Posts by rg287

OpenAI puts Stargate UK on ice, blames energy costs and red tape

It may not. They’re also slowing down the US expansion that Oracle was supposed to be building for them on land they didn’t have with power reservations that didn’t exist.

Don’t forget Altman was also supposed to take 40% of Micron’s DDR output this year, but turns out that was just an Expression of Interest, which they’re not leaning into - Micron’s stock is now tanking because they shuttered Crucial on the basis they didn’t need to retail to pleb consumers anymore if Daddy Altman was going to buy everything.

Microsoft hints at bit bunkers for war zones

Re: LOAC1

MS might be better to try and push all countries to accept that a) civilian infrastructure should never be targeted and b) civilian infrastructure should never be used by the military (this making it a legitimate target), but good luck with that.

It doesn't even need to be all countries really. They can start with their own country. If the US were to accept that civilian infrastructure is off limits, and then imposed that same restriction on states like Israel (as a condition of sponsoring them with all those sweet munitions), then that fix the 80th percentile of issues.

But they won't.

Obviously in this particular case, Iran have been lobbing drones and missiles into surrounding areas, but as an extreme "chaos-monkey" response to the US and Israel bombing civilian infrastructure, schools, and threatening to "wipe a civilisation off the map"... so y'know, maybe lead by example.

Amazon rewards loyal Kindle devotees by closing the book on old e-readers

My wife and I are both in this boat. She bought me a paperwhite gen 1 when we were first dating, having been pushed it on Black Friday. She has a slightly older Kindle. Had I been in the market for an e-reader I'd have had one of the DRM-free ones, I think Sony did one at the time, and IIRC Kobo comes with proper ePub support instead of needing Amazon's proprietary format. I was always lairy of Amazon's lock-in.

Anyway, it is no surprise that the selection of Kindles being discontinued aligns with the devices that still used .azw instead of .azw3. The DRM on .azw is keyed to the device serial number - Amazon would encrypt the file at the point of delivery to a device. Consequently, it's not very hard - using a Calibre plugin - to break the DRM for files downloaded to an older Kindle.

I've done this for a while. My wife has - in the last 24 hours - made a point of downloading her entire catalogue to her Kindle and we'll spend some time at the weekend running those files through Calibre and securing our library. And then we need to find a decent source of non-proprietary eBooks.

Cloudflare, GoDaddy team up to curb AI bot brigades

Love the idea that Ameican big-tech wants to gatekeep bots through an American-controlled registry if you want to get to sites by automated means.

Fuck RSS then!

On the one hand, I fully understand the problems that sites are having, getting slammed by both the AI companies themselves, along with AI-slop vibe-coded apps which will make a thousand API calls instead of downloading local copies of data. I am on the receiving end of that myself, as are many people here.

On the other hand, we're entrenching a big-tech-only internet where you can consume with your eyes, but any attempt to innovate or build your own tooling will be stymied by the requirements imposed by big tech to handle the pressures created by... big tech.

Not that this is new - application signing, walled garden app stores, etc. But it's sad to see the great project decline in this fashion.

Shots fired – literally – over proposal to build datacenter in Indianapolis

You're also very naive if you think "either party" are without corruption.

This children, is what's known as false equivalence.

The Democrats and Republicans both have corruption. No doubt about that. And every politician is looking at their exit, and buttering up some consultancy position for their post-political career, which usually involves doing dishonest things in office.

But it's only under Donald Trump's Republican administration where:

* The POTUS refuses to put their assets in a blind trust, unlike every other POTUS in history.

* POTUS is using his position to promote meme coins like $TRUMP, from which he personally profits.

* Suspicious stock market activity relating to major policy announcements on tariffs, war and foreign policy.

* One Polymarket user made more than $400,000 in profits betting on Maduro’s capture on an investment made within 24 hours of the military action

* Press Secretaries are abruptly walking out of press conferences seconds before betting thresholds.

This administration is the most corrupt in US history. All presidents had conflicts of interest. Name me another who became $3Bn richer during the first two years of their presidency.

This isn't the polite under-the-table corruption of our grandfathers with "there's a board seat waiting for you"-type deals. It's flagrant cashing in at the US taxpayer's expense, with his cohort of bottom-feeding scumsuckers picking up the crumbs by trading their insider knowledge on Polymarket.

This is without precedent outside of tin pot dictatorships and countries that are trying too hard with the "Democratic Republic of..." monikers.

Whilst I do not condone political violence (or violence of any kind really), this very much falls into the "Surprised it's taken this long" category, amid the sombre risk assessment that as long as politicians continue to represent vested business interests over and above the interests of their constituents, this is unfortunately likely to become more common.

I'm also kind of surprised that noone tried dropping a couple of incendiary mortar shells into the "Colossus" xAI datacentres that have been illegally running "backup" natural gas turbines 24/7.

A mortar is not hard to make - metal tube, a bit of propellant. A very simple rig could start quite the conflagration in amongst the turbines and gas storage. Heck, in a country with more guns than people, a few well placed shots with a .308 into the grid transformers or gas-handling pipework would bring the whole thing to a screeching halt with an earth-shattering kaboom.

The Colossus site in Memphis may now be a fait accompli given they've completed construction of a new substation, and some portion of the turbines are reportedly being removed. But as awareness of the issues grow, future abusive developments could well see rather more stubborn resistance if they're polluting the air or dumping brine from the cooling systems.

Brit lawmaker targeted by AI deepfake fails to get answers from US Big Tech

Re: End of satire?

Yes, I believe she did - but they wouldn't have been able to make Spitting Image if she and everybody else depicted had been required to sign releases for their likeness.

It only exists because that fair use principle for fair comment & satire exists.

Re: End of satire?

Since this is UK, the US First Amendment doesn't apply. Does this mean saying goodbye to any impersonation, even Auto-Vox, for comedy purposes? Maybe there needs to be a carve-out to protect a certain level of speech.

People already have a right to their likeness, which is why commercial photographers get model releases, specifying what the images will be used for and any limitations. These rights are generally focused around commercial exploitation of the image or likeness, so do not preclude a "fair use" defence for purposes of journalism or things like satire (Thatcher is hardly likely to have signed a model release allowing Spitting Image to use her likeness).

We also have libel and slander law where stating things that are untrue about a person (or republishing such) may be an offence. Although in this case, the social firms may get off under the "Innocent Dissemination" defence offered to operators of websites in the Defamation Act 2013.

It's not a big step to look at an "impersonation" offence, which doesn't include satire or similar "fair comment", but basically bulks out libel law to explicitly include an attempt to deceive, perhaps tightening regulation on websites to say it's not enough to add community notes - they really have to take stuff down.

Re: Priorities

Funny they put so much effort in a case affecting an individual, but will do nothing about rampant scams being pushed to unsuspecting users 24/7 that are in the grand scheme of things much more damaging.

True... but.

In terms of demonstrating harms in any meaningful way (e.g. in court), it's not enough to wave vaguely at a lot of stuff and say "bad". Not if it comes to judicial rulings or anything with teeth. You often need to get into the specifics of a single case to test the subtance of the allegations - which may then set a useful precedent for other cases or allow other cases to be expedited and bundled in on a "yep, same category" having checked the prima facie details.

It's why class action or large cases sometimes focus on a John Doe as the test case.

In this case, asking about misinformation in vague terms will garner a similarly vague response of "we have systems to minimise it", whereas being able to say "Why didn't you kill these defamatory deepfakes of this person who is sat in this room staring at you?" really lets them needle the respondents and demand specific answers and get into the nitty gritty on a meaningful basis.

It's also why yesterday's ruling against Meta/YouTube in LA is so foundational - it's a major precedent that could affect hundreds of other ongoing cases.

UK wants to know if banning under-16s from social media does anything useful

Re: Banning stuff

Surely rather than deciding if we should ban kids from using it we should first decide if we shouldn't just ban the whole toxic cesspit of social media for everyone.

We should certainly ban it for politicians. A trial period with all 650 members of the HoC would provide double the sample size of this study, with a double-whammy of reducing foreign influence in government!

Stephen Collins on "X" Addiction.

Re: Banning stuff

Or at least ban the algorithmic feeds.

Just to play devil's advocate though, a chronological list of posts is an algorithmic feed. Any sort of sorting or filtering process is - by definition - algorithmic.

Yes, I know what you mean. And I agree. But also, quite hard to legislate for unless you have a list of "approved algorithms".

They can ban targeted advertising whilst they're at it. That's actually easier to define - ads to be contextual, based on the domain/page where the impression will occur. No user data to be collected or provided to the RTB market.

Admittedly UKGov has prior form in this area with the Psychoactive Substances Act 2016, intended to address (non-alcohol or nicotine based!) "legal highs".

On the first pass, the definition of a prohibited “psychoactive substance” being: "produces a psychoactive effect in a person if, by stimulating or depressing the person's central nervous system, it affects the person's mental functioning or emotional state;'

Which of course includes food. Because people get fucking grumpy when they're hungry and suffer declined cognitive ability. They get marginally less grumpy when they're fed (or at least, are grumpy for different reasons!).

So then they carved out an exemption "7 Any substance which- (a) is ordinarily consumed as food". Which is... at best subjective. After all, correct-thinking people don't ordinarily consume marmite. But for some reason it isn't considered a legal high. As for Surströmming... less said the better.

The PSA has become somewhat of a "ways and means" act, relying largely on prosecutorial discretion and a fear of what the courts will tolerate rather than a solid definition of what is prohibited.

Open source isn't a tip jar – it's time to charge for access

Re: Development != Distribution

The unnecessarily complex infrastructure that has grown up around open source, where Sonatype, NPM etc replace the need to manually keep up-to-date with your dependencies (a problem) with the ability to depend on tens of thousands of packages of unknown provenance (a bigger problem).

Absolutely. I use Hugo (a static site generator) for a small site. I set it up years ago. You write in markdown, run hugo and it spits out a static HTML site (including RSS, sitemap, etc) that I upload via FTP. If you're using a theme someone else has written (which many do), you just drop that in /themes/ and specify in the toml or yaml file which theme to pull in. No dev required as such.

Recently I had cause to create a new site, which called for a slightly different (gallery/image-oriented) template. Imagine my surprise when, in trying to download my selected theme, the instructions were all along the lines "add this module to your toml with "https://github.com/<user>/<project>" blah blah blah

Moreover, the "quickstart" now reads "install hugo" and "install git".

No. Feck off. Git is not a prerequisite to using Hugo, but they now go out their way to make it sound like it is. I don't have git installed on my system and I don't want to install it. I don't have anything against it - I can see the value of source control for teams. But I just have a folder of markdown and template files that I want to run hugo against reliably and consistently. If the author unpublishes the repo, I want it to continue working (without update, but that's no problem). My interest (in this case) is in writing, not web development. I want a relatively simple system that builds my writing into a site, without having to manually update menus, article lists or the sitemap. That's what I need the SSG for. It ought to be getting out of my way as much as possible, without trying to get me to build some ever-so-clever CI pipeline for this thing I use once a month...

Now, it turns out you can still do this, but it's less obvious which folders ought to be copied where (unless you already know the "old" way of doing things). The docs and happy path are all "hey, pull stuff in as modules from github on demand". The theme I wanted didn't provide an easy <template> folder. The maintainer had a bunch of folders at the top level, plus the usual repo cruft. There were .sh files to pull in example content (if you were running it as a module), none of which you would want if you're just dropping it into the /themes/ directory.

People have been complaining about this for years now, and the response of the Hugo maintainers has basically been "You're holding it wrong", "It's not for noob. Noobs should try wordpress (php+MySQL stack?!) or jekyl", "people build Hugo themes and templates in lots of different ways, don't ask for a happy path". Ignoring the fact most major/uber-complex themes are going to be used for institutional projects by professional teams. It wouldn't kill them to specify to developers "You basic blog/website/gallery templates that we're going to feature in the gohugo.io site need a folder that can be dropped into the /themes/ directory and more or less just work".

The load comes overwhelmingly from large users, not hobbyists. Fox's analysis shows that 82 percent of Maven Central demand comes from fewer than 1 percent of IPs, with roughly 80 percent of traffic sourced from the largest cloud providers' infrastructure. Now these companies could easily run their own local mirrors, but they don't. Instead, they hit up public open source registries on every build, test, or scan. All of this drives bandwidth, storage, and operational complexity, which eats up cash like an elephant does peanuts.

I should have thought this would be relatively straightforward to address with rate limiting? I'm sure some will try to play silly buggers and rotate addresses, but once you start investing that much engineering effort, you could just spin up a mirror/local cache.

Maven already seem to have some infrastructure to regulate this. At the bottom of their mirror guide, they state:

The size of the central repository is increasing steadily To save us bandwidth and you time, mirroring the entire central repository is not allowed. (Doing so will get you automatically banned.) Instead, we suggest you setup a repository manager as a proxy.

If they can ban someone trying to mirror the entire repository, they ought to be able to ban or greylist IPs repeatedly calling for the same packages multiple times a day/hour/minute (or any combination of packages at a rate which breaches the Terms of Service/Fair Use).

Rate limiting could be quite broad-brush, like max downloads per week, so a hobbyist who bashes away at something for a weekend but then doesn't touch their project for a month is at no risk of tripping over daily or hourly limits, whilst enterprises unreasonably imposing sustained load onto repos will blow their allowance in hours.

You could also rate-limit unauthenticated calls. Companies that don't want to run a proxy could pay for a high-volume api key.

Of course as Androgynous Cupboard mentions, there are two issues being muddled up here. (Ab)Using commons infrastructure commercially without contributing back is quite a distinct matter to the other issue of (not) paying developers to maintain code.

Water company wasted $200k on bad answers from an AI model – so built its own slop filtering system

Re: Maybe That's Her Flaw?

"PhDs and PhD-track scientists" like herself

Um, no? Holmes dropped out as an undergrad student - she was 19/20 years old (born 1984, dropped out of Stanford 2003). Not only was she not "PhD-track", she hadn't even graduated with a Bachelors or Masters. She blagged her way into being a research assistant, which is normally reserved for grad students, but she wasn't one herself.

maybe if Liz had Rozum, then she'd still be running Theranos.

The PhDs weren't the problem. The engineering and the fraud were the problem. Which is to say, the tech kind of worked, but sporadically and unreliably. When investor capital started slowing down, they pivoted to lying about using proprietary tests.

She wasn't convicted of failing to make the product work. She was convicted of defrauding shareholders. And Bulwani was additionally convicted of defrauding patients.

Your startup failing is not a crime - plenty do. Lying about what your startup is achieving (to the board, investors and regulators) and endangering patient safety is not okay.

Yeah... I mean, come on. Even Theranos hired PhDs and PhD-track scientists to do the actual chemistry.

"Not being PhDs in the space, we read relevant academic papers and used LLMs like Grok and ChatGPT to validate our findings.

I don't want to gatekeep startups, because innovation can come from outside the groupthink. But for a chemistry-based startup looking at things like desalination... maybe find a chemist?

Horizon redress still a mess, MPs say – and Fujitsu hasn't paid a penny

The overall amount of work probably hasn't actually changed. It's just that in the old days then you basically just passed all of the work onto the people processing it in accounts.

With computerisation they've presumably eliminated the work from the accounts department and passed it onto the person making the expenses claim.

Quite true. The amount of work won't have changed but people are less efficient at processes they only do occasionally. Consequently, the overall prodoctivity has probably dropped. What took the poster half an hour to submit then likely took an accounts spod another half hour to bash out - because they do it all day every day.

If the poster is now taking half a day to fumble through a horrible unintuitive system, then the net productvity is 25-33% - what was 1 hour of labour (2x 30mins) is now 3-4 hours of labour.

There's something to be said for professional services. Turns out all those "pen pushers" were actually better at their jobs than you or I am!

To be fair any records from 30 years ago are going to be on paper.

When this process notionally started over a decade ago, those records would only have been 20 years old. They've had a decade, there is no "to be fair" about it.

...and when it reaches a certain point, start slinging executives behind bars.

As a general rule, I am a prison reformist - short sentences don't work, Norwegian model has lower recidivism, etc.

I do have an exception though - debtor's prison. There are some people for whom fines make no difference (cost of doing business) or who abuse the system to avoid paying out or hide behind a corporate shield. As if corporate crime is committed by a legal fiction - not real humans doing bad things. What we can take from them is some years of their life, and however rich they might be, that's a hard cost that's the same for everyone.

Only for egregious cases of course. No sense in locking up everyone who has missed a month of some loan payment. But this would certainly qualify - especially since many of the victims were incarcerated as a result of the Post Office & Fujitsu's negligence.

Swiss e-voting pilot can't count 2,048 ballots after USB keys fail to decrypt them

Re: Even Paper Has Its Issues.

For those not in the UK where we still vote by ballot paper (or post), there have been numerous issues with sorting/counting of ballot papers over the years, leading to multiple recounts due to counters miss counting.

Not really. For sure, there's the odd close contest and they have a recount and come out with a slightly different tally due to human error, so have to do a third to make sure. And doing recounts for validation is no bad thing. It adds scrutiny and assurance ot the process. The main thing is that they CAN do a recount. At the end of the day, you've got a room full of ballot papers. Whatever problem you may overcome, you'll sort it out eventually. Not so if you've got an electronic system which only a few people in the world know the ins-and-outs of and even fewer understand the cryptographic basis.

Than there have been allegations of ballot boxes being tampered with, votes going missing.

I call FUD. Go on... ballot boxes being tampered with would be national news. Yet I can't find any history of this (other than unsubtantiated "We're very concerned it might happen" claims from certain parties... you can guess which).

The only news I can actually find is an allegation of tampering in relation to a Labour party candidate selection vote in 2009 - so not managed by a local authority nor covered by criminal legislation.

Tower Hamlets of course is the lightning rod there following the 2014 Mayoral election, and the issues really surrounded fraudulent postal voting, double voting, "treating" and various other issues. There were not problems with the actual count itself. People now like to hold up Tower Hamlets as this exemplar of corruption... except it was one specific election and the council now have some of the most robust and heavily scrutinised polling station management in the country, extensive police presence and availability to polling station officers.

I'm mostly in agreement up to scanning ballots at the polling location.

To be clear, the idea of this is a black box with a scanner, so you scan your ballot, it goes beep, and you post your ballot. There's no output at the polling station except perhaps a count of total scanned. Theoretically it could also issue a receipt to the voter. This could be useful for corporate/shareholder elections but not civic elections where coerced and family voting is a hypothetical concern and you want it receiptless.

This black box is taken to the counting station with the ballots where the counts can be read out and a provisional result found. There are a couple of benefits to this - the tally on the box should match the votes in the ballot box (plus or minus a % margin to allow for mis-scans or spoilt ballots). Anyone wanting to tamper with the ballot box would have to tamper with both in a cohesive way.

In some circumstances you could do a statistically significant manual count - if the computer shows a landslide, you only need to count ~20% of ballots to check. If it's tight then it's a full count. If course for most civic elections a full count is desirable anyway.

In the US, those states which inexplicably take days to return a result would be able to issue a provisional result very quickly. Although they could also just improve their counting arrangements.

Of course, the state of public discourse in the US is such that any variation in the electronic and manual count will be seized on as fraud, even if it's just 0.1% mis-scanned or spoilt ballots and three manual recounts all reach the same result, so not much use there.

Don't do e-voting. Not for anything important. The research has been done and the results are in. Matt Blaze (who co-founded the voting vilage at DefCon) has written extensively on election integrity.

Use paper ballots. By all means use an electronic counting system (e.g. a scannable ballot that is scanned prior to posting in the ballot box) - that allows you to get an instant provisional result when the polls close and perform a manual count if the electronic system fails, the result is challenged, or to verify a close result.

But the actual votes? There's no substitute for a mark on a bit of paper.

Lightmatter says latest photonics will slash datacenter fiber bills in half

Re: "They could have spent 3Bn"

Only if the major cost of that was the cable.

The deal was with Corning, who make fibre. So presumably that's the fibre component, not project cost incl. installation. Also, Harris says they "spent"... when you click through to the WSJ article, "Meta has struck a multiyear deal worth up to $6 billion". And it was in January this year, so they've probably not even taken their first delivery yet.

So they haven't spent $6Bn yet, and if they were to adopt new photonics, they could order less. Although not that much less if they've locked in major architectural decisions.

Also, maybe they are looking at BiDi and the $6Bn is already down from $10-12Bn?

Perplexity Comet hurtling toward Amazon ban

If you've a commodity that can only take limited customers, say concert tickets or flights on planes, it's not unreasonable to try to get mucho money from people who want to go. And then to lower the prices later, if you've not been able to sell at top dollar. And then, in the case of airlines, to bilk the passenger who absolutely needs to fly somewhere tomorrow. This is basically rationing. The people who want it more, pay more. The people who are willing to go to any gig, or fly at any time, can wait until the last minute and then see what wasn't as much in demand and take that for cheap a few days before.

If you've a commodity that can only take limited customers, say concert tickets or flights on planes, it's not unreasonable to try to get mucho money from people who want to go. And then to lower the prices later, if you've not been able to sell at top dollar. And then, in the case of airlines, to bilk the passenger who absolutely needs to fly somewhere tomorrow. This is basically rationing. The people who want it more, pay more. The people who are willing to go to any gig, or fly at any time, can wait until the last minute and then see what wasn't as much in demand and take that for cheap a few days before.

Or.... and hear me out here... you can work out what your cost of product/service is, decide what your margin should be, and then sell your products/tickets/services at that price. Why shouldn't everyone pay the same to get on the same flight or into the same gig? These companies know what their price points are. They know that an Oasis gig will command higher prices than some little band on their debut tour. When the flight is full, it's full. There's no operational reason to ramp the prices so the last 3 seats cost 4x more than the first 3. Or so that they cost more on a Saturday evening when people are watching TV and see an advert and pull out their laptop to book a holiday.

Of course even in the analogue era, there was a level of demand pricing - a hotel would have it's rate card (probably with high/low season pricing), and then they'd have the tariff board behind the counter for anyone who walked in and asked for a room that night. But again, at least it's transparent, rather than looking the person up and down and making up a number on the spot.

However, I wouldn't allow this for trains, or other public transport, in the UK - as we want to reduce car use so shouldn't penalise people for not knowing where they're going in advance.

It's unfortunate that we absolutely do this as a form of demand management, and yeah, it should be banned (and unnecessary). The problem with trains vs planes of course is planes have a booked seat - trains can (in principle) run to standing-room-only. They don't run out of tickets as such (at least not the way we run them) - so pricing encourages people onto quieter services. BUT this is a product of us not investing in our railways for 40 years and running them very inefficiently and at low density. A great shame HS2 was scaled back as this would quadruple the number of local services we can operate once the ICE trains are segregated onto their own line.

I would note that this could be applied to plane tickets as well. I know someone whose mother (in Spain) was taken gravely ill and just had to bite the bullet and get the next flight out, hang the cost. They were in a privileged position that they could, but others aren't. Particularly in the UK, with the issue that Eurostar is not very well run, doesn't run as many services as it should, and charges the hell out of the ones it does. Insane that it's generally more expensive than the equivalent flight (although not by as much as we might think including airport transfers/parking/etc). Aviation is not always a luxury/holiday commodity the way concert tickets are. Of course we need to get continental rail services sorted, but aviation is actually a form of public transport - albeit one that needs to be deprioritised versus rail.

Universal Credit tickets available for £1 -- Manchester United used to have a scheme where you could queue up on a Tuesday morning to buy discounted tickets for Old Trafford matches, to guarantee that local supporters could get in, when prices suddenly jumped up to £70. Bet they don't do it now. Most football clubs limit scarce Cup and European ticket sales to people who've got season tickets or have bought a certain number of tickets in the last season, so the "real fans" get first dibs on the limited resource.

These are just discounts though. They're not the sort of airline dynamic pricing where they bump prices based on whether you've looked those particular flights or destinations before. Moreover, they're transparently declared ahead of time. If you're on UC, you get cheap entry. If you want scarce tickets that badly, join the supporters club. This isn't a perfect system of course because you're discriminating against low-income fans who perhaps can't afford to attend many matches, but would get up at 3am to queue for tickets (when other - better heeled - fans wouldn't, but they can just buy membership or a bunch of minor match tickets to qualify for the scarce matches).

Look at Walker's crisps. -- But I don't think we should legislate for it. How do you define quality?

We've drifted well away from demand pricing here. This is just cost of living, along with supermarket's "consent or pay" shenanigans. Upfront pricing like that is for the markets to decide. If Apple or Walkers are too expensive then people won't buy them. Vertu phones are insane money for what they are, but a handful of more-money-than-sense people buy them. That's fine. What should be legislated against is the idea that Apple charge more because your tracking cookies/advertising ID shows that you've looked at certain other brands and they've inferred they can bilk you for more. Apple are actually fine. That's an example of that I want - they set a price, and that's what you pay and I don't have to waste my life trawling through camelcamel working out that if I logon at 8.33am on the second Monday of the month using Safari 26.1 I can get a discount for <reasons>.

Sure, Apple drop the price on old models, and have their refurb programme, but their prices don't wander up and down algorithmically the way airline prices do.

Of course, if we banned RTB advertising and moved back to contextual ads, then most of the tracking and datasets used to drive dynamic pricing simply wouldn't exist, which is possibly a neater "two-birds-with-one-stone" way to wind ourselves back to some sort of vaguely transparent pricing policies. So there's that.

The bargaining part in particular is in our favor and much needed to counter the mountain of algorithmic manipulations effected by e-commerce sites (including tavel, lodging, vehicle rental, ...) that are so stacked against our best interests and wallets.

This is not in our interests. How would you verify that your bot is negotiating the best deal from Amazon? Is your bot a better negotiator than a different bot, or did Amazon just strategically make you a better offer this time? This just adds to the complexity of that algorithmic manipulation.

There are two ways forward from this, and they both involve cutting the Gordian knot rather than developing higher planes of knot theory.

1. Don't use online souks. Buy from retail stores or independent online retailers (e.g. ordering groceries from a supermarket, via the supermarket website - not with some third party intermediary involved). This admittedly doesn't help the situation where the vendor is using dynamic pricing like travel/flights, etc. But your average independent retailer's website is generally going to have a price and that's the price. I haven't used Amazon in a year now and feels great. But I'm also an oddball deliberately carrying more cash for small (<£20) transactions and generally being a bit luddite in my petty quest to bilk Visa/Mastercard of their fees.

2. Legislate against dynamic/algorithmic pricing (as we should against targetted ads - contextual ads only). There's a price, that's the price. Done deal. There's a bunch of complexity to it, but there are business practices that harm the consumer, and we should just make illegal the same as cartel activity and monopoly behaviours.

RSS dulls the pain of the modern web

Happy user of FreshRSS for a few years. Saves having to try and sync between devices - it's PHP based so just sits on some cheap Plesk hosting I have for other projects and I have it pinned on my phone's home screen. Makes it tremendously easy to scan through a list of headlines and quickly pick out the ones of interest, though I have nowhere near 2000 feeds!

Re: BBC repetition

I subscribe to a few BBC RSS feeds. I expect repetition across different RSS feeds, as the same story is published in multiple feeds that each relate to the story. However, for the past little while (months?), the repetition rate for individual RSS feeds is frustrating.

I can't say I've had an issue with repetition on individual feeds, but I only subscribe to a couple of low volume feeds (BBC Sport Archery & Shooting), so I'm guessing those articles don't get edited or republished with updates like breaking news might. Totally agree that cross-publishing is quite annoying. The Shooting feed will get polluted with people lobbing in gun crime or Countryfile stories even when they're literally nothing to do with the shooting sports (if an Olympic shooter got caught doing something naughty then fair enough, but "Man shot in digbeth with illegal handgun" or "Deer culling to be made easier to protect trees and crops" doesn't have that angle).

I will hand it to the BBC that basically any category on their website has it's own feed. This is not especially well documented, but if you want just the latest Lawn Bowls news, than you can have it (https://feeds.bbci.co.uk/sport/bowls/rss.xml). This is excellent, and a beacon of light in an otherwise dark world (wide web).

For comparison, the IOC website has no RSS feed, so no way to follow their news, much less an Archery or sport-specific feed. That's to say nothing of British Shooting's abomination. A web app masquerading as a web site. Looks pretty, but view the source and it's just a blob of JS. No RSS, no sitemap, no robots.txt. NO FUCKING HREFS. You can't even try and scrape their news reliably because the links aren't links - they're buttons that fire a blob of javascript to call the next page via a fucking API... It's like their agency gave it to some 17 year old intern hyped up on some JS library that was out of fashion before they even deployed it (that agency being Digital Glacier - who also do the UKSport website, which also doesn't have an RSS feed for it's news, although it's news page does at least have some functional HTML and anchor tags to go at).

If Wordpress and Joomla! have one thing going for them, it's that they generate RSS, sitemaps and all that other stuff as by default. The writer or developer don't even need to know it's happening to get the benefit. Which is great for the rest of us who want that feed.

Britain spends £180M to work out what time it is

Good.

Now drop some cash with Trinity House to pick up their eLORAN trials. Trinity House discontinued their Initial Operating Capability in 2015, alongside partners in France and Norway. They knocked off their DGPS network in 2022 on the basis multi-constellation GNSS and RAIM (Receiver autonomous integrity monitoring) were adequate.

DGPS probably isn't actually needed - either you have GNSS (multi-constellation, RAIM) or you don't, and if you do, then it's highly precise these days. For maritime, the integrity/anti-spoof provision within DGPS was as important as the improvements to positioning. If you don't, eLORAN would be handy.

In the scheme of things, the cost of a dozen eLORAN base stations as a backup to satellite systems is trivial.

Re: Is this really necessary?

Every country (of a certain size. Leichtenstein & Monaco less so) ought to have a robust timing capability, so no great coordination needed there. We already have a good capability but a distributed network is no bad thing. £180m is a noddy sum to Government given the chaos that would ensue from time drift (the NHS in England & Wales spends >£350m/day).

What would need more coordination would be rolling out something like eLORAN to lighthouses/nav points as a terrestrial fallback from GNSS. Trinity House discontinued their eLORAN trials in 2015, alongside partners in France and Norway. They knocked off their DGPS network in 2022 on the basis multi-constellation GNSS and RAIM (Receiver autonomous integrity monitoring) were adequate. In the scheme of things, the cost of a dozen DGPS and eLORAN base stations is trivial as a backup to satellite systems.

Re: What is this cesium of which you speak?

I’m not sure I want to know about the King’s Cheese.

US struck Iran with copies of its own drones

Re: Proof this was long planned

I think it is unlikely that the Pentagon has not considered the medium term impact on the global oil trade.

For sure. They've considered it. Trump hasn't.

So, either the US expects the Strait to re-open within a few weeks or they have a plan B.

Lol. Plan B? The first part is right. They thought they'd lob a few missiles, kill the Supreme Leader (Khamenei, not Trump) and then Iran would roll over and elect a new - friendly-ish - leader like Venezuela did. Turns out they've learned absolutely f-ing nothing from Ukraine since 2022 about the new face of warfare (i.e. drones).

There is no strategy from Trump - just (re)action. The Turkish have been itching to kick the US out of Incirlik air base. Taking heat from Iran because they host the US is only going to harden that. They've already been booted out of the F35 programme, so no need to cosy up to Trump for that. Cue various Middle Eastern states saying "hey, will you stop lobbing rockets at-and-around these bases if we kick out the Yanks?".

The only way to do regime change in Iran is by boots on the ground - a trillion dollar commitment no smaller than that of Iraq or Afghan. Trump won't likely do that. He thought he was going to do a "special military operation" and be in and out in a few days, not understanding that what he did in Venezuela was not a military operation - it was a military-enabled policing action. There is a sequence of events here where the US ends up making an ignominious withdrawal, and potentially ending (or significantly scaling back) it's military presence in the region because they're no longer welcome or useful. Korea and Japan are also reassessing whether it's actually helpful to be hosting US bases.

Trump is speed-running the end of the US hegemony, and applying to join Israel as a pariah state.

The other consequence to this is where Israel goes. Netenyahu the nutter has talked about his "seven fronts" - as if Israel can sustain substantive operations simultaneously in Gaza, the West Bank, Lebanon (x2), Yemen, Syria and Iran long term. They already have a problem with reservists not reporting for duty, and there's domestic division where regular Israelis are losing patience at the service exemption for Haredi scholars. The question is how hard Iran goes chaos monkey on Israel and whether they turn from "nuisance" to "Tel Aviv actually thinks about nuking them because they've run out of conventional forces." Israel definitely can't do boots-on-the-ground, at which point we can expect to see an intervention from Russia, India or China laying down the consequences (for both Israel and the USA) if Israel does something truly stupid - this intervention is frankly overdue given that Israel has committed genocide in Gaza worse than the Serbs managed in Bosnia in the 1990s - when the US belatedly entered the fray and convinced Russia to cut support and militarily starve the Bosnian Serb's campaign of ethnic cleansing.

Apple jacks up MacBook pricing with M5 Pro, Max debut

You piqued my curiousity - this definitely does not happen under 15.4.1. I might hold off on that upgrade a bit longer...

Weirdly I haven't managed to reproduce it (15.7.3), so not sure how I upset macOS enough to offer to bin it. It was definitely a raised eyebrow moment.

M5 MacBook Air starting prices, meanwhile, are up $100 to $1,099 for the 13-inch model

I was wondering how long their $999 entry point would last. To their credit, they've held that for years, and maintained it across their (belated) base-spec change from 8GB to 16GB.

Not that it matters. I'm on a work M3 Air that will last a decade, and my next personal system will be Linux. macOS has taken a nosedive in the past few years. Laggy dialogs when network drives are mounted (even if I'm only trying to interact with the local filesystem), the requirement to do command-line-fu to run unsigned applications like LibreWolf and - most recently - the strong desire by the OS to toss any file it's not sure about into the Trash. Last week I inadvertently tried to open an SVG in Preview (which doesn't support SVGs). You might expect an error, and an error is what I got. Along with a security warning and a dialog offering to "Move to Trash"... WTF? Just because an application can't open a file doesn't mean it needs to go in the bin.

I'm not against an amount of "Are you sure" for random applications downloaded from the internet because that prompts my parents (who would otherwise run any old malware) to at least pause a moment. Tech support callouts are much better than the good-old-days of "autorun anything anywhere", but recent changes are now actively obstructive to work.

And this is just on Sonoma/Sequoia, which have been less responsive than Snow Leopard on a 2008 Macbook with 4GB of RAM and a dual-core Intel processor. Tahoe? No fucking thanks.

UK still doodling digital pound while Brussels frets over payment sovereignty

There is a cost to going cashless, but you also get a lot from it. As do the merchants, who don't have to deal with a lot of the hassles and time with cash.

Well yes and no. If you're a coffee shop on the high street, it's very little bother to go and pay cash into the bank down the road - provided you're satisfied that your staff (if any) don't have their hand in the till. I've seen an increasing number of small businesses have signs behind the counter saying "We accept card but appreciate cash - our cards fees were £xxx last month". In a snub to modernity, our local chippy is cash-only (there's a free cashpoint literally next door). I have my suspicions that there's some money laundering in the back because they do seem excessively staffed. But they also do a good fish and chips and a brisk legitimate trade (they were our next-but-one chippy, but our nearest closed at the end of 2022 due to energy prices. Thanks Rishi).

Speaking for myself, I've started carrying cash much more frequently. I'm a member of a sports club and camera club and try to use cash for small transactions (since - proportionally - you lose a lot per transaction on little bits like a fiver). I also just BACS money across for occasional/one-off sums like annual membership and show the treasurer the transaction in the app.

There's a privacy aspect and a small, petty satisfaction in bilking Visa/MC of their fees.

Of course I still use my cards frequently - supemarket shops, anything online, etc - but really try and use cash for the small stuff.

Unless the government choose to set up their own network, in which case they'll either charge for it, or raise taxes in order to pay for it. Although I'd suspect the more likely option is that they'd ask the UK banks to form a joint venture to set up their own version, and then all that would change is who you pay your fees to.

That's exactly what Norway have with BankAxept. If it's worth doing for 5million people, it's going to be feasible for 70m. Yes, there are still fees, but profits/surplus are retained in the UK - not offshored by Visa/Mastercard via a complex tax arrangement that somehow means they don't pay tax anywhere.

It's possible VISA and Mastercard are making excessive profits. But then they've got global economies of scale. I find it very unlikely that a UK-only system will end up much cheaper.

A UK system would have to be cheaper, or else noone will use it. Bank cards would needed to be dual-coded as per BankAxept so they could be used abroad, as would merchant terminals to support foreign visitors. Consequently, if "UKPay" were more expensive than the others, merchants would just lock their terminals to not use that network.

Economies of scale are important, particularly for having a good baseline against which to do things like fraud detection. The Norwegians seem to make it work and the UK is 12 times the size.

We also have to remember that costs are relative. What is the national security risk of having all our transactions controlled by a US duopoly? Entire industries exist at the pleasure of Visa policy on who they will provide payments to. How do we handle (e.g.) a British citizen being sanctioned by the US and consequently debanked (or at least, banked in the UK, but they can't have a card because neither Visa, MC or AMEx will service them, which makes it very difficult for them to pay for.. anything).

If we're going to have a British payments system there's a risk that charges to retailers will actually be much higher.

There's no reason for that to be the case. BankAxept has lower transaction rates than Visa/MC.

Of course we'll give it to Crapita or HPE to implement so.... yeah, not a chance.

What we have to remember though is it introduces competition. Nobody thinks about networks at the moment. The card is on one network and that's what gets used. The only competition between Visa and Mastercard is in bidding for banks to issue cards for them. The user and retailer have no control short of refusing payment (e.g. AMEX) or consumers literally picking a retail bank based on them issuing Visa specifically (or Mastercard). Which noone does.

If you have a choice with dual-coded cards, retailers can choose to go with the network that offers lower fees - "UKPay" or Visa/MC. Competition is generally a good thing.

Re: Digital currency

Could someone please explain to me the importance of this bullshit ?

Our currency is already digital. The number of times I have paid in cash in the past decade are vanishingly small.

So why is this suddenly an important issue ? Banks have already solved digital currency.

By "Digital Sterling", what they mean is a unique digital token that could be traded offline, much as a £10 note has a serial number and can be passed around. Basically crypto, but issued solely by BoE - not a race to unlock coins through proof-of-work or whatever. The supply is controlled the same as currency and bonds and M1/2, etc.

This is completely different from digital banking, where you are simply using digital means to instruct your bank to pay £5.87 to another bank, each of whom update their ledgers. It's the same basic process as wiring someone money, but done from an app or vendor terminal instead of going to a bank and filling in forms at the counter.

Thanks to the wonders of fractional reserve banking - those ledgers are largely honoured on the basis of handshakes, not backed 1:1 by actual sterling issued by the BoE. Which is why the government has to run a Compensation Service in case a bank goes bust - so that people have confidence their deposits will ultimately be honoured by someone.

Consider this, if you pay in some notes at a bank, they don't record which serials you deposited. If you later transfer £10 to a friend, they don't transfer one of those specific notes to your friend's bank. They just update their ledger. "Your money" in "your" bank account is just a ledger entry. Not actual cash. This takes some getting your head around, and the government would prefer people didn't realise just how much of our banking system is based on handshakes and good faith.

With "digital currency", you're actually passing around a unique digital token that in theory is as good as a pound coin or a £10 note. No bank or server need be involved in the transaction, noone's ledger is updated - it's just your digital wallet holding some number of cryptographically signed tokens.

So no, our currency is not digital. But our banking is - and frankly that's good enough - as you say, all we need is a "UKPay" payment network to secure ourselves against Visa/MC.

Re: One option - ban them

Like it or not Mastercard and Visa control the UK and EU payments systems at present. There were alternatives like Electron (i forget the name so that might be wrong) but they never suceeded.

Payment methods are not the topic of discussion here - digital currencies are a completely different beast (to whit: what's in your bank account is a record in your bank's ledger and isn't directly backed 1:1 by actual Sterling that's actually issued by the BoE, which is how fractional reserve banking works. Of course you can swap it for cash at an ATM - provided there's no run on the bank (Northern Rock) and they haven't suspended withdrawals). The idea of a digital currency is one in which you trade in currency directly minted by the BoE, same as cash. This has the issue of "singleness" - the concept of an individual digital token that can be passed around, not simply a ledger saying you have 157 and Tim has 562.

However, to address your point, Visa Electron was a Visa debit card product. It didn't fail - Visa just discontinued it. The distinction between it and Visa Debit was that is required online electronic authorisation - which is not to do with "using the internet" but meant that funds had to be guaranteed available at the time of transfer, whereas with Visa debit, they didn't necessarily check you had the cash and (to a point) allowed transactions "offline", with your account then going overdrawn when payments were reconciled overnight.

Visa Electron cards generally used for accounts that didn't allow overdrafts (teenagers, poor credit score, etc). Mastercard had the same sort of thing with Maestro, which also covered pre-paid debit cards.

What we NEED more than digital pounds is a national, sovereign payment network much as Norway has in BankAxept (run non-profit by a consortium of banks, with cheaper vendor fees than the US pair). Cards are dual-coded to BankAxept and then Visa or Mastercard, which covers their use outside Norway. Merchant terminals accept either (to support foreign visitors). Something like 99% of domestic transactions are done over the BankAxept network, never touching Visa/MC.

Probably be a lot quicker and easier to set up than a digital currency as well. Albeit there will be a multi-year phase-in period as cards are reissued by participating banks.

Users fume at Outlook.com email 'carnage'

Re: OVH

Similar.

My employer currently hosts on OVH (no outbound mail from there!), but we do the exact same for Digital Ocean and a few others. Our API access is basically allow-listed, since we sell to a relatively small number of corporate clients and we can keep on top of their access points. The web/portal services we provide have a pretty extensive blocklist in CF from both certain countries, and then from ASNs that belong to hosters, since those services should only ever get hit by humans.

Altman: You think AI is wasted energy? Try raising 100 billion humans

Re: 20 years?

1) moving, 30+ hours of sitting still a week and we wonder why kids look like beach balls.....

Not sure those two are strictly connected. We didn't have a child obesity crisis in the 1970s when... kids also sat for 30+ hours a week. More to do with falling nutritional standards, cheap carbs, aggressive marketing. Plus an over-commercialisation of sport, and a failure by school sports programmes to offer a diversity of sports - for which I certainly don't blame individual teachers. They do what they can on a shoestring budget. But basically all school sport is track & field or sportsball. Things that can coached/supervised on a 30-1 staff-student ratio with minimal equipment - balls, or communal kit like a small set of shot puts, javelins, etc. So football, basketball, hockey, etc.

In some respects this is about "get them moving for 90minutes a week and stave off childhood obesity". But this creates a clear divide between "sporty kids" (who fit into the normative "athletic mould") and "non-sporty" kids - even though the latter may well find great sporting success in fencing, archery, martial arts or one of the "bridging" sports that are half sport/half outward-bound like canoeing.

Much has been made of Norway's approach to youth sports, which encourages broad participation and doesn't even count scores until 13+. I guess it helps when you literally engage in cross-country skiing to get to school in the morning. The most important thing is that Norway encourages kids to try lots of things rather than specialising early - "You look like a football player. He's tall, he needs to go play basketball".

Norway makes it fun. Sport is something lots of people do even just for recreation as adults. The commercialisation of sport in the US in particular (but also increasingly visible in the UK and other areas) turns sport into entertainment - if you don't play professionally, you are a spectator (please buy your gamepass and merch. Also, do you want to pay £££ to participate in branded quasi-sporting events/fitness regimens like Tough Mudder or CrossFit).

Or you could just join the local fencing/harriers/sunday league club or team. This is not to say there is no recreational sport in the US or UK. Of course there is. But the governmental approach and general cultural background is moving towards being a thing that you watch professionals do, or take your kids to do (soccer moms), but not something 30-something adults do.

As for the other points. Yeah, when metrics become measures and teachers are performance-reviewed on exam results, they are incentivised to get their kids through exams and nothing else. The job is less about education (which isn't measured) and about preparing for specific exams (which is). And unfortunately this is top-down and can't be changed by individual teachers, because kids need good grades to tick the box for the next step in their life - whether that's apprenticeships, college or university. Doesn't matter how well adjusted or educated they are, they need to prove they know the content of a specific textbook. Everything else is a bonus.

Euro hosting giant hiking prices by up to 50% from April Fool's Day

Re: its a conspiracy

I'm waiting for the bubble to burst and produce baskets of surplus components marked "slightly used only one stupid owner".

If only. But a tidal wave of used HBM and AI-oriented GPUs aren't going to help the rest of us. It's not like we'd get a glut of DDR5 DIMMs and x86 onto the market (although there's obviously some of that). It's all soldered onto single-purpose SoCs. A tsunami of single-purpose ewaste instead.

Founder ditches AWS for Euro stack, finds sovereignty isn't plug-and-play

Re: VPS

"I doubt anyone reading this will need to scale-out their web application to multiple servers due to the number of users."

Are you serious?

Have you ever worked in a large business, say one in the FTSE100? One with tens, hundreds or even millions of customers/users?

They probably haven't, and neither will most people. Because most businesses are not FTSE100s, and most people work for companies with fewer than 20 employees.

So there are two points here:

1. Who cares about tens of thousands of users? FTSE100s, sure. But not most people.

2. Modern silicon/servers are wildly, hilariously powerful and storage dense per U, when considering basic line-of-business applications like web hosting, mailboxes and file storage. You can host >20k mailboxes on a 1U server. Even for moderately Medium sized enterprises, you're looking at a couple of mail servers (for redundancy), couple of webservers, couple of fileservers. The whole thing is less than a quarter rack. But then you want another half rack in another location for failover. These are probably going to be Proxmox clusters running VMs rather than bare metal, but even with the inefficiency of bare metal (in so much as some U will always be significantly underutilised), it's not a lot of boxes, nor a particularly deep architecture. Bakc in the day, I started out at a company that had everything on a single server. They got a bigger one, then a bigger one. I thought "this'll be the last general purpose box we get. We'll have to branch out to DB/web/media servers next time". But Moore's law and storage density kept up. They split out stuff for redundancy and architectural purposes eventually, but the whole lot could still have run on a single box in terms of CPU/RAM/storage capacity.

I've been through a company that had a single server hosting their entire production environment, with a second (in a different DC) for failover, on a hot-cold basis. Their (third party) CDN ran health checks and transparently failed across if Prod went away. That was it. Then there were a couple of VMWare boxes for build nodes and some other small stuff that didn't justify bare metal. You'll say "can't have been very significant", and you'd be right - they were (are) a small firm <20 people with like, 10 customers that you won't have heard of - a mid-west concern in Illinois called John Deere, some insecty outfit called Caterpillar. A niche Korean bunch called Komatsu.

All live traffic was being served (at the time) off one 16-core Windows box, running IIS, SQL Server and a couple of other bits. Kind of terrifying, but also a testament that you can do a whole lot of very real, actually useful processing with modest resources.

In all seriousness, most businesses could host their entire infrastructure on a single 1U cPanel server - web, mail, etc and have empty drive slots and capacity to spare. Yes, you would like some redundancy, but in terms of the actual resource required to run the applications? That's it.

We get carried away in IT with racks of gear and five-9s. Especially when it's not our money. For a small business, their email falling over for an hour or for a reboot is of no concern. Certainly less important than making sure they have immutable, ransomware-proof backups. They receive less than 50MB of mail in 24hours and delete most of it immediately. That's the reality of the real world - in so much as most of the real world is shops and car garages and hairdressers getting on with their business, not software houses with a big old CI/CD build infrastructure, multi-cloud load balancing or any of that stuff.

UK to demand social platforms take down abusive intimate images within 48 hours

These images are at least potentially illegal. Would you expect 48 hrs notice if they were posted to a site that you hosted? Do you think the police would "understand" your workflow?

If I was at work and some images were posted at 2.30pm, and I checked my mod notices when I got home at 6pm, then that's 3.5hours. But of course I might be on holiday, or asleep in another time zone. Such services are necessarily "best effort". Yes, it would be nice to be "immediate". But deleting them "as soon as I become aware of them", which may actually be 6-24hours after they are reported/flagged, is hardly unreasonable. And of course if they're illegal, then I shouldn't be deleting them - I should be removing them from public view, but archiving them for Police.

To criminalise a gap of a few hours for a best-effort service is inane - unless your name is Mark Zuckerberg and you want the entire internet to be four American hyperscalers and the rest of us just bow and scrape and don't host anything of our own ever.

It's also worth noting that these things scale. On a very small service where I am the sole mod, I probably know everyone on a personal basis. It's highly unlikely that anyone is going to post that amongst people they know (unless I'm deliberately hosting a CP service, in which case the whole question is moot).

This is a very different risk profile to one like Facebook which is massive, and also has extensive paid moderation. Alas, the OSA doesn't really make a distinction, even though it should.

You realise the OSA doesn’t just apply to billionaire social media right? It also covers PHPbulletin boards and Other traditional forums, often run by an enthusiast for a community at their own expense in their own time.

12hours is a minimum since… y’know people have jobs and need reasonable opportunity to check their forum for mod notices and action that review. 48hours is more realistic.

Now, if you want to specify an “immediate” action for platforms with more than £100m in global turnover then that’s a different discussion. They should have 24/7 mod teams. But for small sites? Be realistic. If you legislate on the basis all operators have equivalent resource to meta, then the entire internet will be only Facebook/Insta/Tiktok - because everyone will say “I can’t comply with that” and close up shop (the OSA does actually make allowance for “small” services, but doesn’t define what a small service is, and OfCom have refused to provide meaningful definitions, so everyone has to assume they’re in scope for the whole caboodle).

Microsoft throws spox under the bus after Parliament testimony on ICC email kerfuffle

I'm confused by this article. What was the actual correction?

The first four paragraphs all talk about how Milward insisted that the ICC pulled Khan's service in response to US sanctions, not MS directly.¹

Paragraph 5 says ""We have apologized to the Business and Trade Committee for the inaccuracy and asked for the record of the hearing to be corrected." which I presume means they were admitting "yeah, it was totally us", per the sub-headline. But this not explicit in the article body. There's no statement to the effect "Microsoft have now admitted that they pulled the plug on Khan's access, not the ICC". Perhaps MS haven't? Are they still trying to defer blame through pedantic semantics?

At no point have we been told what the correction actually is. Merely that certain statements were inaccurate and are being corrected. Would it be that hard to explicitly say what MS are asking the record to be changed to, not just what it's being changed from! Rather than expecting readers to infer the accurate thing from the inaccurate thing? I know journalists like to add analysis and value. Rehashed press releases are not journalism. But in this case it wouldn't be unreasonable to quote-dump the entire MS letter (if available) so readers can see exactly what was said. Because sometimes the precise wording and nuance is too important to summarise.

1. Which we all knew was bullshit prima facia. Why the fuck would they do that? The US doesn't recognise the authority of the ICC, so why would the ICC respect US sanctions against their own staff?

The UK government isn't spending much taxpayer cash on X

It would be good if they published everything on their own websites with an RSS feed too.

They do, or, y'know Atom - but most RSS libraries from the past 15 years parse Atom equally well. GDS and gov.uk have always been very good at those old-hat, "boring" bits like sitemaps and RSS that React-bros forget when they're building fully-custom blobs of JS.

e.g. The Atom feed for https://www.gov.uk/search/news-and-communications is linked at the bottom of the list of posts, next to the "Get emails" button where you can subscribe to email updates too.

Not only are 19.2 million British citizens registered with X, but 10.8 million families use X as their main news source;

I call bullshit, and want to see her sources.

Maybe at it's peak. And still not monthly active users.

I don't know anyone with an active account. My MP still has an account but it says "Not in use, please use <email>", which prevents squatting - since Musk decided it made sense to allow people to re-register deleted usernames and impersonate defunct accounts.

I wonder how they've distinguished corporate and institutional accounts from personal accounts of citizens? The ratio of public sector and HMG accounts tweeting out to a diminishing pool of actual users gets less favourable by the day. And of course thank's to Musk's dismantling of any sort of assurance or safety, we can't trust how many accounts are bots.

"It is incredibly important, in an age of misinformation and disinformation, that facts are available on the platforms people are using, as opposed to the platforms *WE WISH* people were using, which is why the Government will continue to post organic content on X"

This is a very fair point, and one I made to a sports body in 2015 who were resolutely refusing to engage with the membership, or even publish competition results digitally! The conversation is going on with or without you. Might as well try and set some tone through an official presence. They had the - not entirely unfair view - that it was just people arguing, to which I had to explain that they didn't have to join in the arguing, but being able to post their news and side of events would provide balance.

However.

HMG never had accounts on Truth Social or Parler, despite there being a great deal of misinformation in need of correction on those platforms. There is no obligation to maintain a presence on all platforms, especially on a platform which comprehensively fails every SAFE principle, not through carelessness, but through the designs of it's owner. Hate speech and anti-democratic messaging is not merely rife, but propagated by the platform owner. Illegal activity has not merely been tolerated by the platform, but enacted by it (through GrokAI). And yet we're making excuses why we're holding press conferences in a Nazi Bar?

Most businesses and individuals have jumped ship. The way in which InfoSec Twitter lifted-and-shifted to the infosec.exchange Mastodon server in the space of a week was an impressive bit of social engineering, and demonstrated how a few key accounts at the core of a community web can implement change by declaring in unison that they're all moving platform all at once.

The reality is that most UK Twitter users are politicians and journalists, locked in a Catch-22 where they all have FOMO and can't move until the other one does.

UKGov actually has that level of influence where they could just unilaterally declare tomorrow "Today is our last day on X. As of Monday, all press releases and communications will only be on the relevant website and by Facebook/Instagram. We're not using X any more.".

Assuming this applied to all UKGov accounts, Ministerial accounts and Labour MP accounts¹, that would be that - the journalists would abandon it because the conversation would be gone. For sure, the Tories and Reform would probably witter on for a while, but without anyone to bait, and with diminishing press attention, it would lose traction, or become a far-right echo chamber like Parler where a few tweets from UKGov accounts aren't going to make any difference or convince anyone to change their mind anyway (arguably X is already there, which is why it's galling to see thousands of pounds of taxpayer's money being funnelled to Musk so departments can scream into the void).

What Ruth Anderson is glossing over is that people are using X because UKGov and Ministers are there. She's putting the cart before the horse by claiming they're following the market when actuallly, they're now the market-maker, and are propping up the UK userbase of the personal political project of the world's richest man. They could just... not.

1. Professional accounts of the type "Bob Jones MP". Gov can't stop Bob Jones having a personal account, but they could put out an edict to all MPs to kill their official "MP" accounts.

AI vastly reduced stress of IPv6 migrations in university experiment

The tool includes configuration validation checks and automatic rollback triggers to help prevent errors.

...

The second test, which involved an interactive simulation of a migration, saw cognitive load fall by 72 percent, completion time fall from 9.4 hours to 96 seconds, and firewall configuration errors disappear entirely."

Dare I say, non-AI based migrations might also enjoy fewer firewall config errors if they were suitably equipped with tooling to run validation checks and trigger automatic rollbacks?

I have absolutely no doubt that AI can do something moderately sensible in terms of writing an Ansible playbook (clearly defined structure, limited syntax). But the loud part being said quietly is that they invested time into wrapping some testing and safety around the process. Something which engineers are not always afforded in a manual migration. Shocker, there were fewer errors when the config was forced through validation checks.

I'd also ask, whilst the 96 second completion time is jolly impressive, do the engineers actually understand the final outcome? Sure, it works. If it stops working, do they - having watched some output splurge forth in a minute and a half - have a good enough mental map in their head of how their system is configured, so that they can go fix it? Or do they have to beg the AI to find the fault and fix it?

BBC bumps telly tax to £180 as Netflix lurks with cheaper tiers

Re: Spot the difference

She's been out of front-line news reports for a long time, and it's a dwindling demographic who remember her TV reporting.

Completely off topic, her autobiography "The Kindness of Strangers" is a very interesting view into the early days of local radio and local journalists figuring it all out as they went along (with occasional help from the Army to drag an OB van out the mud with an engineering recovery vehicle!). Worth a read if you find it in a charity shop (the publication date is shown as 2019, but I must have read it 10 years before that, so I don't know if the version currently on sale is an updated edition, or merely a reprint that got a new ISBN, maybe a new introduction or something).