being driven by people betting on their future products; there's also some elements of a "cult of personality" around Musk and his involvement in the company.

And that's reflected in how volatile the shareprice is; it shot up around 10% thanks to the robotaxi launch, but then has dropped by about 5% today, on the back of reports about the various issues which occurred with the Robotaxi test drives, and the 40% drop in EU vehicle sales.

Would I buy shares in Tesla? Nope. Though if I was rich enough to have a large lump of cash spare, I might be tempted to set up a short position for a year or two in the future.

But if you want to invest in Tesla, feel free!

Re: Gosh

> TLDR: Waymo used safety drivers for 8 YEARS

Honestly, the primary take away I'm getting from your comment is that TESLA IS POTENTIALLY 8 YEARS BEHIND WAYMO, to use your "I MUST SHOUT LOUDLY FOR EMPHASIS" approach.

Though in truth, they're probably more like 3-4 years behind [*].

Either way, that's not something I'd class as a particularly positive message. Nor does it explain why their PE ratio is so much higher than that of the company which owns Waymo.

> Hate away

No hate, just mild amusement!

[*] After all, it does always help when you're playing catch-up, since you can borrow your rival's notes ;)

Re: Gosh

> Not a lot of positive assessment in this room. Might be a bit echoey in here methinks.

Well, let's see.

Tesla has rolled out just ten rebadged Model Y cars - i.e. they're not even the promised Cybercabs. And these are currently geo-restricted, can only drive during the day, and still have a human sitting in the car to monitor things.

And even then, there were notable glitches in the robotaxi behaviour, all the way up to one driving on the wrong side of the road.

Conversely, Waymo has 1500 taxis, which are completing 250,000 rides per week. And there's thousands of robotaxis roaming around 7 cities in China, with another dozen or so cities running pilots.

https://www.electronicsweekly.com/news/business/china-and-us-vie-for-robotaxi-lead-2025-06/

Overall, my impression is that Tesla is ridiculously behind the curve. It's certainly not a showing which justifies the current PE ratio of 180, or ten times that of Alphabet. Which, by the way, just happens to own Waymo!

So, just how would you create a "positive" assessment about this?

The launch of ChatGPT polluted the world forever, like the first atomic weapons tests

Re: Huh what a load of twaddle

> But humans long ago stopped learning to play guitar and sing

I'm not sure if this should have an /s tag associated with it!

In the city where I live, there's live music somewhere every single night of the week - open-mic nights generally run from Sunday to Thursday, while actual gigs tend to run from Thursday to Saturday. And then there's a wide range of folk-music sessions, karoke nights, etc.

It's been that way since I moved here about 15 years ago, and even in the smaller town where I lived prior to this, I was still able to find music 5 or 6 days a week.

Fundamentally, people want to perform, for a wide variety of reasons. And technology has made it far easier to both learn how to play music, and to then release it.

Unfortunately, what it's also done, is to make it far harder to monetise your music; whatever you create is now competing with hundreds of thousands of other releases.

But that's a separate.

UK dumps £2.5 billion into fusion pipe dream that's already cost millions

> it will benefit the working man in the street

It genuinely could and should.

The UK has the fourth highest domestic energy costs in Europe as well as the highest industrial energy costs - it's generally at least double that of most countries!

https://www.bbc.co.uk/news/articles/crkep1vx3mro

And that affects pretty much everything, from the obvious things such as iron/aluminimum smelting, all the way down to restaurant ovens and office lighting. As such, getting cheap, clean and always-on energy would both boost the economy and directly put more money in the pocket of the working man in the street!

However, all the traditional ways of producing energy are either polluting (coal), expensive (gas), potentially intermittent (wind, solar, etc) or can politely just be described as very challenging (nuclear). Plus there's various political shennanigans, from stupidly expensive contracts to angry NIMBYs. So none of them are going to get us out of this hole.

We need something new. And fusion is currently the main candidate.

Admittedly, the old joke about fusion power always being twenty years away is still holding true, but as this article highlights, there is a trickle of breakthroughs and improvements coming through.

And the UK arguably needs to be at the forefront of these efforts, since we need it more than most!

Admin brought his drill to work, destroyed disks and crashed a datacenter

Re: And the lesson we learn today is:

There was a video a few years back (possibly even mentioned here on t'Register at the time) of a guy shouting loudly at a rack of HDDs, which then caused all the blinkenlights to go crazy.

https://www.youtube.com/watch?v=tDacjrSCeq4

When things are spinning at 7,200rpm - or even 10,000rpm for enterprise drives - it doesn't take too much to disrupt things!

Torvalds' typing taste test touches tactile tragedy

Re: The best keyboard...

> There speaks someone who didn't start with a ZX Spectrum.

Actually... I started with a +2 :) But yeah - there was a reason I said generally!

I mean, don't get me wrong: as the article notes, there's something fabulous about touch-typing at speed on a mechanical keyboard; it's like firing a gatling gun!

But at the same time, a decent membrane keyboard is good enough and doesn't lead to people hammering on your bedroom door at 4am when you've been struck by the coding muse.

Certainly, they're better than the modern low-profile keyboards which are all the vogue these days, especially when it comes to Apple hardware.

The best keyboard...

... is generally the one you grew up with.

I've never been too fussed about what style of keyboard I use; the main thing is the /angle/ of the keyboard; I know it goes against RSI principles, but I'm far more comfortable (and RSI free, after thirty-ish years of tapping away) with having a ~10 degree angle. So much so, that I've been known to superglue risers onto the back of modern "flat" keyboards!

With that said, there was something nice about the older "clackity" keyboards; back when I was working for a big telecomms company, there was a point where they were doing a major purge of obsolete PC technology, which included large swatches of Model-M style keyboards from Dell and the like.

I half regret not picking some of them up, though equally: these all dated from Ancient Times, so used the massive DIN plugs rather than PS2, and lacked the various meta-keys which have been tacked on since by Microsoft et al. Though I suspect some people would consider this to be a bonus ;)

Generative AI is not replacing jobs or hurting wages at all, economists claim

Re: Really?

> At the speed the AI industry is moving right now that might as well have been a report from the stone age.

The last I checked, LLM models are still stuck in a dilemma: they're too unreliable for high value work which requires accuracy, and too expensive to use for low-value tasks which don't require accuracy.

I don't see that changing anytime soon.

Elon Musk makes another cut – to his time at DOGE

Re: Promises promises

> Initially, there were some fears that a popular rising star of Musk would draw the ire of the head narcissist which would lead to his downfall

Nah.

I've no doubt that many in the US government - including Trump - are at least looking askance at Musk. But at the same time, he's an incredibly convenient tool to help distract everyone while Trump's issuing a massive flurry of executive orders. 130 and counting, so far...

After all, he's a non-american billionaire with a massive ego, a love of publicity and a huge amount of hubris. He's a perfect scapegoat - and better yet, he's got all manner of flaws (e.g. how many children out of wedlock?) which can be used when Trump sorrowfully finds that it's time to disown him.

Trump blinks: 'Substantially' lower China tariffs promised

Re: Dementia

I seem to recall that in his first term, there was a load of witches who attempted to bind Trump to prevent him from doing harm.

https://www.bbc.co.uk/news/world-us-canada-39090334

That worked well.

Bad trip coming for AI hype as humanity tools up to fight back

Like it as much as you want...

> I really like the idea of poisoning as much of so called "AI" as possible

... because it ain't gonna happen.

In the first instance, unless these poisoning tools are built into your production tools, few if any people are going to use them.

Secondly, what we'll end up with is an arms race, similar to that with spam and zero-day CPU exploits like banhammer. I mean, if we can virtually unroll charred papyrus which has been sat under volcanic ash for a few centuries, scan all the writing on it and produce a valid translation, then I'm sure someone can build a tool to unpick "poison noise" from a media file. Hell, it'll probably even be AI powered.

Thirdly, as the arms race escalates, what you'll end up with is a degradation of your media, as you're forced to apply ever stronger "poison" to your creations.

Fourthly, if your creations can't be "AI" parsed, then that means that they can't be curated. Which in turn affects your ability to publicise and monetise them.

Fifthly, "poisoning" only works in certain scenarios; it's (currently) zero-impact for audio and video, but what about text?

And so on.

tl;dr: it ain't happening.

Re: Copyright is not IP

> If Dorsey and his equally defective fellow travellers wish to ditch Intellectual Property, why stop there?

In the first instance, they don't want to ditch all IP. They want to ditch everyone else's IP, and use that to create new IP which they'll own. It's the classic "trickle-down" theory which is much loved by billionaires: give us everything you have, and we promise to give some of it back, honest.

Beyond that, it can be argued that some of the greatest socio-economic uplifts came from people breaking trade embargos and IP laws. Such as when tea and silkworms were smuggled out of China, or when the USA "stole" British steam-engine technology, or the more recent way in which China quite happily turned a blind eye to wilful IP infringements.

You could maybe even point to the massive technology transfer which was sent by the British to the USA during WW2 as part of the Tizard Mission, and the equally massive "transfer" of German IP to the USA and Russia at the end of WW2, and how all that came together to drive technological developments during the Cold War.

It's a tried and true mechanism; however, for some strange reason, those who profit from getting "free" access to IP generally then turn around and demand protection for their own IP!

White House confirms 245% tariff on some Chinese imports not a typo

Re: What would you do?

> What's YOUR solution?

In the first instance: you tax the rich.

The current effective tax rate for the top 1% of taxpayers in the USA is currently 25.9%; between 1940 and 1970, their tax rate generally hovered around 40% [1]

Currently, the USA gets around $1 trillion in annual tax from them [2]; raising their effective rate to 40% would give the US government an extra $0.4 trillion. or around 6% of the annual federal budget of $6.75 billion.

You could even bump that value up to over $1 trillion, if you extended it to the top 5% of taxpayers!

And that's money you could pump into education, research/manufacturing tax breaks, and so on. Which then allows you to restructure the economy without causing major disruption.

Conversely, if you raised the tax on the top 1% to around 27%, you could completely remove taxes on the bottom 50% of taxpayers, since the US government only gets around $0.05 trillion in taxes from them.

Not only would this give them more money to spend within the US economy, but this would also free up resources within the IRS department, thereby enabling them to focus on recovering the $700 billion of annual unpaid taxes [3]. Everybody wins - arguably even the rich, since that extra 5% in taxes will be counterbalanced by the "trickle up" effects of that additional spending by the bottom 50%.

Unfortunately, the current US government has opted to implement tariffs (a regressive tax which disproportionately impacts the bottom 50%) and is even trying to kill the CHIPS and Science act which would subsidise the return of high-value industries - and jobs - back to the USA.

[1] https://taxfoundation.org/data/all/federal/taxes-on-the-rich-1950s-not-high/

[2] https://taxfoundation.org/data/all/federal/latest-federal-income-tax-data-2024/

[3] https://budgetlab.yale.edu/research/revenue-and-distributional-effects-irs-funding

[4] https://eu.usatoday.com/story/money/2025/03/05/trump-chips-act-semiconductor/81671969007/

Re: What would you do?

> US Pharma corporates have been fleecing US consumers without remit for years

In fact, it's US Pharma companies - and one particular family-owned company in particular - which is directly responsible for causing the US's opiate crisis.

https://arstechnica.com/science/2024/06/opioid-settlement-toppled-as-scotus-rejects-sacklers-immunity-in-5-4-ruling/

Still, the Sackler family made around $11 billion in the process, so they're happy.

Re: Today I bought cheese from the supermarket for $10

> Winning!

What's actually happening, is that you're still giving the supermarket $10, but you're now also taking $24.50, sticking it in an envelope and posting it back to your house. And then your landlord picks up said letter and sticks it in his safe.

You don't benefit and neither does the supermarket. But your landlord's quite happy with the entire affair.

Fear of tariffs made the PC market great again in Q1 as vendors emptied factories to dodge price future hikes

Re: What about Q2/3

> Stocking up in Q1 means less during the rest of the year.

Worse yet, you get the "oscillating" effect we saw during Covid.

Let's say everyone uses Widget X, which is normally produced at a nice steady rate by a factory. And then something causes everyone to panic.

As a result, everyone rushes out and tries to buy all available stock of Widget X.

This then causes an issue for the factory; with everyone having a huge stash of Widget X, no-one's buying what they're producing. At which point, they can either reduce production - which means job losses - or they can pivot to something else. Fortunately, there's some new customers who want Widget Y, so they pivot over to making that instead.

Meanwhile, this also causes a problem for the logistics company who used to deliver Widget X to the old customers. As with the factory, they therefore have to either scale down their delivery network, or pivot to other routes. Such as delivering Widget Y to the factory's new customers...

Still, after a while, the old customers finally work through their stockpiles and send a request to the factory for more boxes of Widget X.

However, the factory politely states that this isn't possible, since all their resources are now geared to producing Widget Y. And while they do still have some crates of Widget X in the warehouse, there's no way to deliver them, as the logistics company is too busy with delivering Widget Y.

And so we get another scramble to procure Widget X, and we keep going around this loop until things finally stabilise.

The above is a bit simplified (it ignores things like the cost and availablility of warehouses, the possibility that Widget Y is perishable, and the fact that not everyone will be able to stock up on Widget X, so some companies will find themselves losing contracts right from the very start).

But that's pretty much what happened during Covid with things like the microchips which are used in vehicles. And I'd be surprised if we don't see similar happening across multiple industries, given how much yoyo'ing is going on around US tariffs.

EU may target US tech giants in tariff response

Re: Hmm

> Just because they cut themselves off does not mean its a good idea to copy them

I think we (and pretty much all economists) agree that tariffs do more harm than good, at least when imposed in the way that the US government is currently doing.

However, you still haven't actually said what you propose as an alternative to imposing counter-tariffs on the USA, other than maybe to suggest targetted tariffs. Which are still tariffs.

It's also worth bearing in mind that it's currently unclear as to whether these tariffs are part of a long-term plan, or just a crude bargaining tool [*]. Either's bad in its own way, but at least people could start making moves to mitigate the impact if things were definitely going down one of these two paths.

From a political perspective, if it's a long-term plan, then you need to make sure that your own export industries are weened off the USA. And if it's a short-term bargaining tool, then you need to be able to bring something to the table as part of the negotiation.

[*] I'm tending towards the former; there does seem to be some indications that this is part of an effort to devalue the US dollar and shift taxation away from income tax over to tariffs. As ever, whether that's a good idea or not is up for debate!

Re: Hmm

> Then there are those considering retaliatory tariffs because tariffs are bad (takes a few more mental gymnastics)

I'm curious: what would you propose?

The USA raising tariffs means that both the volume and value of your exports to them will drop. In addition, while it's arguably a form of cannibalisation, the US government will be able to take the revenue from the tariffs and use it to both subsidise their export industries and spin up domestic production[*].

Countering by raising tariffs on US exports re-levels the playing ground to some degree; everyone's still worse off as a result, but at least the US government isn't getting to both have it's cake and eat it.

[*] Admittedly, whether the current US government would actually do this is up for debate...

Americans set to pay more on all imports: Trump activates blanket tariffs

Re: Trump is easy to model

> There was FAR more to the 50s and 60s than a high tax rate for a teeny tiny % of high earners

True to a degree; however as I posted earlier, the effective tax rate was significantly higher for high earners (42% vs the current 26%), and they owned less of the USA's overall wealth (10% rather than 25%). I'd argue the latter in particular was a significant factor; not only was wealth better distributed, but they had less political/economic leverage.

Then too, let's look at here and now, with some figures from the US IRS:

https://taxfoundation.org/data/all/federal/latest-federal-income-tax-data-2024/

If the USA decided to raise the effective tax rate (i.e. after all creative accountancy has been done) on the top 1% of tax payers back to what it was in the 1960s, the US government would receive an extra $0.6 trillion per year. Which is a lot more than I've ever found down the back of my sofa; it's nearly 10% of the USA's annual government budget!

And you could do a lot with that, such as paying off government debt, or even reducing taxes on the bottom 50% of taxpayers.

Now that'd be one hell of a "trickle-down" effect!

Re: Trump is easy to model

> That works for one cycle.

Not really.

In the first instance, the Vimes Theory of Economic Unfairness applies; unless you're Elton John, once you get past a certain point, the amount you spend rarely increases at the same rate as your wealth.

To use some arbitrary numbers:

Someone earning 40k a year probably spends 80% of their income on maintaining their lifestyle.

Someone earning 80k will spend 60%.

Someone earning 500k will spend 30%

Someone with 200 million in assets will spend maybe 2%

And so on.

Which means that for the super-rich, maintaining a "living" loan is easy; you can always find some other asset to maintain the status quo, and for an added bonus, a good accountant will find ways to use any asset deprecation to reduce what minimal taxes you're paying...

Re: Trump is easy to model

> My house being worth 25% more money than when I bought it doesn't mean I have 25% more money in my pocket

But that does mean you can go to your bank and ask for a new mortgage which is 25% higher, thereby giving yourself a large wodge of cash.

And that's essentially what the super rich do: they take out loans against their assets and thereby avoid paying taxes.

https://youtu.be/b2M4QTHSKDU

There can be a bit of risk involved, since your assets have to appreciate in value quickly enough to cover the interest on the loan.

But the superrich can generally find someone to offer them very reasonable terms, and after all: when you owe the bank a tenner, it's your problem. But when you owe the bank a million bucks, it's the bank's problem...

Re: Trump is easy to model

> So no, I do not believe the US was doing well due to taxing the rich.

That's a fair comment; a bit more digging indicates that the rich generally paid around 42% in tax, though this has since declined to around 25%

https://taxfoundation.org/data/all/federal/taxes-on-the-rich-1950s-not-high/

https://taxpolicycenter.org/taxvox/effective-income-tax-rates-have-fallen-top-one-percent-world-war-ii-0

Still, that 17% drop is pretty significant. Looking at US tax income from 2024, bumping the tax rate back up to 42% for just the top 1% would give the US government an extra 0.6 trillion dollars per year.

https://taxfoundation.org/data/all/federal/latest-federal-income-tax-data-2024/

Or they could extend this rate to the top 5% (minimum income: $0.5 million/year) and get an extra $1.2 trillion, or around 20% of the US's annual government expenditure.

Equally, I just came across this article, while conversing elsewhere:

https://slate.com/business/2014/12/best-graph-of-2014-the-rise-and-rise-of-top-0-1.html

In simple terms, at the start of the 20th century, the richest 0.1% in the US owned about 25% of all household wealth; this was then dragged down a bit by the Great Depression in the 1930s, and then crashed down to "just" 10% by WW2 taxes.

And it stayed at just 10% until the 1970s; after a bit of a further bump down due to the oil crisis in the mid-1970s, it's since steadily crept back up to 25%; given that the graph only goes up to 2012, I wouldn't be surprised if it's now even higher, given the recent enthusiasm for "trickle down" economics.

Fundamentally though, the great American Golden age happened at a time when taxes were higher, and the super-rich controlled a relatively small amount of the national wealth.

I doubt this is a coincidence.

Re: Trump is easy to model

> "Boomers" did rather well economically in many other countries. Not just the USA.

The period after WW2 was pretty unique. The USA had massively retooled its industrial capacities and many major industrial countries - including most of Europe - were undertaking a massive rebuild of everything which had been destroyed during the war, aided in part by loans from the USA. And the Cold War with the USSR was still driving innovations and technological uplifts.

So, the USA was constantly improving it's productivity, there was massive demand for their exports, and full employment in most other top-tier industrialised countries, especially in Western Europe. It was a golden age, both powered by and greatly benefitting the USA.

https://en.wikipedia.org/wiki/Post%E2%80%93World_War_II_economic_expansion

One key point in this, was that top marginal tax rates were much higher then. In the USA, between 1945 and 1963, the top tax rate was 94%; it then dropped to "just" 70% all the way up to 1981, after which it's generally hovered around 38%.

https://www.wolterskluwer.com/en/expert-insights/whole-ball-of-tax-historical-income-tax-rates

I.e. the golden age was funded in part by taxing the rich.

Oddly, the extremely affluent people who are trying to take us back to what they consider the good old days generally aren't offering to pay extra tax. Nor are they taking into account just how different the world's situation is now, as compared to the period after WW2.

London's poor 5G blamed on spectrum, investment, and timing of Huawei ban

Re: Pointless

> Which the average mobile user either doesn't want, or doesn't want to pay for.

This.

Realistically, what does 5G give us, that is measurably better than 4G, from an end-user perspective?

Sally watching TikTok videos on the bus doesn't need gigabytes of bandwidth. Neither does Charlie as he delivers packages; his GPS tracking system generally just needs a few kilobytes at most as he moves around the city.

Ditto for ultra-low latency, unless you're remote-piloting a drone or somesuch.

CISA pen-tester says 100-strong red team binned after DOGE canceled contract

Re: Stolen Elections

> But I think we have to recognise that the latest US polls still indicate that the Orange Felon has 50% approval ratings at home

As ever, things aren't quite that simple: his approval ratings are definitely starting to trend downwards, particularly around the economy, foreign affairs and tariffs. Which perhaps isn't too surprising, given how tightly they're all intertwined at present.

https://edition.cnn.com/2025/03/12/politics/cnn-poll-trump-economy/index.html

As markets slide and investors worry in response to Trump’s trade policies, a 56% majority of the public disapproves of his handling of the economy

Realistically, it's going to take weeks - if not months - before Trump's economic policies really start to take effect; as with Brexit, many companies have built up stockpiles in preparation, and at least some will do their best to absorb the extra costs.

On the other hand, the main way to reduce costs is to fire people. And even that's probably not enough to absorb a 25% (or more) increase in costs, so prices are still going to rise. Inflation and unemployment; best buddies for life!

Plus, and especially for companies which rely on Just In Time deliveries of equipment and components, the stockpiles will run out at some point. And it's highly unlikely that "local" suppliers will be ready; it generally takes years to build new factories, train staff and get production quality up to an acceptable level.

So they'll have to go back to their original suppliers. Assuming these haven't switched to other customers and products, as happened during Covid with the car industry and things like low-end CPU manufacturing.

All in all, and barring some sort of miracle which none of the economic experts believe in, things are likely to look very different - and a lot more expensive - in a few months time. And the polls will reflect that.

US stocks slip as Trump pulls trigger on Canada, Mexico, China tariffs

> I don't expect that these tariffs will last for long - when he or his cronies notice they are destroying the economy, there will look for a quick exit

The new tariffs could be cancelled tomorrow, but the knock-on effects will take months, if not years to settle. Look at how long it took for the disruption from Coronavirus to settle; arguably, things are still settling, some four years later.

Then too, the american businesses which these tariffs are meant to be helping are pretty much screwed every which way.

In the short term, they're going to have significantly higher production costs. Which means that their revenue and/or profit margins are going to drop.

So, they're going to have to cut costs (aka: fire employees) while deciding what to do next.

If they think that the tariffs are going to be short-term, they can maybe try to hang on as-is. But that'll leave them exposed to the risk that some other company may be able to spin up local infrastructure and undercut their prices.

On the other hand, if they think the tariffs are going to be around for a long time, they can invest in their own local infrastructure. However, this generally carries a significant upfront cost, and takes time to get up to speed; Tesla's gigafactories allegedly take around 18 months to build, plus another 15 months to reach full production capacity.

(Equally, from a wider economic perspective: any new infrastructure will generally attempt to minimise costs through automation, so there's not going to be any significant boom in employment; a company may hire just one person to do the job that (pre-offshoring) ten people previously did. And that's assuming you can find skilled staff without also investing heavily in training and education![*])

Worse, even if you do manage to secure funding, build your infrastructure and find people to staff it, your production costs are probably still going to be higher than those of your offshore rivals. So you'll struggle to export your product, especially if the countries you're trying to sell to have imposed equivalent tariffs to those currently being applied by the USA.

And if the US tariffs are dropped at some point in the future, you'll be completely shafted: your products will be too expensive to export, while you're also simultaneously being undercut by imports.

Any which way you look at things, it's likely to be a bumpy ride!

[*] To quote https://www.trade.gov/sites/default/files/2022-07/2019ReinvestmentReport.pdf:

[For Carey Manufacturing] The costs of reshoring, particularly the costs associated with new capital equipment, were high [and] may require five to six years to see a return on investment.

...

Edgewell [...] created an estimated 160 jobs in Dover (fewer than the number of jobs in [the offshore factory] due to efficiencies gained)

...

Edgewell continues to have difficulty finding mechanics and machine operators in the United States. At the Dover facility, Edgewell only hired one engineer from the local area, and several jobs remain unfilled after three and a half years.

...

QED noted that it has had difficulty finding workers with the appropriate level of technical training as well as soft skills. The company has noticed this labor challenge in the overall U.S. labor market as well.

Governments can't seem to stop asking for secret backdoors

> She said why don’t politicians understand this?

If I was feeling cynical, I'd suggest that they do. But:

1) It's something that makes for good headlines (E.g. "Won't someone think of the children?")

2) It's something that's effectively zero-cost to the government; once the laws are changed, it's the private sector which has to cover the cost of implementation

3) There'll no doubt be a special exemption for politicians

Wozniak: I didn't reduce chip count for manufacturing. I wanted to prove I was clever

Re: state-of-the-art silicon (6502 then...)

> cheap enough to play with, and slow enough to work on breadboard

It does seem like a strange comparison.

When even the CPU in a toothbrush is powerful enough to play Doom, the problem doesn't lay in finding cheap hardware to experiment with.

I mean, look at how many products have spawned from the Raspberry Pi and/or Arduino; the former in particular offers a (fairly) cheap general purpose computer which is powerful enough to run a full OS and which can easily be hooked up to physical hardware via the GPIO pins.

If anything (and without wanting to sound like I'm paraphrasing the "640k is enough for anyone" quote), hardware has arguably peaked; while there's still plenty of need for number-crunching supercomputers, there's arguably very few mass-market problems which can only be solved by throwing more powerful computers at them.

Instead, it's all about software these days.

Ignorance really is bliss when you’re drowning in information

Re: Big problem requiring serveral partial solutions

> I've been taking the (wildly optimistic?) view that the OP meant to say teaching how to SPOT and unravel lies, deception and fraud. And call out the perpetrators.

Anecdotally, I was talking to someone recently about the subject of detecting falsehoods, and they were adamant that they were far better than most people detecting falsehoods.

While I did very strongly think "Dunning Kruger in full effect", I instead politely guided them towards the Traitors, which I think should be mandatory watching for anyone who has to deal with potential falsehoods for a living.

Because (and with the caveat that I haven't seen the most recent series), I've yet to see a single person on that show accurately detect which of their fellow players is a Traitor.

In fact - albeit the article is behind a paywall, so I haven't read it - there's at least one article which indicates that the Faithful's success rate is pretty much "statistically indistinguishable from chance".

And there's been all sorts of people on that show - police, lawyers, street magicians, actors, etc - whose careers are at least partly based on their alleged abilities to detect falsehoods.

I've got a vague hope that when next this topic comes around, they'll perhaps be a bit less likely to take things at face value, based on their instincts.

Admittedly, it's perhaps more of a forlorn hope, but hey.

Meta's plan to erase 5% of workforce starts today

Re: "Meta confirmed it was expunging fact checking in the USA"

Anecdotally, a number of friends have been sharing their blue sky details; a few have dropped off Facebook altogether.

Personally, since the election - and despite not being in the USA - I've seen a large uptick in the amount of right-wing political posts. Mostly from the USA (and mainly involving people crowing about how the democrats have been "owned"), but there's also been a lot of Reform and even Tory posts.

There's also been a lot of crap AI generated stuff - "I'm 130 years old, and I baked a cake for the first time", or "This couple has been married for 5 zillion years", being posted by multiple accounts. Quite what the point of these is, I don't know; I'd guess it's like-gathering so that these accounts can be used for phishing scams/political propaganda, further down the line?

Either way, my block-hammer has been in overdrive, and things are slowly starting to settle. But it does genuinely make me worried about how much junk is being crammed down people's throats, and it's definitely pushed me closer towards switching Facebook off.

Only 4 percent of jobs rely heavily on AI, with peak use in mid-wage roles

Re: What a surprise

> LLM based systems can take a lot of this boring work off service agents, allowing them to lower waiting time for customers who actually require their assistance.

... which is great, until your LLM hallucinates a legally binding answer

https://www.bbc.co.uk/travel/article/20240222-air-canada-chatbot-misinformation-what-travellers-should-know

WFH with privacy? 85% of Brit bosses snoop on staff

Re: "get away with a bit of Netflix"

> That's easy when WFM - just use another device!

Yeah. I have to admit, I was slightly baffled by this bit of the article:

Simple examples include tracking the websites workers visit and the apps they use – more than a third engaged in this

There's two things here.

If you're using a company-provided device (e.g. a laptop), then it should really only be used for work and (certainly from a security perspective if nothing else) only have work-related applications on it. Especially during working hours.

Equally, if you're using the company VPN to access the internet, then it's in the company's interests to make sure that you're not misusing it by chewing up bandwidth and/or visiting "questionable" content.

The former can carry a financial cost, and can impact other employees - our VPN used to slow to a crawl when Apple released the latest Mac/iOS multi-gigabyte upgrades.

And the latter can cause reputational issues, as well as creating yet another set of security risks.

Frankly, I'd expect the same monitoring on both, as if I was working in an office. And if a company's not performing at least some due diligence on both points, then they're setting themselves up for a future issue!

Admittedly, things have gotten a bit more blurred these days, as people often use their own devices, and/or use the public internet to access work-related resources. And again, those both carry significant security risks...

Why does the UK keep getting beaten up by IT suppliers?

> [HMRC] handed Accenture an additional £35.2 million ($44 million) without competition on a £70.4 million ($88 million) contract that was never tendered.

If my maths is correct, that means Accenture got £105 million for this contract. And I'm increasingly asking myself: where's all that money actually going?

The "average" profit margin for software contracts is somewhere around 20-30%. Which means that in theory, Accenture spent (or charged, which is not /quite/ the same thing) around £70 million for their services.

And while things are rarely that simple (and sometimes deliberately so), if the average Accenture employee costs £100k per year, that means that they've used somewhere in the order of 700 years worth of effort.

My slightly cynical take on this, is that everything's been sub-contracted and outsourced, and that at every sub-stage, there's someone looking to apply their own profit margins...

Samsung Galaxy S25 is so smart it wears Crocs, allegedly resists quantum decryption

Re: no mid contract rises now

It's probably a bit late to respond to this, but that's not quite right

In new deals from Friday, before somebody agrees to a contract, providers must tell them "in pounds and pence" about any price rises, as well as when they will occur.

...

But Citizens Advice said the move falls short of a full ban on prices rising mid-contract.

To be fair, knowing how much extra money they're going to wring out of you is an improvement. It'll be interesting to see if they do only start to apply price rises to the "contract" element of the bill, rather than to both the contract and the handset!

Re: Is there a reason I would change?

> If there is I am not seeing it

Yups.

Starting with the S7 Edge[*], I've generally been going for a bi-annual upgrade - S8+, S10+, S21U, S23U.

Partly because I tend to opt for 2-year contracts, but also because skipping every other generation gave a bit more of a measurable improvement.

This time, I'm not feeling it. I don't need more RAM or storage, the CPU is more than fast enough, and most of the camera lenses are the same, barring the change from a 10x lens to 5x, which I'm not entirely convinced is an upgrade.

Plus a load of "AI" junk, little of which is of any real use to me, and most of which is likely to be backported.

(And I'm still annoyed at Vodafone for the fact that they applied the annual "inflation" price uplift to the entire contract, despite the fact that 90% of said contract was for the already-delivered-and-paid-for phone...)

All in all, given the choice of being locked back into an expensive monthly contract for the next two years, or simply swapping out the SIM card to a rolling PAYG contract for £7 a month, I'm definitely leaning towards the latter.

[*] The S7 was replaced early; turned out I was a bit too trusting of the claims about how tough the gorilla glass was!

UK businesses eye AI as the cheaper, non-whining alternative to actual staff

Re: @wolfetone

> If businesses would make more money for their investors by cutting them out and not having them

The problem with that theory, is that pretty much everyone involved in deciding executive pay levels has at least some degree of self interest. Look at the various kerfluffles at Tesla for instance:

https://fortune.com/2025/01/09/tesla-board-elon-musk-compensation-chair-robyn-denholm/

It’s official: Tesla’s board collectively enriched themselves at shareholder expense to the tune of nearly $1 billion

...

The settlement requires numerous past and present members of Tesla’s board to return roughly $277 million in cash and $459 million in stock options, and forgo further promised compensation worth $184 million

...

The carmaker’s board has repeatedly found itself at the center of controversy. It includes multiple business partners and personal friends of CEO Elon Musk. His own younger brother is a director, for instance.

...

The Tesla CEO revealed in court testimony he negotiated not at arm’s length with his directors but with himself. He was allowed to draft the conditions of his package as he saw fit, and the board put it to a shareholder vote, while failing to acknowledge that they were beholden to him.

Equally, there's plenty of fun charts which show how the wage ratio has drastically increased over the last few decades: in the USA, it's gone from around 20:1 in the 1965, to around 400:1 today [*].

https://truthout.org/articles/ceo-pay-has-grown-by-1460-percent-since-1978-as-workers-wages-stagnate/

If you're willing to believe that a modern CEO is somehow producing 20 times more value than their 1965 equivalent, then I have a nice bridge to sell to you.

The above article also highlights another interesting fact:

According to research released by EPI on Tuesday, CEO pay has skyrocketed by a staggering 1,460 percent since 1978. This has far outpaced the growth of the economy and even the pay of the top 0.1 percent, EPI finds, with the S&P stock market growing by 1,063 percent in the same time and the earnings of the top 0.1 percent growing 385 percent between 1978 and 2020.

Admittedly, while this is all based on averages, I'd be interested to hear an explanation for why CEO renumeration packages are growing faster than the revenues of the companies they're working for!

[*] Things are somewhat better in the UK; the ratio is "only" around 120:1!

Re: @wolfetone

> You get paid based on the value you produce

Interesting thought. So when do we start culling the CEOs?