Posts by Sproggit

Moscow slaps Telegram founder on wanted list, Durov responds with one-finger salute

False Flag?

The problem with the internet today is that it is getting much harder to know whether to trust published reports.

But I note that in June of last year, US Media outlet Newsweek published a story suggesting that the Telegram Messenger application had ties to Russia's FSB agency.

Here's a link to the article:-

https://www.newsweek.com/telegram-messenger-russia-fsb-ties-report-2083491

Here is a link to the underlying reporting that formed the basis of the Newsweek story:-

https://istories.media/en/stories/2025/06/10/telegram-fsb/

So which is it? Is Telegram run by someone with a real axe to grind against their country of birth, or is this latest reporting a false flag, designed to lure the unsuspecting to use Telegram because the FSB have now managed to back-door their way in?

And yes, I'm extrapolating a bit there... but, as we all know, in war, the first casualty is always truth.

Not sure I believe any of this - no offense to el Reg, Newsweek, or any of the other authors... It's just too hard to tell these days.

Openreach nudges another 112 exchange areas toward full fiber

That should not be a problem - and certainly not enough to discourage you. OpenRetch are rolling out monomode fibre in order to future-proof the infrastructure for future speed upgrades.

The minimum safe bend radius for monomode depends on the exact specification of the fibre you are deploying. Standard G.652D requires just 30mm radius at rest, while bend-insensitive types such as G.657 will allow a bend radius as low as 10mm (A1) or 7.5mm (A2).

You don't mention further detail regarding these five 90-degree bends, but if I could be so bold as to offer you a tiny little bit of advice from real-world practical experience...

1. If your 90-degree bend happens to be on an outside wall, then if you lay the fiber level with the mortar between two rows of bricks, you can actually "cheat" slightly and chisel away a small amount of mortar where your corners are located, to enlarge the bend radius and soften the curve slightly. [ Yes, you can get a pretty tight radius with current specification monomode, but if you don't push to that tightness, you'll reduce the risk and hopefully get a more resilient connection].

2. If your 90-degree bends happen to be on an inside wall [inside corner] then my advice, if you can tolerate it, would be to either 3D print - or find a friend with a 3D printer - and fabricate a mini "cable corner piece". Basically all you need is something that looks a bit like a shelf bracket laid horizontally. Make the distance from the corner to the furthest ends say 10cm or so. Fill in the corner with a flat piece of printed plastic, but then scribe a quarter-circle on the inside of your 2 edge pieces. You can 3D print a snap-in lid to keep it neat and tidy. You can choose a filament colour that matches your wall... You can fit the things on top of a skirting board, or up at ceiling height. And if you 3D print a quarter-circle groove on the inside of the bracket, your fibre won't kink or bunch or move once you lay it in place.

Finally, if you're still keen to proceed but can't see a way forward, a small specialist [data] cabling company might be worth a try. Chances are that a one-off might be prohibitively expensive, but if you can find one local to you, you might be able to get one of the guys to moonlight the job. Or you could try one of the local "Sky" installers... same sort of skills required.

Re: Only available for Single-Dwelling Structures

Looks like the process has changed - again - recently. Here's a link to the latest submission site:-

https://www.openreach.com/help-and-support/obtaining-wayleaves

Re: Only available for Single-Dwelling Structures

Do you have power in your loft? Could you bring the fibre connection in to the property via your soffits and terminate it there? You could then if necessary go back outside the building with some exterior-rated CAT7 and route that to your preferred entry point?

[Caveat - if your loft has only a single electrical feed for the traditional naked-bulb light, maybe not. You can't connect appliances to a lighting circuit.]

Failing that, you could run an exterior mains cable up to your loft from the back of a wall socket lower down the building [take the socket out of the wall and make a hole from the exterior to the back of the socket to add the spur]... With a mains supply run up the exterior and through your soffits to feed a couple of sockets in your loft, you'd then be able to provide power to the fibre transceiver - the wall box that OpenRetch need to fit on the inside wall of your property. The private side of that connection is presented as RJ45, so then you could get a few metres of "exterior" shielded CAT7 and route that back down the exterior wall [parallel with but not too close to your mains feed, unless both are well shielded] and introduce the RJ45 at a location of your choosing.

Re: Only available for Single-Dwelling Structures

That's entirely fair.

What that approach DOESN'T entitle OpenRetch to do, however, is LIE to people with claims that they have "contacted the landlord and are awaiting permission to proceed with cabling up the building", which is what they told me. It also doesn't explain how my fellow director was able to go to them with full, documented approval to proceed - *and they refused anyway*.

The point here is that there is some kind of performance criteria or contract stipulation for them to make fibre available for a certain number of premises... They have decided that the time required to figure out what is needed for an MDU degrades their time-per-dwelling performance statistics, so they are treating anyone living in an MDU as a second-class citizen just so their directors can get their bonuses.

Re: Only available for Single-Dwelling Structures

On November 20th, 2025, Bilal Akhtar of Openreach wrote to me concerning a UK property that I own; it is part of a leasehold apartment building where the freehold is owned by the leaseholders through a simple company structure.

The specific reference ID quoted to me by Bilal was QIS750618.

The claim put forward by Bilal was that Openreach had contacted the landlord for the property in question and requested permission to access the building in order to be able to deploy fibre cables to the 37 properties within the structure but stated that the landlord had declined to reply.

I *am* the landlord for the property, because I am the chairman of the Board of Directors for the Limited Company that owns the freehold [with each leaseholder owning one equal share. So I knew for a fact that Bilal was flat-out lying to me, because no attempt to contact me or the administrator/company secretary had been made. No paperwork had been left when Openreach deployed fibre to the road outside and no Openreach employee spoke with any resident.

After pointing out to Bilal that not only was the information he provided materially false, it was probably intentionally misleading. I asked him in writing what steps were needed to be taken to trigger the necessary on-site survey and get a fibre team to install the service to the property. I was directed to a non-advertised page on the OpenReach web site that does not appear to be linked from any default or public landing page (meaning, a bit more explicitly, that if you do not know the page exists and have not been given the URL at which it resides, there's no chance you will find it).

I completed all of the required information on November 20th last year and as of today - 28th July 2026, I am still waiting for a response.

I am part of an active community of landlords and after I shared information concerning my experiences, I was contacted by the chair of another apartment building in the same county. His experiences were even more interesting. He happened to see the Openreach team when they came to his street and began to install the fibre to individual detached freehold homes, so he walked down the road and spoke to the men concerned. They explained an essentially identical set of circumstances. He obtained a link to the same information portal and had everything completed and submitted well before Openreach had made it down the road to his property. He went out and spoke to the installers, provided them with paper copies of all the material he had submitted and asked them to ensure that they completed the installation to the roughly 50% of the residents of the block where he lived and who had requested connection. The roadside team immediately refused - perhaps understandable, since they would be following instructions - so he asked for a contact number for the manager of the installation teams.

I don't know the detailed specifics of that conversation and heard it from my fellow director; as he explained the story, he was told that a decision had been taken at the highest levels of Openreach to only deploy FTTP to stand-alone dwellings, but there was some confusion as to the motive. At first the Openreach person said something about the terms of the agreement they were deploying to had some kind of stipulation that a certain percentage of properties had to be offered connectivity by certain date milestones, but then apparently back-tracked. I don't know the specifics of what was originally claimed or where the retracted/revised statement ended up.

Bottom line is that Openreach will tell you if you ask that there is a request process, if you live in an MDU, to have their installers return and provide a connection. That process is - as far as I can tell - non-existent fiction. The entire process being followed by Openreach is to run their cable past as many dwellings as possible, but they are ignoring "MDUs" because of the perceived complications of getting landlord permission and of any potential complications caused by getting access to interior ducts within the building. From other conversations with specialists, I understand that this concern is highly relevant when the apartment building concerned is a tall multi-storey, since there is no practical choice but to use internal ducts to reach higher floors. However, for other properties, such as though with just 2 levels [ground and first floor] and where no interior ducts exist, the only possible solution is for Openreach to run the fibre up the side of the building to the first floor apartment and then simply go through the wall.

The issue seems to be that nobody in Openreach management is either aware [or possibly cares] that not all MDU's are 20-storey buildings constructed in the 1960s - and as a result Openreach have zero interest in applying common sense or customer service to a rollout being funded at least in part by UK tax-payers through the Project Gigabit initiative.

Quite a disgusting level of service from one of the UK's largest critical infrastructure providers.

Only available for Single-Dwelling Structures

Bear in mind that Openretch [sic] have an internal company policy - this is not mandated anywhere in Project Gigabit, which is sponsoring a lot of the costs for the FTTP deployment - that Openreach will NOT connect up any property that it deems to be an "MDU".

That's a "Multiple Dwelling Unit" to you and me.

Which means that if you live in a block of flats, you won't be given FTTP and you won't even be given a date for when it *might* become available.

Friend of mine had full fibre installed down his road over a year ago - but because he lives in a large older house that's been converted to 2 flats, Openreach have refused to connect either property. All it needs is to drill a small hole in a wall, but they are refusing.

Meanwhile, BT cut the original 80/20 FTTC broadband down to 67/18 to cope with the extra load caused by COVID... but now they've massively increased their network capacity, they won't even restore copper customers to the previous 80/20 rates.

Bastards, the lot of 'em.

Hugging Face rebuilt a third of its infrastructure after OpenAI agents ran amok

Cover the Cost

The very least that OpenAI should do - right now - is publicly state that they will cover the full cost of all work that Hugging Face have to do to tidy up this marketing-ploy-gone-wrong.

Not because I'm demanding some sort of altruistic behavior from them.

But because, if they don't, other companies are going to be rightly cautious of deploying OpenAI models in security capacities, out of concern that the model will simply "go rogue" and start to perform *Criminal Acts* against other organizations.

https://www.justice.gov/jm/jm-9-48000-computer-fraud

https://www.legislation.gov.uk/ukpga/1990/18/contents

etc.

The marketing ploy for OpenAI would raise some concerns: "Deploy our Model for your cybersecurity needs - and it will break the law, so you don't need to!"

Personally, I would like to see the DOJ open a case against OpenAI, under the Federal Statute, then take it all the way to a prosecution [if the facts merit]. Not because I think OpenAI need to be singled out for special or particularly harsh punishment. I'd be happy with a token fine.

I think this is needed as a "warning shot" across the bows of all AI model developers and a reminder that if you use technology to break the law, it's still you that's breaking the law. The companies clearly lack either a sufficiently strong moral compass and/or basic common sense.

We should look upon this incident as an opportunity to prevent future repeats - and a big part of that is issuing an unmistakably clear warning.

Uncle Sam needs you to fight for 6G leadership and security, lest Beijing get there first

Let me get this Right...

... President Trump has attacked...

NATO members, for failing to provide military support for his war against Iran (despite the fact that the US didn't invoke Article 5)...

Allies the world over, with unprovoked and unjustified tariffs, which he then diverts from the Treasury general fund to pay for e.g. expansion of ICE...

The European Union - both collectively and individually...

,,, and he continues to support the US veto of International Tax Agreements that would curb the excessive "profit shopping" that huge US multinationals use to spirit billions of dollars out of the economies of other nations without paying tax...

And he wants all those nations that he has been attacking and harming to join him and develop 6G so that the US can spy on allies rather than allowing China to spy on them?

Yes, yes, I can see how this would be an extremely popular move.

Snort.

Leading AI models (even Grok) are all a bunch of leftist punks

Depends on where you define "The Center"

I need to put a bit more time in reviewing the data this article reports on... but after an extremely hasty review, the one thing that stands out is the question of where do you define the baseline.

If you look at the history of politics in the US and UK since, say the Reagan/Thatcher era, then what has in fact happened is that the political baseline has move significantly and steadily to the right. For example, after Tony Blair led the Labour Party to government in 1997, he went on to Privatize NATS - National Air Traffic Services [in 2001], converted Railtrack from a failed private enterprise in to a not-for-dividend private company [rather than re-nationalize it] and sold off the government's remaining "golden shares" in privatized utility companies [such as water and gas] which would have given the government more control over their pricing structures and compensations for senior executives. Finally, he removed the remaining state vetoes over foreign and private takeovers of utility companies.

None of those actions would be considered traditional "left of center" policies, yet they were all delivered by the UK's supposed left-of-center Labour Party.

Interestingly, all of the above elements are linked to the then "Clause 4" of the Labour Party manifesto - part of the Labour Party's constitution and written in 1918 - which committed the party to common ownership of public services. Blair drove through that change in 1995, before Labour won office in 1997.

I offer all of the above not to critique or express views "pro" or "con" any political viewpoint, but to underscore the fact that our definition of "centrist policy" has been moving steadily to the right for at least the last 40 years - and in some periods, moving *rapidly* to the right.

It's also worth making the point that AI models have to be trained and that training material has to come from somewhere [apparently from just scanning copyright materials, cough, cough], the premise here would suggest that the developers of the models themselves in fact consciously and selectively fed training data to those models. Had that been the case, I can't help but think we'd have heard about it before now...

If we try and combine these two data points, I'd suggest that it is more than entirely possible - I'd go as far as to suggest "quite likely" that the models listed in this article have been fed data from a variety of public sources; that the data they have been fed dates back across the "digital age" (i.e. the last 40 years or so) and that in the same time period, western nations have observed a steady swing to the right in national politics.

If you combine those elements, then it seems reasonable - even likely - that the "left wing bias" suggested here could in fact be more accurately viewed as a reflection or consequence of the steady right-ward trend in western politics over the last four decades.

None of this is right or wrong - I'm just looking for other explanations. The reason for doing that is because in order for the article to have some foundation in fact, it would be necessary for all the companies and models identified to have [independently or in some grand conspiracy] have chosen to train their models with left-wing bias. That just seems to be implausible - for one thing, the owners of the companies producing these models are clearly all out-and-out capitalists.

In fact the data shown - especially that graph - illustrates the one obvious outlier here: Grok. That model absolutely seems to have been trained with conscious bias... and the result of that is clear to see.

FOSS smashed one Microsoft monopoly. After 20 years of failure, it's time to smash another

Re: Remember the Approval Process for OOXML?

Fair...

Re: Remember the Approval Process for OOXML?

"Not even MS can have that many national HQs."

True, but what Munich's experience teaches us is simply that Microsoft are more than willing to do anything which is not outright illegal in order to get the outcome they want.

Microsoft are a currently valued at $2.84 Trillion, but have been as high as $3.85 Trillion (Yahoo Finance). So they certainly have pockets deep enough.

Would Microsoft also be willing to bend or break the law? Maybe history gets to be the judge...

Re: What's the bet ...

98% might be a *bit* of a stretch... but I definitely think you're on the right track.

It might be a little closer to the truth to say that the 80/20 rule applies - that 80% of documents written in word use 20% or less of Word's features and functionality.

Remember the Approval Process for OOXML?

Rupert wrote:

"That means the tooling wrapped around the spec has to account for proprietary dependencies that get smuggled in, such as fonts. It also means actively and continuously tracking the ground truth of Microsoft's evolving products and services."

It's worth noting a little bit of the history around the adoption of OOXML and it's acceptance by ECMA and the more heated battles when it became the ISO/IE 29500 Standard in 2008.

In September 2005, the Commonwealth of Massachusetts announced that it would migrate all documents to an open document standard (specifically declaring the intent to move to ODF). Microsoft had been flat-out refusing requests from large customers to do this - and they reversed course just weeks after the Commonwealth made their policy announcement.

Microsoft responded by writing OOXML as a "Standard" ... and depending on which version you consider, the resultant specification spans between 6,000 and 7,229 pages. By comparison, ODF 1.4 covers a total of 1,057 pages, split across 4 discrete parts:-

Part 1 - Introduction (8 Pages)

Part 2 - Packages (34 Pages)

Part 3 - Schema (802 Pages)

Part 4 - Recalculated Formula (213 Pages)

Within the OOXML Standard itself there are many "BLOB" structures, which Microsoft initially claimed were required to support legacy file formats and structures from older versions of their Office products, but which experience has shown actually build reliance (lock-in) on native Windows-specific API elements, system-level font-metrics and graphic-rendering logic (such as GDI coordinates, for example).

None of this is by accident - Microsoft produced something that was a "standard in name only" and then seemed to have relied upon the creation of a document specification that spanned 7,000 pages in order to bamboozle reviewers and obfuscate that what they had done was create a lock-in dependency on proprietary API's and rendering logic provided by Windows. Microsoft did the exact same thing when governments demanded that Internet Explorer be split from Windows - they simply moved the browser's logic in to OS-level DLLs as API calls and claimed that the browser was thus inseperable from the OS.

However, there's a deeper concern hidden in those 7,000 pages of weeds... and it concerns integrated macro logic. Unless you're doing something very (*very*) simplistic with your office suite, your spreadsheet is going to include more complex structures, including formulae, graphs, macros and charts. These have proven to be a minefield for portability, because Microsoft didn't even pretend to support these outside those proprietary BLOBs - and the relevant Standards Committees turned a blind eye. Translation of VBA, in particular, seems to still be a struggle [if attempted at all] and that might be the reason why there is no real alternative to Access. The closest still seems to be LibreOffice Base, but that product can't handle the forms or macros written in Access.

Only one thing is going to change this - which is for a nation somewhere to have national government agencies standardize on LibreOffice.

Do you remember what happened when Munich, Germany made the declaration that they would move approx 15,000 municipal desktop PCs to Debian and OpenOffice? It started out so well:-

https://www.omgubuntu.co.uk/2014/07/munich-city-saves-millions-going-open-source

Then Microsoft went on the offensive:-

https://www.omgubuntu.co.uk/2014/08/munich-city-linux-switching-back-windows

What is less obvious from the historic record about the goings-on with Munich is the way that Microsoft played dirty [because they could afford to]. So, for example, when it became apparent that Munich might be the first domino to fall, they immediately announced plans to relocate their German HQ to Munich - and completed that within 2 years. Just a coincidence, nothing to worry about.

That history is going to rinse-repeat unless or until a much stronger insistence on a 100% open and transparent standard is set and enforced. Which is a tall order, especially given macros and Access and the current lack of equivalence in the open source world.

On the other hand, the sophistication and polish of the FOSS ecosystem has transformed beyond all recognition over the last 10 years. Over the same time period, the only significant change from Microsoft is that their products have become more bloated and slower, making them more expensive to run.

Maybe the time is right for a lean, clean competitor to come in and shake things up - as long as there is a 100% level playing field, that is.

The last Space Shuttle returned to Earth 15 years ago

Bear in mind - and this is an important distinction - that SLS was *not* a fixed-price contract, but a cost-plus contract - which meant that NASA agreed to pay the consortium who built SLS the cost of designing and building the vehicle, plus a "commercial profit".

Which goes a long, long way to explain why the project came in so many years late and so many billions over budget.

Worth noting that this isn't NASA being stupid either - more accurately to say that this is Washington D.C. getting their slice of corruption added to the bottom line.

Microsoft seeks Supreme Court lifeline in pre-owned license battle

The Magic Trick

It would be interested to ask a specialist in the various elements of the law - contracts, assets, licenses - for their views on this.

But it seems to me that what Microsoft are trying to argue here is that [their] software is "a whole new thing" [to quote Chris Harris]... such that the basic principles of "offering for sale" and "purchasing" no longer apply. They want the courts and the world to believe that software is somehow "magic", that it represents a new and entirely unique construct that means that it is somehow excused from the basic right of resale that exists with all other property.

Interestingly [but not for Microsoft], when corporations write up their balance sheets for legally-mandated accounting purposes, purposes which have to adhere to e.g. GAAP or IFRS accounting practices, we see that software licenses are listed in the asset column. Which means that since the first piece of commercial software was purchased by a company, everyone involved believed that software was an asset, *purchased* by the customer for their use. Now Microsoft are trying to rewrite that history and claim otherwise.

I'm sure it's no coincidence that outside of software and media, the closest thing we find today with a legal parallel to this is the concept of "leasehold properties" - and it's no coincidence that the UK government has recognized that leasehold is being systematically and shockingly abused by landlords.

That's not a coincidence.

Hopefully Microsoft will be told where to go.

SpaceX just about nails Starship test flight 13

Re: You can say what you like about Elon...

One of the key differentiators that we've seen from SpaceX since the early F9 flights has been their willingness to learn iteratively.

SpaceX recognized that the hardest part of what they were attempting to do was to re-land their first stage boosters so that they could be re-used... But, the moment they got to the point where they were confident they could get a payload to orbit, they started to put commercial loads in space. This not only meant that they were funding the cost of the R&D needed to accomplish the much more difficult landings, but that they could leverage every single paid-for launch as an experimental landing.

So what if an F9 booster had a RUD on the landing pad? They got critically important data.

If there has been a single defining piece of genius innovation from SpaceX, then that was it. Whether you're looking at ULA or Blue Origin, they have stuck with the traditional approaches... and they are a long way behind. A long way? Yes - in 2025, SpaceX accounted for more than half of all commercial launches to space, *worldwide*.

It's abundantly clear that lessons learned from F9 haven't scaled smoothly up to starship - the complexity of the vehicle, the kinetic energy and forces involved, all of these things will take significant effort to overcome. It's also worth noting that one critical piece in the Starship/Super Heavy infrastructure is brand new - and that's because SpaceX never attempted to recover any F9 second stage. With Starship and "100%" re-use SpaceX are in "undiscovered country".

The fact that this most recent launch included actual working Starlink satellites suggests a level of confidence with the portion of the test profile that includes getting starship to an orbit suitable for satellite release - and *that*, I suspect, might herald a tipping point. If SpaceX can demonstrate the reliability of that part of the mission profile a few more times, I think they will be in a position to attract paying customers for the launch process ... That, in turn, would mean that Starship/Super-Heavy starts to self-fund, rather than each launch being a $100Million+ expense on the company's bottom line.

We'd be crazy to try and predict the outcome of future launches, but if history is anything to go by, getting Starship to the point where paying customers will be confident in using it might be the biggest leap forward of all...

Ireland stalls €1B Microsoft tender amid digital sovereignty questions

The Value of Ireland

There are two main reasons that Ireland is so popular with US multinational companies.

The first is that Ireland speaks English - albeit with a lovely accent - and that means that they can converse with their Corporate Overlords in the US without having to parse everything through a second language... It means that US-based lawyers can read contracts and legal briefs without having to pay and wait for translation. There is a significant advantage to US companies through having a foothold in the Irish Republic.

In case you read that and wonder why those companies would not simply use the UK... it's because the EU tried to get a bit "cute" with big US companies in response to Brexit - for example they required that financial institutions wanting to transact Euro business must have an EU *headquarters"... There was definitely a hope for cities like Paris and Frankfurt that they could attract a significant amount of investment and high-paying jobs this way - and those nations were gutted when American companies chose to go to the Irish Republic.

The second reason is that Ireland have been in real trouble with their finances and were desperate for any way to bring employment to the country - so they set up tax schemes that were extremely favorable to US multi-nationals. A lot of that has since been leveled out - the other EU nations took umbrage - but by then the US companies had already put their roots down in Ireland and weren't about to incur more expense moving just to satisfy bureaucrats.

Support newbie figured out how to avoid all-nighters, and his colleagues hated him for it

From the back page of Computer Weekly...

Back in the 19980-1990s - more specifically at some point between 1986 and 1998 - Computer Weekly ran a very similar story to this on the back page of one edition of their magazine.

The story as printed concerned a bit of hardware - pretty sure it was a laser printer - that had a habit of failing "every now and then", as a result of which a sheet or two of paper would get jammed up in the works and the operator would call a technician from a central IT Team over to fix it.

This particular operator, however, was single, female and stunning, so the attendant engineer clearly didn't mind having to make repeat visits...

Except that in the case of this particular tekkie, things unraveled when they took a vacation. The central IT Team were a bit stretched, so they hired in an engineer from one of their main suppliers to act as a "backup"... and it was this individual who attended the failing laser printer. They spotted the fault - an internal part had been "over-adjusted" and would fail on a particular type of heavier paper, used one day a week for specific correspondence - and made an adjustment that resulted in a permanent fix.

They also reported what they'd found to management...

As the anecdote was told via the back page of CW, the engineer who had done nothing but remove the jammed paper without fixing the actual problem was reassigned, and the guest engineer ended up dating the very attractive user of the printer.

As it happened, we had 4-5 different hardware engineers from a couple of companies that looked after our kit at that time, one of whom was very much a "ladies man" ... so we photocopy/enlarged the article and pinned on the notice board by his desk. He took it down - of course - but that's what photocopiers were made for... and because as we told him, he was "one of the team", we made sure that a fresh copy of the article was waiting for him every time he came on site.

Google meets the neighbors and gets both barrels over its new UK datacenter

Re: Suggested Condition

Three observations in response...

First is - the lack of "inward investment" that you rightly point out would be a consideration might be significantly offset by tax receipts from companies that decided to stay.

Second is - the foreign multinationals have clearly remained because they see a market here - so if they decided to quit because they didn't like a level taxation playing field - well, that would leave an established market open for UK companies...

Third is, "half a loaf is better than no bread" - if a commercial entity had to choose between making a profit and paying some tax on it on the one hand... or making no profit on the other... which do you think they would choose?

I've heard the "inward investment" story done to death... but the UK is the 5th largest economy in the world. Do you really think multinationals operating in the UK would simply up and leave in a fit of pique? Put another way, shareholders have ousted CEOs and/or file lawsuits for less. Not going to happen.

Suggested Condition

How about...

The government enacts a law that says that if any multinational corporation wants to conduct business in the UK, they must form a wholly-owned subsidiary and all UK-related business must be conducted in the UK, under UK law. The provisions of this new UK law would further state that taxation of said company profits would only permit *UK* expense deductions and NOT cross-border licensing deals.

In other words, that would prevent Google - and others (Starbucks, Facebook, Amazon, Microsoft, et al) from spiriting away their profits and bleeding the UK Exchequer of legitimate tax revenue.

Because - let's not forget - anything that corporations don't pay in "corporation tax", we will end up paying in "Income Tax".

It is way past time that the government closed all these tax loopholes. Way pays time.

And please don't tell me that this isn't possible. If we have to pay 20% VAT for goods and services that we buy off e.g. Amazon, then the purchase takes place in the UK, under UK tax law. So Amazon can damned well pay corporation tax in the UK as well. This one change would put *billions* a year back in to the UK Exchequer...

Fujitsu joins £14.9B UK framework despite public sector bid freeze

Re: and how much compensation for Horizon have they paid so far?

Your question gives us at least one - and obvious - precondition before the UK government should ever consider Fujitsu for a future contract - which is that all compensation claims by injured parties in the Post Office fiasco must be paid - in FULL before they are even allowed to bid for another contract.

Just remember - an official inquiry found that as many as 13 people may have taken their own lives as a direct result of Fujitsu's lies.

If there were actual *justice*, accountable managers of that company would be serving serious prison time. As an absolutely BARE MINIMUM, they ought to be paying 7-figure restitution to each injured party.

And that would be letting them off lightly.

EU telcos ask: Huawei going to afford to replace Chinese equipment?

The Three Hegemonies

Seems to me that the world is trending towards 3 hegemonic doctrines: The United States, China and Europe.

I appreciate that's an oversimplification, but the concerted economic power required to survive in an adversarial climate means that an isolated Russia and the Oceanic nations don't stand a chance.

Whether smaller nations like it or not, they are going to have to build alliances with one of those three power centers. The problem is that right now, technology such as this are only really available from two of the three hegemonic powers - the US and China.

It's possible that Europe could mitigate the risks of adopting technology from one of the other two powers with some robust ring-fencing and aggressive monitoring of such untrustworthy devices, but when you think about it, something like a 5G router is so ridiculously vulnerable to abuse, simply because all you'd need to do to compromise it would be to send it a carefully-crafted SMS message.

The EU may not like the obvious conclusion... but "the only way to be sure" might in fact be for the EU to develop their own technology industry, where they control the entire development cycle, start to finish. Is that a stretch? Well, high-end plants for CPU's and RAM cost in the order of billions to manufacture... so it won't be easy. But it's definitely possible.

Re: There should be a rule

"Not so much as a cable."

Part of Edward Snowden's disclosures in 2014 included details of "Cottonmouth"... an NSA-CIA hardware approach that was based on the then new USB technology - and in particular USB printer cables, among others. The TLAs involved had "special" USB cables manufactured that included a hidden transmitter that transformed the cables in to short-range radio antennas. The range was limited, but it meant that if your computer had one of these cables connected, then a "covert agent" could sit in their vehicle in the car park in front of your building and remotely connect to your PC, without your knowledge.

The software that supported this was bundled in to Windows 95 by default [which was the Windows OS version that introduced USB support].

See here: https://en.wikipedia.org/wiki/ANT_catalog

So: you're not wrong.

Intel working to ‘leapfrog’ ARM and AMD designs, says CEO Lip-Bu Tan

Conflicting Strategies

This strikes me as an acknowledgement that Intel find themselves in a uniquely challenging position.

For decades now they have battled AMD at the "top end" with regards to delivering powerful server and workstation processes, but the 15th Gen family (Ultra 7 265 and Ultra 9 285c etc,) proved to be *slower* than their 14th Gen counterparts in critical tests. [ Qualification - the differences are likely so minor that it's unlikely you'd notice - for example in gaming frame rates dropped by around 1-3% on average - but the fact than an entirely new micro-architecture (the Lion Cove P-cores and the Skymont E-cores) delivered poorer performance was kind of a marker that things might not be going entirely as expected ]. Meanwhile, AMD have continued to develop and enhance their flagship processors and continue to hold the top spot in sheer performance.

At the other end of the street, Intel's "T" family of processors (the parts with 35W TDP in each major generation) have been hit by a "double whammy". First, in order to deliver measurable improvements in performance, the T processors have now got such insane "turbo" modes (up to 112W) that they're rather making a mockery of the "low power" claims. Second - and perhaps more important, we're starting to see more and more mainstream OS ports across to ARM and ARM-like [Apple] Silicon. Microsoft, in addition to Intel, use the Qualcomm Snapdragon in their latest Surface and Surface pro models. Framework - the user-repairable laptop folk - now have an ARM based motherboard for their Framework 13 laptop.

Let's be fair - the ARM-on-the-desktop has a way to go before it will be able to compete directly with either Intel or ARM... but on the other hand, we really don't need 5GHz processors with insane core counts to run 99% of typical desktop requirements.

And while we're in the neighborhood, the obvious response to the previous statement is: well OK, that's fine - but what about performance users - people who use graphic design software, or video editors, or want access to complex multimedia? Here Intel took us in a weird direction when they killed off the HEDT families - the chips with more PCIe lanes and significant overclocking headroom - and switched to much more limited configurations. (Pretty sure that was in order to achieve higher base clocks).

Intel are besieged on all fronts: AMD has them beat at the top end; ARM has them beat at the bottom end; and in the middle they have been making weird design decisions that are slowing down [or failing to improve speeds] for processors between generations.

I wish them well, I really do - I want AMD and ARM to have serious competition...

But it's been sobering to see how quickly ChipZilla has shrunk to WimpZilla.

Year-long Russian attacks infect users as soon as they look at an email

Re: 'collaboration suite'

Obligatory Reference:

https://www.youtube.com/watch?v=q0POXW4V1_k&t=18s

You're welcome.

China advances plans for national single-stack IPv6 network, and its own surveillance-friendly version of the protocol

Re: Every time I see <whatever>+

Mathematically, you might even argue that this is

1 / <whatever>

given that the perversion of the original basically inverts something that started out as being useful and completely undermines that practicality, until it does the opposite of the useful thing it was designed to do. You know, like public utility regulators, competition and market watchdogs, consumer protection agencies, environmental protection agencies... OK, gonna stop there - too depressing.

Re: Let's Not Forget BGP Abuses

Thanks - fair challenge - have an upvote...

But in this instance I'm going to have to counter that this is one of those scenarios where "less is more"....

Let's Not Forget BGP Abuses

One of the other dimensions we might need/want to think about with regards to IPv6 concerns the fact that it is also vulnerable to BGP (Border Gateway Protocol) abuses in the same way as IPv4.

This, if you're unfamiliar, is the wheeze that Russia were caught doing a while back, where they deliberately misrepresent the "cost of routing" traffic across Russia's portion of the internet, advertising to their peer connections with other countries that the "cost of routing" traffic through Russia [for certain network addresses] was suddenly "cheaper" than the actually most sensible path.

This had the effect of redirecting massive blocks of internet traffic, for controlled periods of time, so that the content flowed across Russian-controlled connections before being handed on to destination networks. So to give you a terrible and contrived example for illustration, Russia manipulated their border gateway routers to tell internet traffic in Ukraine that the quickest/fastest/shortest/lowest cost path to Finland was via Russia, even if that wasn't actually true...

The reason Russia was doing this was generally thought to be that they were literally "bulk collecting" traffic for local storage. This would allow them the luxury of time to brute-force encryption on any of that traffic, or to wait for the encryption scheme to be broken (by e.g. quantum techniques being applied to non-quantum-safe encrypted traffic), so that they could access the contents at a later date.

One of the under-reported issues that we have with the networking layer of the modern world is that neither IPv4 nor IPv6 were designed for a world in which participants in the supra-national infrastructure layer would become so abusive.

Similar sorts of concerns look as though they will soon also be applicable to China - if that is not already the case, of course. And to be balanced it's worth noting that western governments are not far behind, albeit with slightly different approaches. The UK, for example, is forcing internet providers to keep internet communications data under the Investigatory Powers Act of 2016, which requires ISPs to keep "Internet Connection Records", or ICRs. These log all web site domains you visit, apps you connect to, timestamps and IP addresses - and requiring this data is kept for 12 months. It's worth noting, however, that there would be little to stop the UK or any other government from:-

1. Having a list of "blacklisted" web sites and demanding more detailed data capture for any access to those endpoints

2. Capturing details of all UK citizens using VPNs

Citizens of western nations are kidding themselves if they believe that the surveillance to which they are subject is less than China's current approach.

UK pumps £708 million into its future fighter jet

Re: What is this for?

As a rule-of-thumb, I would not generally be in favor of committing a military to any operation not based on repelling or deterring a direct attack on the nation to which the military belongs.

Sadly, history teaches us that the world is simply not that cut-and-dried.

For a start, Article 5 of the NATO Treaty (the principle of Collective Defense) obligates all signatory nations to go to the defense of any member nation that is attacked. (OK, in fairness, the article doesn't mandate military participation - a response could be economic sanctions* - but depending on the circumstance that might be a moot point).

Which I think means that the premise for your post is undermined at the get-go, if only because it is slightly mis-framed. It doesn't have to be a military threat against the UK, it could be a military threat against any other NATO member - say Finland - and the UK may become involved by default.

* One of the aspects of the recent US/Israeli strikes against Iran that I thought was under-reported was analysis of President Trump's displeasure that other NATO members did not provide the US with the level or sort of support that he expected. But Article 5 is clear: it does not commit member nations to either a specific course of action, or, indeed, to simply agreeing with specific requests from the nation that invokes it. And just for the record, the United States did not invoke Article 5 prior to opening hostilities with Iran, a small detail which meant that none were under any obligations to do *anything* in support of that conflict.

The big difference about a military strategy based around drones and one based around aircraft is that Ukraine are building combat drones for a few hundred or low thousands of dollars each, using commercial parts [like batteries, motors and propellers]. The Tempest program is going to cost billions in development and individual aircraft are going to cost tens of millions each to buy - and the same again to maintain over a service life of e.g. 30-40 years. Ukraine's approach allows them to iterate designs cheaply and to innovate rapidly. Literally they can take the results of a sortie back to a manufacturer and have a modified design ready to test in a week or two. Off-the-shelf parts, capable CAD and 3D printing has transformed the lifecycle process. That's simply not something that can be done with a full-sized aircraft, because of the cost of a single unit - and the design, engineering, tooling and assembly costs for each unit built.

What I find interesting about this announcement is that it is setting a clear direction towards support for a domestic industrial base, as opposed to simply "buy American".

*That* strikes me as the most interesting part of this announcement, by far - and one driven by the change in relationship that the US is showing towards former allies. It seems certainly possible, potentially likely, that Washington's recent announcements have heralded the end of trans-Atlantic cooperation on defense procurement. If President Trump's remonstrations about defense spending from NATO allies was meant to encourage them to buy more US weaponry, well, it seems to be having the opposite effect.

It's also interesting to see that this is being pegged as a British development and not pan-European. The Eurofighter Typhoon is a capable bit of kit and no mistake - supersonic without reheat (afterburners), exceptional performance, good endurance and seemingly a well-balanced set of compromises to produce a multi-role aircraft. But, lest we forget, the design and development was a tortuous process filled with delays, changes and cost over-runs. The French got so fed up with the Eurofighter bickering that after the Eurofighter program was underway, they commissioned Dassault to produce a domestic-only twin-engine, canard-wing multi-role fighter. The result was Rafale, a Gen 4.5 omnirole aircraft, which was flying before Typhoon and which is very well regarded.

So it can be done right.

Whether or not Britain's omnishambles of a government can deliver... well, that remains to be seen.

Stats watchdog prescribes stronger caveats for NHS Palantir claims

The COVID Angle?

Lindsay wrote another excellent piece on this debacle of a story back in the middle of June - you can read it here:-

https://www.theregister.com/public-sector/2026/06/16/nhs-palantir-claims-face-scrutiny-after-data-suggests-uneven-results/5256198

After reading this update I dug back through the older Reg articles to see how the story had changed in the last 5 weeks or so... but then spotted something of possible interest in that earlier story that I hadn't registered first time around.

Lindsay wrote, "Palantir won the £330 million contract to provide the NHS Federated Data Platform (FDP), which the UK government said was vital to improving NHS productivity and recovering from the long waiting lists for elective care caused by the COVID-19 pandemic."

I don't for one minute buy the idea that you need the FDP to recover from the long waiting lists for elective care caused by COVID. You just need to return to pre-COVID levels of elective care - then maybe add in some temporary additional resource to help you work through the backlog.

But then it occurred to me - what if the claims being made by Palantir had played a bit fast-and-loose with the start and end dates they used to compare their "Before" and "After" performance boost?

That's relevant, because if the dates used for "Before" had absolutely *any* overlap with the period where elective treatments were suspended by the NHS out of concerns for patient safety [during the COVID pandemic] and the need for the NHS to divert absolutely 100% of their resources to handling the unprecedented crisis that COVID caused... then the net effect for Palantir would be to show a massive uptick in throughput - but that increase would be as a result of the deliberate blocks on elective treatments being released.

Just the sort of thing the Palantir - and government officials taking back-handers for letting this go through as a no-bid contract - would be happy to do.

One thing is becoming increasingly clear to citizens of all western nations - the corruption of the mechanisms of government in democracies are now every bit as bad - if not worse - than those of dictatorships. Which is NOT a suggestion that if authoritarianism is less corrupt it's preferable, far from it.

It's the observation that unless or until we get a check-and-balance against government corruption, this won't just continue, it will get worse.

Google Cloud is killing it

Maybe...

... that one of the reasons that Google Cloud is "killing it" is because they only ever pay a "token" Corporation Tax.

For example, Business & Accountancy Daily reported in 2020 that Google paid just £44 million in Corporation Tax on £1.6 billion in UK revenue.

https://www.accountancydaily.co/google-revenue-hits-ps16bn-pays-only-ps44m-corporation-tax

That works out at a tax rate of 0.275%.

How do they do this? Their tax charges are only applied to "strictly localized net profit figures"... Put another way, that means if Google Cloud claims that the service they're selling a UK client is "served" from somewhere in the US, then, well, that is not "strictly localized" in the UK... and they don't pay UK corporation tax.

But you can probably see the problem now... Google can argue [perhaps fairly] that the services paid for by UK consumers can be served from any of their hardware, located anywhere in the world. Then they book their sales via wholly-owned entities in jurisdictions with zero-rate corporation tax.

Unless or until the way that corporation tax is charged is cleaned up, multinationals are going to continue to "cook the books" and not pay their fair share of tax.

So yeah, maybe Google *are* killing it. But they're doing it by killing the economies of countries other than the US, where they are not paying their fair share.

Raspberry Pi goes large with 10.1-inch Touch Display 2

Re: Priorities?

A Pico is an amazing piece of technology - and represents good value for money given the price point.

But a Pico is an entirely different class of device compared with even a 3B or 3B+, never mind a 4B or 5. It would probably be more reasonable to compare a Pico with [or at least put it in the same general tech category] as the microcontrollers like the Arduino. If you look at the most popular projects for a Pico, then you'll find them being used for robotics and general (DC Servo) motor control [classic microcontroller application], model railways [classic microcontroller application], programmable macro keypads and specialist input devices [microcontroller] and even custom gamepads.

What you don't find for the Pico are the sort of general compute projects for which the Pi is such a perfect fit - a family mail or web or private (e.g. NextCloud) server.

So I can't dispute your observation that a Pico is a potent and competitively priced piece of technology... but that potency and price are sadly meaningless if it doesn't provide the general compute capabilities of something like a Pi.

You've given us a chalk-and-cheese comparison. Your argument in favour of a Pico is valid for the niche it services, but not as a Pi replacement.

Priorities?

I find this development interesting not because the foundation have now produced a 10-inch touch screen, but because of what they're *not* spending their available effort on.

It isn't just the cost of the Pi range that continues to climb without limit, but the power consumption and TDP that goes along with each new model released.

There aren't many nice things I could say about Intel these days, but one genuine note of appreciation I'd offer them is their steadfast consistency in the production of lower-powered CPUs at known, predictable TDP thresholds - like the 35W and 65W models we see in every range - like the 14900T [which, health warning, has a default TDP at 35W but can boost up to 106W under max turbo loads].

I have a hunch that many people using Pi hardware have as much of a challenge cooling their machines as anything else. So - bearing in mind how long the machines have been in production - it would be really, *really* nice if they would be willing to offer to produce a Pi 3C and a 4C, which deliver the same processing performance as the current "B" models, but, by being built with the latest-generation processors, do it with a worthwhile reduction in heat output.

Honestly - and I'm not just saying that because: heatwave - I'd much, much rather have lower-power machines than something like this. When it comes to peripherals, the foundation is competing with dozens of resellers. It would be nice if they would stick to their core competence and bring us cooler-running models. I feel obliged to caveat/qualify this statement...

Currently the only way I'm aware of getting e.g. Pi4B performance at "lower thermal cost" is in fact to buy a comparable Pi 5 and then underclock it to bring the power consumption down. The A76 in a Pi5, for example, runs natively at 2.4GHz and is a massive upgrade over the 1.5GHz (or 1.8GHz in later machines) offered by the 4B. A Pi 5 clocked at 1.8GHz drops it's typical thermal output down from 12W to 7-9W and you still get a slight performance edge thanks to a newer die design.

The catch? Being able to justify buying a Pi 5 when you only need the performance of a 4B...

Auditors tell UK government to do the math before banking on £45B AI savings

Re: A moment of your time, if you please

...he falls asleep for.

There, FTFU

OpenAI admits it was the source of the agent swarm that attacked Hugging Face

Re: 100% they are lying

Does it still play tik-tak-toe?

Re: 100% they are lying

It's OK.

We don't have to be concerned about that - unless they try and get access to W.O.P.R.

Linux kernel team publishes 432 CVEs in two days

Regression Testing

One thing I haven't seen [probably thanks to not looking too far, to be fair] in the fallout conversations from the significant up-tick in patches for AI-detected vulnerabilities... and that's regression testing.

If you're a company or a team or a developer, especially someone working in the middleware of the GNU/Linux ecosystem, you could historically be pretty relaxed when it came to patching your Dev and UAT environments. Typically you'd freeze patching of your machines within sprints... and then between sprints you'd allow updates... which you did so that you didn't end up with false positives or waste developer hours chasing down obscure issues.

This strategy was perfectly fine - even in agile shops - when the update frequency for code on which you had hard dependencies was low. Yes, there are lots of patches being released all the time, but the frequency of patches for {infrastructure code X} remained pretty low / manageable for most development shops.

That world no longer exists - we've moved to the vulnerability patching equivalent of "Everything, Everwhere, All At Once", which is already playing merry hell with development, because it leaves you with a dilemma. Either you update your dev environment more frequently, which means you risk lots of pauses while you figure out why something you had working yesterday just stopped working... or you keep with your configuration freezes outside of release candidates... and you're left with a huge mess to sort out at the end of the cycle.

There's no magic bullet here. At least [corrected] there's no magic bullet *that I can see*.

My sense is that we're all just going to have to adjust to working with much more volatile environments; we're going to have to put serious thought in to automated regression testing [if we're not already doing that] and companies offering those sorts of tools are going to be in high demand going forward.

Projects will recalibrate and will find the sweetspot in their release cadence; perhaps they will get better at creating a handshake between Agents finding a vulnerability and then Agents pulling down the open source code, locating the vulnerability, and preparing a patch for review by maintainers. [And I know that won't be universally popular].

But... like King Canute watching the incoming tide, we can't sit this one out.

The vulnerability tsunami is here.

The only choice we're left with is how we handle it.

UK's seventh prime minister in a decade says he'll ditch digital ID scheme

Re: Hooray!

That and (likely) the fact that he realises that any such project will turn in to another central government money-pit, result in one-sided contracts locking the nation in to paying {insert US mega-corp here} for years to come... and the acceptance that there are more important things for the government to be spending money on...

Airbus takes flight from AWS. What happens next is critical

I think you might be over-stating slightly by framing your point as the decision resting with a spreadsheet as the application. I'd say that the application is a mix of business logic and presentation layer - and that what matters is the data structure you're using for the actual processing you need to perform.

I'm not sure if you're aware of this... but there was a point a while back - region of maybe 10-15 year - where Microsoft starting telling their folk that they would be officially dropping Access from Office and instead migrating all the functionality to Excel. Working for a huge multinational I got access to all sorts of early release stuff and some of their strategy documents and the messaging that came with the comment suggested that they had been doing some user surveys and concluded that most large-scale clients - big corporates - were using Excel to crunch data as much if not more than numbers.

They literally fell back on "fields will be columns; records will be rows" ...

I can't tell you what ever became of that idea - sorry, though perhaps another reader can fill in the blanks - but really my point here was to observe that the real power or limitation from Excel [or any other app for that matter] tends to start and end with the definition of the data structures it uses. And yes, I'm aware that this likely contradicts any premise of user-centric design, but there's correlation between intuitiveness on the front end and the cleanliness of the data architecture on the back end. Get the latter wrong and you're screwed on the former before you've even started.

Let's not get carried away - it's a rare case of modesty.

Microsoft didn't even claim "Sentence..."

Kind of feels like there's a rare, deeper truth in there...

Microsoft cuts OneDrive support for older Windows 10 versions next month

Hang On

I understand why a company like Microsoft would terminate support for older versions of it's software. Realistically they can't keep on supporting every release forever. That's reasonable when it comes to support for technical issues.

But this might be a bit different.

As the article says, Microsoft will continue to support OneDrive on Windows 10 22H2 until 2028, but will stop for earlier versions from August 15, 2026. The issue here isn't - or shouldn't be - the version of Windows the user is running. The issue should be the version of the OneDrive client the user is running.

To suggest that we'd need 2 different versions of a OneDrive client - basically a proprietary WebDAV-for-Windows - to run on *two different patch levels of the same version of Windows* seems to be an awfully tall ask. For this to be a legitimate reason for Microsoft to withdraw support - and I'm sure such reasons exist - this issue should not be the OS but the OneDrive client codebase. If there is genuinely something operationally different for OneDrive that means it works differently between the two Windows 10 releases would smell a bit rotten.

This doesn't look like Microsoft withdrawing support for e.g. Office 2000 because they're migrating they're no longer providing training for their staff on a product with just a few hundred users. This looks like Microsoft looking for a lever to withdraw support for older versions of Windows 10, using client code - written to a common API no less, as the hook.

This is exactly the sort of garbage they used to pull with DirectX on Windows and Triple-A Games that came out that absolutely HAD to HAVE the latest version of DirectX, even if the first thing you did was disable those features because your graphics card wasn't powerful enough to run them. In short, it was just a built-in mechanism to strong-arm users to upgrade to the latest version of Windows.

If there are genuine, technical reasons for Microsoft doing this, then the issue should link to the version of OneDrive, not the underlying OS.

I smell a rat.

I hope the EU takes a look.

OpenAI admits GPT-5.6 occasionally deletes files – but it's an 'honest mistake'

Re: an "honest mistake."

Your comment reminded me of a perfect quote, which I've seen attributed to Robert X. Cringely:-

"If the automobile had followed the same development cycle as the computer, a Rolls-Royce would today cost $100, get a million miles per gallon, and explode once a year, killing everyone inside."

Don't worry about the bit where this goes bang... look at the low, low price... a *million* miles per gallon... Sign here, please...

Insert token to continue, says AI. Yeah, about that...

Re: AI.

Just bear in mind that that isn't the AI per se.

It's the humans that define the way the AI interacts with it's users - and the technique is taken directly from Transactional Analysis, which is a robust and well-evidenced psychological framework.

What the human designers of the commercial AI clients have done is to teach them how to give responses to human users that generate the dopamine "hit" of a compliment. Obviously you have to bear in mind that AI is still learning how to do this with subtlety, but if you think about it from the other side... Imagine an AI that you ask for help to triage an error message in a log file. If your AI told you the answer but then said, "You do realize that this is documented ... here, on this page, in this section of the reference material? I'm surprised that you're so incapable, incompetent or lazy that you weren't able to look this up for yourself, loser..." ... that you're not likely to continue to use it.

I'm not suggesting you're wrong; I'm not critiquing your observation as misplaced... I'm suggesting that there's a simple reason that the models are behaving like this and it's all down to the fact that the model creators are trying to get their users hooked. Tobacco companies use nicotine addiction; the AI companies are using dopamine.

Once you understand that and once you develop a filter to ignore it, you can still get useful output.

And if you still don't like it, have you considered a per-session instruction: tell it to stop with the effusive praise.

Even HP resellers think the price of toner and ink is too high – so HP India facilitated an illegal cartel

Better than Fines

By all means impose some fines on HP if you need the currency.

A much, much better move would be to ban HP from selling any of their products, either directly or indirectly, in India, for a period of 5 years. There's enough competition in every market where HP does business to give people alternates. You could explicitly allow a transitional support period where an India-owned and India-operated licensed importer could bring in supplies and consumables to support existing in-country HP units, but prohibit the import of all assembled hardware - laptops, printers, MFDs, the works.

In the grand scheme of things, India is likely a rounding error on HP's global revenues, but the mere act of kicking the tossers out of the country for five years would send the HP board in to a spin. So quietly add some new laws including immediate arrest and custodial sentences for the C-suite of any company convicted of the practice... and then let HP send their Sales director to India to try and resolve the issue. Throw that person in prison and then, maybe, these people will get the message.

Once one country successfully implements that kind of sanction, others will see it as an opportunity to give a boost to their domestic companies and will follow. At which point these egotistical multi-nationals might get taken down a peg or three.

TSMC's $265B US fab pledge is the outline of a concept of a plan

On the other hand...

... if the EU / UK knew what was good for them, they would be working triple-time to convince TSMC to break soil in the UK or Europe.

The UK is the birthplace of ARM, of course [despite it being sold to Softbank yonks ago] and it might be surprising but not inaccurate to suggest that, in actual fact, the UK/Europe looks one whole heck of a lot more stable, both economically and politically, than the US does right now. Of course TSMC is going to be threatened with restricted market access... but it doesn't take a stretch to look at the protests against data centers in the US and the way that political violence is growing to realise that making such a huge investment in the US might not be all that smart after all...

Airbus migrating 70 critical apps from AWS to France's Scaleway amid digital sovereignty push

"The more you tighten your grip...

... the more star systems will slip through your fingers."

Props to Carrie Fisher.

I've been sitting here trying to figure out how to work in, "Governor Tarkin, I should've expected to find you holding Vader's leash. I recognized your foul stench when I was brought on board"

but frankly, there are just so many options I'm not sure which substitutions to go with!

Photovoltaics are still running after a year under Swiss trains

Interesting Photo

I note that the picture with the article is one showing railway with an overhead "live" wire for the train to pick up using a rooftop conductor [the ground wire being the tracks themselves]...

But the presence of the pylons required to carry that electrical feed means that lengths of Swiss track have gantries overhead already.

Maybe an alternative approach would be to simply build out a more substantial suspension mechanism for the the live rail - make it from aluminium frames on to which solar panels could be placed "on top".

If this were done, then instead of simply having one row of panels "lengthways" - parallel with the track - it would be possible to lay panels 10-20 units across. It would also make it possible for the orientation of the panels - and their elevation - to be better tuned to point towards the sun... though of course it would only really be viable where the compass heading of the track meant that the panels could be set to point south [you'd need track running east-west for best results].

This would mean that not all track would be optimum for deployment, but, where it was possible, you'd potentially be getting 10x 20x the panel density of the solution taken.

You'd also see a reduction [not elimination] in the amount of dust and soot coating the panels [thought the downside would be that they would be harder to clean].

Interestingly, as the cost of electricity gets steadily higher, this is going to become more and more cost-effective. Bring it on.

Re: Looks like the trial was in the perfect spot

The separation between the rails, combined with the tightest turn that is permitted on railway track [which is not that tight] means that there should be plenty of room to fit existing 400W panels with relative ease. Which doesn't address the other problems, but does I think render the specific concern you raise to be less of a worry, even with the narrower gauge of Swiss railways.