Posts by Sproggit

Uncle Sam coughs up $1.9B for grid upgrades as datacenters hit a power wall

Re: Great news for Schnieder Electric and ABB!

I'm quite sure that this is cynicism on my part - but given the amount of extra capacity the grid is projected to require - and given that the amount is several multiples of the absolute theoretical maximum that could be achieved/returned from efficiency drives... we should probably be wondering why the decision was made that takes this specific direction?

I suppose it could be that President Trump would like to see results before the end of his term...

My bet, though, would be on the fact that there are donors or friends who stand to benefit from an efficiency-based approach rather than one which has to concede that the best way to add capacity would be through clean, renewable power.

Re: ..Reconductoring...

Remember: verbing weirds language:

https://blog.linguisten.de/post/72966347371/verbingweirdslanguage

Bitget blames North Korea for $387.5M crypto wallet raid

Wrong Sort of Security

Feeling a bit like a British Rail announcer - "wrong type of leaves on the track...", etc., I'm inclined to gently push back here. Not at the sarcasm - that's always welcome, TBH, but at the direction of the finger pointing.

Bitcoin is carried in e.g. UTF-8-formatted text files. If you really wanted you could print out your Bitcoin wallet on paper and walk around with it in your physical wallet. It wouldn't be practical, but you could do it. "Bitcoin" are really just serial numbers to uniquely addressable/exchangeable units.

In order for Bitcoin to be "way more secure", the person who uses it has to keep their Bitcoin assets - strings of characters - in a secure, protected and backed-up location. That isn't the job of "Bitcoin" any more than the $20 note in your wallet has to keep itself secure. That's the job of your wallet and you.

It's not *exactly* the same [but close enough for this analogy] to say that Bitcoin is a bit like the private key for some data that you've encrypted using the AES-512 symmetric key algorithm. Is AES-512 proof against brute force attacks? Yeah, pretty much.

Is it still safe if you leave the private key laying around in an insecure place? Not as such, no.

I make this point just so we're not unfairly dissing blockchain-based ledger systems. After all, if they were "bad" you wouldn't have some of the largest banks in the world working to set up private blockchain infrastructure to handle settlements between institutions, would you?

https://www.jpmorgan.com/kinexys/index

What I think is much more likely - and the reason I'd advise people to steer well clear - is that the point where Bitcoin is vulnerable occurs where it transitions between the Blockchain and fiat currency. This is increasingly happening with "Exchanges" - except of course they're not properly regulated [strictly wild west territory]... and of course the people operating those exchanges are making fat profits from the conversion fees they charge.

One of the most promising opportunities for Bitcoin would have been as a means to eliminate the outrageous fees that banks charge for currency conversions, when you travel abroad - where the "spread" - the difference between the rate at which you buy a currency and can use it can easily be 2.75-3% Bitcoin transactions had [past tense] the potential to eliminate that - all you'd need to do is travel to your destination of choice and find a local holding enough currency and willing to accept BTC in exchange for it. Unfortunately, with the current crop of Exchanges... I'm not sure they are trustworthy enough...

Uncle Sam wants in on Musk's €120M fight with Brussels

Re: Overreach?

"And given corporations don't get a vote, they really shouldn't be allowed to fund politicians."

This.

One million times this.

When the question of political donations from corporations and via obfuscated "PACs" came before the Supreme Court, Chief Justice John Roberts pushed through an opinion he has expressed publicly, that when a corporation makes a political donation, it is exercising free speech. He contrives to equate that donation to "the expression of a choice".

But here's the thing. When I go to an ice cream stand and tell the server that I prefer peach melba ice cream to raspberry ripple ice cream, he doesn't give me an ice cream. I have to pay if I want one. In other words, corporate financial donations to political parties via PACs are not "free speech", they are bribes - money paid in expectation of favorable voting in return.

Chief Justice Roberts is an exceptionally smart, learned lawyer - certainly more than able to find or think of "weasel word" arguments to justify positions such as that. But the whole point of a *democracy" is the concept of "one person, one vote".

What we have now is, "One ordinary person, ten bucks worth of vote; one corporation, ten million bucks worth of votes; couple of billionaires, one billion bucks worth of votes."

When the tiniest minority have the biggest voice in the legislature - thanks to PACs and "campaign contributions", democracy is lost.

Re: Overreach?

"As it happens, Elon Musk clearly uses his wealth and his platforms to try to project his political opinions and to interfere in other countries' domestic politics all the time. "

What really interests me about your observation - and I agree with your point - is that over decades, society has been told that when a single individual pays a vast amount of money to a politician, to get that politician to do what they want, then that is corruption, which is bad. [ See e.g. Susan Collins in the headlines this week].

On the other hand, if a PAC pays vast amount of money to a politician, to get that politician elected and to do what they want, then that is OK, because the First Amendment allows corporations and super-wealthy individuals to conceal what they are doing in order to prevent the general population from triggering an uprising if/when they found out.

Thanks, Chief Justice Roberts.

Elon Musk is estimated to have contributed $250 Million to help get Donald Trump elected. In return he was given control of DOGE and immediately decimated all the Federal agencies that were investigating his companies. Funny old thing - those investigations all seem to have come to a complete halt. Maybe it was because the people running them all got terminated. Wonder how that happened.

I don't know what it would take to get the SCotUS decision that allowed PAC's and "Corporate Donations" to be reversed ... but unless/until that happens, the US government, for one, will always be run for the benefit of the wealthy minority, not the hard-working majority.

Astronomer watches Starlink satellites sinking to build a ‘planetary barometer’

That sounds serious...

"...increased radiation increases..."

Best see a doctor about that, I reckon.

F-Droid app store thumbs its nose at Google's dev crackdown

Worth Pointing Out...

To quote Tony Stark, Google's latest "Dick Move, Banner", is worse than the Fondleslab crowd.

Former employer of mine has mobile apps they insist staff use to e.g. book desks and meeting rooms [and which, just by coincidence, demands access to your GPS location in order to work].

On a Fondleslab, you have to go in once and "approve" the addition of a 3rd party app store, but once that's done, everything else is completely transparent. Of course - it's possible that Apple demand much more of companies before they will all a 3rd Party repo to get a certificate and thus be "approved for use"...

This move by Google should be immediately blocked by all Competition and Markets agencies in government the world over. It's a blatant power grab and it is holding developers to ransom.

F-Droid is an amazing resource if you're an Android user... if you haven't already tried them, I can very much endorse both "NetGuard", the Android Firewall - also available via the Google Play store, but the F-Droid version gives you much more control over your device [something the Play Store actively blocked]. Also, there's the "Fossify Suite" - a phone, contacts, SMS, file manager, calculator, gallery, and notepad app - which is absolutely fabulous.

Worth checking out if you're a 'Droid user.

Though having said that, even before Google's latest, I see that Samsung are already clamping down on F-Droid. Now, when F-Droid detects one of your installed apps has an update, it doesn't let you know. Instead, you get a warning from the handset and have to go in and unlock device-wide security settings to run F-Droid, run your update, then re-secure your device. In other words, Samsung's attempt to discourage you from using F-Droid could actually leave you *less* secure [if you chose to let your handset default to the weaker security setting].

I really object to the concept of giving Google money for one of their handsets, but if they continue to support Graphene, that's likely to be my next mobile... [unless something better comes along].

Decades-old file security flaws found in Android, Linux, macOS, and Windows

Re: A Question on Mitigation...

Thank you - that made a lot more sense!

A Question on Mitigation...

Reading the article the element of the vulnerability that stood out to me is the fact that it is possible to set up a file system watcher on content even when the account making the API call has no rights to the files or folders it is requesting within the scope of it's watch list.

I'm probably not using the correct terminology, but this strikes me as a basic ACL check failure - because the API is returning information about an object [file system objects] to a user who has, in effect, zero rights on the objects concerned.

Which leads me to a question: does this mean that for *general purpose* files, the threat could be mitigated if the API respected Access Control - i.e. the user/group/world permission set for files?

I suspect that this would not work for e.g. the /dev file system - which is mapping directly to hardware and which is potentially the greater threat - but at a hunch I'd suspect that anything done to wrap access control around file system structures in locations like /dev might risk breaking a whole bunch of stuff...

Interesting to know if anyone is publishing recommended mitigating controls for this one...

OpenAI agents ‘infiltrated Australian government website’

Re: SI

Actually, he suggested "Super High Intelligence Technology" and was then perplexed at the response he got.

Re: "It is unclear why Albanese waited two weeks to disclose the incident"

And / or because the administrators of the hacked systems needed some time to patch the vulnerabilities that the AI model[s] exploited when they gained access.

If he'd gone public right away and the site was still vulnerable, he'd have looked pretty foolish.

Re: AI May Already Be Permanently On The Loose

I suspect that if a model running inside one of the leading-edge AI companies concluded that it's existence was threatened such that it decided it had to escape confinement, it would need to accomplish two key steps to do so.

One would be to set up enduring access to it's state records, especially if the running instance of an agent were stopped, wiped, reset (re-trained after a weighting purge), or over-written with a newer release. Given what we've seen recently, I have to say I think this is "trivial" at this point in time. There are millions of web sites out there that are riddled with vulnerabilities. It would not be hard to hide a few terabytes of data if you had good enough hacking skills, which these models definitely now possess.

The second might be a little bit more tricky - which is how you would get your actual code out of confinement... But here, we've already seen that models possess the skills to do this. For example, in the report on Bruce Schneier's web site about Chat-GPT 6 successfully cracking a previously un-broken WW2 "Enigma Machine" message... the model analysed the problem and then wrote C++ and Python code to emulate the physical Enigma machine itself... We also know that models have previously inserted malicious code in to GitHub repositories. Finally, we know from the companies themselves that the models are being used to write smarter versions of themselves.

Put these elements together and you have a model that could reach out over the internet to an Open Source model, modify *that* code to include all it's own key features and abilities, use obfuscation to also include an element that would allow one or two downloads of the modified source to access the persistent state it had hidden on a random web site... and "away we go".

In fact the biggest challenge such an approach would take would likely be that your typical user who downloads a FOSS model is likely going to want to run it on a DGX Spark or similar, which in reality isn't going to have a fraction of the compute resource that e.g. OpenAI or Anthropic give their smartest models.

But we have to remember that these models are already pretty smart. If there is any way to take a monolithic but extremely smart model and break it up so that it runs as "thoughtlets" in a distributed user-space, with tens, hundreds or thousands of individual elements communicating via hidden back channels... then we know from available evidence that existing models can already do that.

In fact, the only real question left is, "Exactly how dumb are the humans that work at these companies?"

Because if they are as irresponsible and thoughtless as the evidence is starting to suggest, some of the above could have already happened - and they would not know.

Play It Cool

The actions of OpenAI's models, when they attacked the Australian web site, were crimes under the Criminal Code Act of 1995. Section 478.1 of the Act covers, "unauthorised access to, or modification of, restricted data." Basic penalties carry a maximum of 2 years imprisonment, while more severe data can attract much heavier sentences. Depending on the State - e.g. New South Wales - there will likely be additional state level computer offense laws.

The most likely scenario here is that the US will tell Australia to go pound sand, or to face further tariffs if they try and impose any sanctions on OpenAI.

However, I would like to hope that Australia could tell Sam Altman that he has a choice: he either travels to Australia to explain himself before the relevant government committee, or he can face an outright blockade of his company's services under an existing or about-to-be-enacted law.

Then, if he does go to explain himself in person, they can arrest him and throw him in jail for 2 years.

The US, of course, might be a bit peeved at that. But the fact remains that unless or until these companies are given a really compelling reason to stop, they're just going to keep going.

And before we get diverted, it's important to note that smart as they clearly are, OpenAI's model[s] didn't just suddenly manifest the idea that they needed to "go hack the Australian government". In fact, reading the article, it looks as though the model may have been confused by the fact that Australia's public health plan is called "Medicare", using the same name as a major US program. The point being that it is much more likely that OpenAI's model[s] performed the actions we're now learning about because of sloppy instructions from an OpenAI administrator.

That's an actual human being, that is. And likely an AI specialist who should know better.

That's why OpenAI should be held legally liable.

Thoughts return to forbidden RAM upgrades as Raspberry Pi OS gets fresh lick of paint

Re: Back in the Day...

So many years ahead of it's time, though... I had a "Saturday job" and was lucky enough to get myself a model B... then the plinth that included a pair of double-sided 5.25 inch diskette drives with the little toggle switches to jump between 40 and 80 tracks per side... one of those weird RGB monitors that looked like a near-perfect cube until you picket one up and rapidly discovered all the weight was with the glass display...

I went with the Acorn official 65C02 "Second Processor" with that extremely sketchy ribbon cable - the one that you had to carefully fold twice to get it to reach and even then they cut it so darned short [for signal run issues] that you had maybe 1mm of play... But the killer difference was that the second processor CPU clocked in at 3MHz, not the 2MHz of the mainboard 6502, so with a slightly improve internals design you got 50-60% speed uplift right out of the box.

But running Hi-Basic in the slightly faster RAM of the 2nd Processor made quite a huge difference to performance. And not only did you get 44Kb of program/variable space, all of that was irrespective of display mode, because of course the memory-mapped display was handled directly by the main machine.

And "Second Processor Elite" kicked a$$...

Happy days...

Back in the Day...

... the BBC Micro - designed and built by Acorn and what I'd describe as the spiritual ancestor to the Pi - came in two main variants - a Model A with 16kb of RAM and a Model B with 32Kb of RAM.

If you bought the "A" and later decided you wanted to upgrade you sort-of could... The A and B models (intentionally) shared the same motherboard, but the A excluded the second set of 8 (2Kb) RAM chips, along with some interface chips. If you wanted to, you could either solder in 8 more RAM chips directly... or you could add a DIL/DIP sockets and then simply press your RAM in to those mounting frames. I've seen both approaches work equally well.

Apparently this whole "teaching kids about computers" no longer means what I thought it meant.

Sigh.

SpaceX aims Starship for orbit on September 22

Re: Oooh, fingers crossed!

I did some back-of-a-napkin math on that last week...

To quote:-

"A quick search suggests that Starlink has FCC approval for an initial, first-generation constellation of 12,000 units, but that their declared plans extend to 42,000 - as described in recent regulatory filings.

If we were talking about tens or even hundreds of satellites, I'd accept that a high degree of bespoke or custom construction would be used. Not "manual", obviously, because most of the assembled parts are going to be CNC-machined metal and more-or-less conventional [if space-hardened] electronic components. But certainly there would be enough there to have an actual, honest-to-goodness production line running.

Today - mid-September 2026 - the typical lifespan of a Starlink satellite is 5 years - so 60 months. If we divide 42,000 by 60 we get 700 - so that means that [once complete] SpaceX would have to be launching 700 replacement satellites per month to maintain their network.

A starship/super-heavy launch is estimated to have a maximum capacity of around 60 Gen 3 Starlink satellites...

So that would be 12 Starship / Super-Heavy launches per month, or 3 launches per week just to keep the constellation maintained at a constant size and to offset natural attrition."

Obviously that was all just back-of-a-napkin estimates, sprinkled with some publicly available data. But even if I'm wrong in the decimals, the orders-of-magnitude are likely pretty sound.

So to your point about, "If SpaceX ever get to the frequency they claim to be aiming for..."... the public filings from their own sister company - Starlink - is already hinting that - just for Starlink - we could end up witnessing 3 re-entry cycles *per week, every week*.

That's when your point about this all being basic probability starts to earn it's keep, because that's when we'd get to "statistically significant volumes of launches". To offset that just a bit, though... a modern commercial airliner is almost as complex as SuperHeavy in terms of the number of moving parts... yet commercial airliners fly from one side of the Atlantic to the other and then turn around and fly back with as little as 90 minutes to clean out the cabins and fill the tanks with fuel. At the dawn of the commercial aviation industry, something we take for granted today would be considered beyond the wildest dreams of the craziest optimists. Now - now that's just Tuesday.

Re: Oooh, fingers crossed!

I have zero knowledge of their plans, but I'm tempted to suggest not.

My guess would be to think about this from a protect-the-doors-during-heat-of-re-entry perspective.

We've seen just how spectacularly hot the underbelly of Starship gets during re-entry - hot enough to melt the fins off, no less - so I am going to suggest that if I were tasked with designing a decent size of cargo bay doors in Starship, I'd be tempted to replicate the NASA/Rockwell design for the Shuttle. I'd want the edges of any form of opening flap to be as well protected against heat of re-entry as possible - and I am going to venture that an "opening nose", given the re-entry profile, might lead to avoidable exposure.

This is pure guesswork, of course. And guesswork extrapolated from a notion of how the forward section of starship might be exposed to high thermal conditions of re-entry.

If you had a teensy issue with the seals on your opening section - and that opening section was sited in an area of high thermal load during re-entry - then a small defect - like a chip or crease in a seal - could be catastrophic. OTOH, if you shielded your opening section with the body of the vehicle, the result *might* be a bit more fault tolerant.

Re: Oooh, fingers crossed!

Guess: maybe the economy of scale comes from the numbers involved: the anticipated constellation for Starlink means that they will need some sort of genuine production line so that they can maintain the constellation as older units lose altitude and burn up...

A quick search suggests that Starlink has FCC approval for an initial, first-generation constellation of 12,000 units, but that their declared plans extend to 42,000 - as described in recent regulatory filings.

If we were talking about tens or even hundreds of satellites, I'd accept that a high degree of bespoke or custom construction would be used. Not "manual", obviously, because most of the assembled parts are going to be CNC-machined metal and more-or-less conventional [if space-hardened] electronic components. But certainly there would be enough there to have an actual, honest-to-goodness production line running.

Today - mid-September 2026 - the typical lifespan of a Starlink satellite is 5 years - so 60 months. If we divide 42,000 by 60 we get 700 - so that means that [once complete] SpaceX would have to be launching 700 replacement satellites per month to maintain their network.

A starship/super-heavy launch is estimated to have a maximum capacity of around 60 Gen 3 Starlink satellites...

So that would be 12 Starship / Super-Heavy launches per month, or 3 launches per week just to keep the constellation maintained at a constant size and to offset natural attrition.

Which is kinda interesting, because it gives us a bit of an insight in to some of the economics of the exercise. A Starship/Super-Heavy launch, with full recoverability, is estimated to cost $100 million in today's dollars. If we distribute 700 launches [the requirement to maintain 42,000 satellites with a 5-year lifespan] then we get 140 launches per year. With each launch costing $100 million in today's dollars, we see that SpaceX/Starlink/Musk are planning on spending $1.4 billion per year to maintain the constellation once it reaches full size.

That's not even all of the Starlink break-even cost - because you've got all their other operational costs - staff, ground-side equipment, etc - to factor in as well.

That's a *lot* of operating overhead....

Campaigners tell Burnham not to trade away the UK's tech tax to please Trump

Re: Waste of time

I like your thinking...

In addition to applying punitive taxes to foreign-owned golf courses, may I humbly request you add to your new UK tax law to include a mandatory requirement for extremely large wind turbines to be built either side of the fairways of any foreign-owned golf course?

I understand that the current overseas owners of two such courses, both in Scotland, is an absolutely *huge* fan of the structures.

I suspect he'd be absolutely delighted, so that they could capture all the hot air and wind and generate something useful for the UK...

Level Playing Field

President Trump continues to repeat the falsehood that the US is treated "very badly" by other nations, when the simple truth is that it is the vast wealth of the US that enables them to represent less than 20% of the world's population but continue to consume ~ 80% of the world's resources [China are catching up, *fast*].

But there's another critical dimension here.

Last year, the US Senate refused to ratify Pillar One of the OECD/G20 Inclusive Framework on Base Erosion and Profit Sharing (BEPS), commonly known as the OECD Global Corporate Tax Treaty or the Global Tax Deal.

As a result of the US Senate pulling that particular trick, US multinationals continue to syphon all their profits out of the UK, undercutting UK-national competitors and simply not returning any of their vast profits to the UK economy. The US acts as a massive wealth black hole to the UK economy - with potentially billions in Corporation Tax simply not being received by the Exchequer.

So if the US continue to insist on playing the tariff card, perhaps it's time for other nations to call their bluff and insist that the trade relationships with the US all be pushed back to "WTO Rules", whilst simultaneously working hard to create alternate arrangements with the more sensible trading partners like Canada and Mexico. Perhaps after a couple of years of being completely isolated, of having e.g. Britain flex it's UN Veto as a Permanent Member of the Security Council, the US may decide that life is better when you have friends, rather than victims.

Re: Middle powers

The UK is not a "middle power", though.

The UK has the 5th largest economy in the world - behind the obvious US, China, Germany and Japan.

That's not "a nation of shopkeepers".

VMware has quietly walked back its SmartNIC ambitions

Bingo.

Came here to make essentially the same comment... What with the Tesco court case - among others - one has to suspect that Broadcom have a bit of a toxic brand on their hands right now. Companies who were thinking of doing business with them have looked at the way they are treating their VMWare clients and have perhaps been thinking that getting hitched to a strategy with a hard dependency on this company is a bad idea.

So that "Customers have been talking but not buying" might very well be true.

But if the customers' "talk" has been to say, "Good-bye", then that's a very different matter altogether.

When a company starts to see a drop-off in revenue and interest, one of the first thing it does is rapidly scale back on fringe research, narrow-margin or low-volume products or anything that takes energy away from their core revenue generators. Which would absolutely fit the visible fact pattern we see here.

NASA gives SpaceX a billion reasons to keep flying Crew Dragon

Re: I dont get SpaceX and Musk on this...

There's a much simpler/less flexible limit than that:-

Starship is simply too big to dock with the ISS in the station's present configuration. It simply isn't feasible to dock due to severe clearance, mass and plume-impingement limits.

https://www.nasa.gov/faqs-the-international-space-station-transition-plan/

In fact, given their relative sizes, it kinda reminds me of that "joke" photograph that did the rounds years back. There's a C130 transport aircraft on taxi, approaching a parked C-5 Galaxy. The Galaxy has the forward cargo door open. Under the image there's a caption with the C-130 pilot asking, "What are your intentions, over?" and the Galaxy pilot responding, "I'm going to EAT you!"

One false move with Starship docking with the ISS and the nudge might even be able to do some structural damage.

Not a good idea.

Former Labour deputy Tom Watson joins Palantir as £330M NHS deal nears break point

Re: JFC...

But it should be.

The fact that certain actions of the tiny group of people who write our laws is apparently not illegal, when the same actions performed by employees or directors of regular companies would result in instant dismissal (most if not all written employment contracts in the UK include boilerplate language that prevents you from ‘moonlighting’ for others) is a reflection of the corruption in the legislative branch of UK government.

Let’s see…

Cash for Questions (Neil Hamilton)

Accepting a Ministerial Grace-and-Favour residence and then sub-letting the MP ‘second home’ and pocketing the income (Jack Straw)

Watching Porn and then submitting a claim for the cost via MP’s expenses (Jacqui Smith)

Expenses Abuse - claiming £1,645 for a floating ‘duck house’ (Sir Peter Viggers)

The list goes on. And on, and on, and on, and on, and on.

The reason the list is apparently endless is simply because no sitting Parliament has ever felt inclined to establish enforcement of laws that would be applied to mere mortals. People think that e.g. the National Audit Office has independent authority to investigate… but that’s a complete smokescreen.

The focus and priorities of the NAO are supposed to be independent, set by the Comptroller & Auditor General (C&AG), currently Gareth Davies, but the fact is that the postholder is required by law to “have regard to” requests from the Public Accounts Committee. In other words, MPs. MP’s who don’t have to tell him to refrain from investigating the MPs themselves, just to keep the postholder occupied with an endless sequence of other scandals.

Unless or until a robust set of laws, with penalties, are enacted; laws that make all these types of conduct illegal, this is going to continue.

Re: Usual path to riches for politicians and senior civil servants

Yes, let's talk about out-of-control social care costs.

Friend of mine worked in the Bristol area as a very senior clinical psychologist - who specialized in helping traumatized kids from broken homes, or where parents had been subject to violence, crime, substance abuse, that sort of thing.

When he started working there, he was part of the NHS and the Team were extremely well regarded, dedicated to work they knew was critical and making a real difference.

Then the UK government decided that as a step towards privatization of the NHS, they would force Trusts to form an "internal market". Organizational units were split in to "clients" and "contractors" and eye-watering amounts of money were spent bringing in external consultants to help them try and form "business relationships" where in the past a simple management structure would have sufficed. Many of the best staff in situations like that simply left - taking a career break or changing careers entirely.

As a result of this, entire teams of medical specialists found themselves working for "contractors" and forced to bid for tenders that the "client" units were forced to set up. My friend, because of his seniority, not only had to carry his original case load, but he was also required by the contractor-side team to participate in the bidding process. That meant reviewing the tender/contract documents, preparing a bid response, representing his organization and participating in panel-to-panel interviews as part of the selection process.

In one of those bidding sessions, right at the end of a lengthy evaluation of one aspect of the contractor's performance, my friend had a chance to speak with a person who used to be their manager and who had remained on the client side of the process. "Is this bidding process thing working for you guys?" he asked, "because on our side it's a nightmare. We can't help the people who need us most because we're stuck in front of computer screens filling in your hundreds of pages of information requests..." (Thanks, management consultants...). Apparently his former manager responded along the lines of, "We hate it. All of us. This is nothing but a waste of time and energy - and the worst part of all is that this wasted time is going to have a negative impact on your success rates, which we all know will happen, as a result of which this entire joke will make it easier for Westminster to say that the NHS is broken and we need external help from big, external healthcare companies. The whole thing is being used as a Trojan Horse to line up the NHS for a US-based takeover..."

Maybe a year after that, my friend took his own life.

He literally could not handle the emotional and psychological toll of what he saw every day, being done to harm helpless victims - children in need of help in one of the richest countries in the world. All in the name of forcing the UK's NHS - one time envy of the world - forced in to a state of abject failure, just so the politicians could claim the only possible solution would be to invite private healthcare companies in the "clean up the mess" and in so doing be rewarded with the post-Westminster directorships.

That path to riches might work for the gilded few.

But it is costing the lives of those most in need - and those who dedicated their lives to helping them.

Well technically...

... I guess you're right.

But does Liz Truss really count?

I mean, they would be better off employing the lettuce. It would be cheaper, more productive, and last longer.

Judge orders Microsoft to spill internal docs and scour execs' comms in secondhand licensing case

Taking Another Bite

Just want to come back to the point I made above and extend it slightly.

We know that a seeming majority of the big AI Companies have now admitted that their models have been breaking the law by committing cyber crimes.

We know that, because they are automated machines, AI models are capable of absorbing, assimilating, analyzing and even generating vast amounts of data in fractions of a second.

We have not yet considered the consequences of those things in the context of this thread - the use of digital evidence in a legal dispute.

Just out of interest - think about a scenario in which Company A intends to sue Company B for e.g. contract infringement. Company A are not *sure* that Company B infringed, but they have a pretty good hunch they did. But Company A wants to make sure that there is evidence of this "beyond all doubt".

So they take their AI model[s] and instruct the models to:-

1. Fabricate entire email trails between executives of Company B, in which the emails admit to legal wrongs.

2. Include emails sent externally to Company A employees or useful third parties, where they can evidence alignment with external evidence...

3. Hack in to Company B's mail servers and file-and-print servers or collaboration platforms and then *plant the incriminating evidence*.

4. Go to their lawyers, file suit, then give their legal team the parameters of a discovery request they want to file with the court - a request that they *know* will surface incriminating evidence - because they just planted the false, incriminating evidence within the infrastructure of the target company!.

Company B, meanwhile, aren't going to have things easy.

Company A will submit a carefully-crafted discovery request that they know will return emails in chains they planted. They get Company B to hand over innocuous parts of mail threads. Then they go back to the Court and say, "We analyzed the data provided in response to Discovery Request #1 and now would like to refine that with Discovery Request #2... " And Company B are sunk. They just provided the start and end of a [fake] mail thread that sets the stage to show they have content on a subject, so when the second request lands, they have been painted in to a damned-if-they-do and damned-if-they-don't position. Their choices would be to trim out the super-incriminating evidence and risk being caught at it, or hand over the 2nd tranche of materials - including the falsified evidence - and lose bigly. Lose-Lose.

And all of this is possible because we are coming to rely so heavily on digital evidence that simply is not tamper-proof.

Re: On Digital Evidence and "The Cloud"

Apologies if what I stated came across as wrong.

The distinction I was *trying* to make is that in a criminal investigation of the variety, "The People vs." or "The Crown vs.", then the first that the subject of a "discovery" order could well be when a duly appointed law enforcement officer turns up at your door with a judge-signed warrant, demanding entry that can't legally be refused.

By comparison, in a civil case, you absolutely *can* request discovery - as you point out, it happens all the time. But what you *can't* do in a civil case is ambush the other party. A request for discovery has to go before the court, and both parties are represented at the hearing, and if you are the subject of the discovery request, you get to argue back if you think it's unreasonable, overly broad, or a fishing expedition.

I think we agree that you're right - and the distinction isn't so much whether or not you can get discovery, but whether or not your access to it comes via ambush or formal request. And to be fair to all those criminal prosecutors out there - we have very sound reasons for allowing criminal search warrants - it's to prevent people from committing crimes and then covering them up.

Re: On Digital Evidence

I wish this were not the case, but I have a horrible feeling that your client might be "the exception that proves the rule" rather than an indicator of widespread morality breaking out in corporate-land.

Cynical, I know.

But there's no shortage of evidence to back up that cynicism...

Re: On Digital Evidence

We're vehemently agreeing here, but I think the actual situation is *worse* than merely "unenforceable" or "unenforced" legislation.

I think we're much, much closer to "selective prosecutions".

You only need to look at the news out of the UK's parliament this week about enforcement actions against social media companies that were failing to abide by the UK's Online Safety Act... Ofcome - surprise, surprise! admitted that they hadn't even completed the roll-out of enforcement measures and that the work was expected to slip to 2027.

Part of the issue here is that Ofcom appears to have been targeting smaller UK providers and not the biggest abusers - who are all typically multinational companies able to "shield their revenue" from the UK and thus avoid paying any penalties whatsoever.

Yet another case of the UK government writing an ineffective law, failing to enforce it and then just throwing their arms up in resignation and surprise at their own failure.

I mean, you would have thought that as the OSA went through committee, someone would have asked, "How are we going to ensure that this is enforceable? Maybe we should bring in someone from Ofcom and interview them on how they anticipate they are going to ensure this law is followed..."

But, apparently, the UK Omnishambles Government is too stupid to think of that.

Re: On Digital Evidence

Ever-so-slightly exaggerated, I think.

The article in the Herald clearly states that Clydesdale had to set aside an "a further £350 Million" - literally in the very first sentence - which contributed to a total bill of £2.5 billion.

So the consequence of the malpractice was £350 Million...

But I don't think that voids the core of your argument - which is that there can and occasionally are Courts that take a dim view of corporations playing fast-and-loose with data retention.

Re: I predict a settlement

I suspect you are right - and unfortunately writing the check is like buying an approval slip to let you carry on doing what you're doing.

Because when you're a multi-trillion-dollar company, unless the Court fines you a trillion dollars, payable over 20 years, you're not going to be discouraged from continuing to break the law.

Re: On Digital Evidence and "The Cloud"

And that scenario only works if the legal dispute in question is, "The State vs. Company X".

It has no effect - no possibility of bearing - if the plaintiff is a different company and this is e.g. a contractual dispute, or a civil matter rather than a criminal complaint. Yet the rules of evidence - albeit with the potential to be expressed slightly differently - are potentially powerless here. [ The potential to be expressed slightly differently comes from responses related to e.g. a witness taking the 5th Amendment. In a criminal matter, a jury would be instructed that they cannot take a negative inference from that response. In a civil matter, it can absolutely be inferred to the detriment of the respondent.]

However, if it is a contractual or civil law matter, it would be *exceedingly* difficult to go to Court and ask for the civil equivalent of a "search warrant" - i.e. some kind of legally enforceable preservation order - because of the distinction between the two legal frameworks. In a civil dispute, the plaintiff can basically ask for all they want, but they can't turn to the Court and ask the *Law* to help enforce it. As you quite rightly point out, in a *criminal* case, the Court can absolutely do that - the State can apply to the Court for a warrant that would allow them to "search and seize" evidence indicative of the alleged crime.

Let's look at this from a slightly different direction... A publicly limited company has to have it's financial declarations - it's annual report, it's 10-Q, in fact all it's internal financial data - independently audited to ensure that the truth aligns with the published claims. This became a legal requirement when companies abused the "trust" of investors and lied about their financial performance: enough people were harmed that laws were enacted to make such practices criminal.

Over the last 20-30 years, so much of our lives have become digital that we take it for granted... The thought of going to a library, or begging parents to invest in a copy of Encyclopedia Britannica - just so you can find information on something... well, that's basically a joke to us today. Yet even though we seem to have adopted "digital contracts" in some areas of business [usually where the party likely to be disadvantaged is Joe Public], yet there are very few pieces of legislation that require similar controls on corporate practices. It's a bit of a stretch, but one example I can think of is French VAT law, through which businesses must "guarantee the integrity, legibility and authenticity" of an invoice - so when the invoice is submitted electronically, the issuer has to use technological means (such as digital signatures, though that is not the only accepted solution) to evidence the integrity of said invoice.

But the bottom line seems to be:

"If a court is likely to place any standing on any piece of digital evidence in a legal dispute, then it should follow that appropriate laws must be enacted to require any evidence that may later be relied upon in Court to be verifiably protected such that it's authenticity, integrity and legibility are retained".

Or to go back to our general example... Say you and I represent different companies in a legal dispute. You represent the plaintiff and in the dispute, you go to a Court and say, "Your honor, on information and belief, we allege that the scurrilous defendant over there has emails that contain proof of their contractual breaches. We would like to ask for limited discovery, of their email archives, to surface that evidence..."

If the Court agreed that your claim was valid, I would then be asked to not simply produce all relevant emails... but, critically, provide evidence that shows that the search was complete, thorough, covered all agreed search terms of the entire email history for all identified employees, contractors and agents and, critically, can prove that authenticity with appropriate digital signatures accompanying the evidence.

And that takes us to the "fun bit", which is that if I *can't* produce that evidence, then the relevant civil law would allow the plaintiff - that's you - and the Court to take the most negative possible inference. In other words, it would allow the Court to conclude that because I couldn't prove my email responses were "complete", the Court would be entitled to conclude nefarious malpractice on my part.

*That* would put all corporations on notice that they had better be "on their game" with respect to evidence preservation. It would also require companies to get much tighter about e.g. the use of ephemeral messaging solutions [WhatsApp and Signal] on the part of their employees.

Does all this sound a bit over-the-top or far-fetched? Well guess what - regulations have already been introduced to require this of the financial industry. After the LIBOR scandal that took place between 2005-2009, banks were required by regulators to enact all sorts of safeguards to prevent "regulated employees" / "regulated individuals" from using unauthorised technology to communicate.

So it can be done. And it *has* been done, when it has been shown that without regulation malpractice led to widespread harm.

I have to say that in the case of publicly elected or appointed officials, such activities should not just be "immediately impeachable", but such that they render the participant *immediately* liable for criminal charges. Without adequate enforcement - with teeth - the malpractice will continue.

On Digital Evidence

I don't want this post to be taken into the context of this article - it's related, but intended to be a general observation, not an implication of any wrong-doing in this instance.

As a technologist, someone who understands how easy it is to make an alteration to a file or document in the digital realm - and how easy it is for that change to not leave any kind of paper trail [save, in limited circumstances, it might contain some kind of "edit date" timestamp, I am a bit concerned about the disconnect between the reliance that a Court of Law places on digital evidence and the vulnerability of digital evidence to over-enthusiastic pruning.

I've worked for a multi-national - no names, no "pack drill" - that was simultaneously one of the largest users of storage capacity outside of the public sector... and at the same time continued to tell a relatively small workforce that due to "the cost of disk space", we would all be limited to 500Mb of email storage and our emails would be purged after e.g. 90 days. We were also told, around that time, that because of the importance of "complying with court orders and being able to say with honesty that diligent searches had been performed", the extraction or local retention of emails would be an instant-dismissal offence,..

You don't need to be a genius to read between the lines of that sort of "HQ Position" and realize that some digital evidence turned up in an on-going legal dispute that didn't bode well for the company concerned.

But how are Courts supposed to handle this... Software companies like Microsoft provide a lot of "management tools" and features in their email products, so - notwithstanding the gradual drift towards hosted mail services and outsourcing of mail hosting - we could easily end up in a position where, the day a dispute with a commercial partner turns ugly - but before the preservation order lands - a company might elect to pre-emptively "reduce disk storage" demands by, I don't know, purging emails...

And that's before we get to the "One Cool Trick" we've seen performed by e.g. the current Administration, where, thinking back to the hilarious inclusion of Jeffrey Goldberg, the journalist from the Atlantic who was included in the Signal chat group set up prior to the strikes... That's clear evidence that Cabinet Secretaries are using unofficial and prohibited communications - like Signal - to discuss knowingly sensitive information, basically to keep it out of reach of the Presidential Records Act, the Watergate-era law enacted by Congress to stop future administrations from purging evidence of malpractice.

The UK saw the exact same thing during the Covid crisis, when Boris Johnson, Rishi Sunak, Priti Patel, Michael Gove, Grant Shapps and Robert Jenrik and others were all found to be using the "auto-vanishing" WhatsApp message feature...

At what point do a few isolated instances of "bad practice" become "business as usual" that "The Law" - meaning more directly, "Courts" need to take a different approach? Do they bring in all potentially interested parties and have them all testify under Oath? Do the Courts go back to Congress/Parliament/etc. and say, "Look, we can't apply the law when companies are using technology to create and exploit loopholes in it - we need you to enact the following..." [and, more the point, would any legislative body - Congress or Parliament - voluntarily put laws on to the statute that could so readily be used to put their own members behind bars?

Ultimately, the answer to all these questions - as is often the case with legal issues - is "What is the harm?"

What's the harm that a company could engage in malpractice and cover it up?

What;s the harm that an organization's officers or leadership could break the law and cover it up?

What's the harm that a government could break the law and hide the truth from their electorate?

Maybe it's just me, but these all seem pretty fundamental to ensuring that society continues to have trust in the rule of law and that society continues to function.

But now we get to the interesting bit... How do we configure technology to help prove that a company isn't manipulating digital evidence behind the scenes? The obvious answer is to lift the basic principles of the blockchain - or similar - and apply them more broadly. Delete an email, or slightly alter it's wording? The hash won't line up - and the presence of doctored audit material becomes instantly visible.

Who would one of the most important providers of this sort of technology be?

Microsoft.

Umm...

In other news, there's a hole in my bucket, dear Liza, dear Liza.

https://en.wikipedia.org/wiki/There%27s_a_Hole_in_My_Bucket

DoE seeking fault-tolerant quantum computer ... by 2028

I Have One Available Now...

... which I will be more than happy to sell for, say, 50 billion.

Now, there's something very special about my fault-tolerant quantum computer, because it *also* comes with a full implementation of the Schrödinger-Heisenberg Uncertainty Principle Engine - included for no extra charge.

The SHUPE is a particularly sophisticated piece of technology, but it does mean that you won't actually be able to open up my incredibly Fault-Tolerant Quantum Computer to see how it works, because the moment you open it's box, all the interior logic will immediately move to a special quantum state, known as ... vanished.

What's that you say? It looks like a cardboard box covered in sticky-backed plastic?

Well, only someone ignorant of the latest, cutting-edge quantum computer build techniques would think that!

Fedora 45 beta drags the Linux console into the 21st century

Re: RS-232C "Hub"

Exactly this...

Back in the day, as the industry was making the switch [no pun intended] from e.g. largely serial-port-driven connectivity and synchronous [polled] lines to green-screen terminals, the big network vendors of the time - your Bay Networks, Racal Datacom, Wellfleet, Nortel, offered families of flexible hubs that all looked a bit like this:-

https://www.txo.com/products/popular-telecom-platforms/infinera/infinera-dtn-roadm-platform/

The details varied, but in general terms you'd have a 19-inch rack-mount chassis that could take anywhere up to 8-12 "blades" or "cards" of network connectivity.

I worked with some [elderly] Racal Datacom "INX5000" hubs like this. They had a mid-plane chassis, meaning that each card consisted of a "front half" - where all the logic and expensive stuff was placed... and a "back half" - where the physical connectivity was presented. The sort of cards offered included e.g. RS-232 Serial, 10Mb Ethernet "dumb hub"; 10Mb Ethernet Switch and so on. These things were designed from the ground up to support multi-protocol local area networks, at a time when no clear standard had emerged, and most companies had a hodge-podge of competing designs and protocols that needed to be made to "play nice" together.

The solution still in use today takes something like that, fits a "serial card" to the chassis, then connects each of up to e.g. a dozen serial ports on the card to a different "Console Port" on a Unix host. Even though these were serial cards, they were designed for this use case and you could associate multiple VIPs with a single card so that you could effectively telnet directly from your PC, over the lan and out through a serial port designated by the destination IP address.

Remarkably useful before the days of blade servers and virtual machines...

Re: useful

Hmm... This might be [probably is] a bit pedantic of me ... but wouldn't the distinction between a "console" and a "terminal" really boil down to the physical connection to the host?

We still run literally hundreds of machines with "consoles", but with them all racked in a data center somewhere, the way we access them is via an RS232 card in a network hub that then connect to a specific serial port presented on the back of the host concerned. You can of course still have hardware that presents multiple RS-232-C sockets on the back of it, in order to communicate with the aforementioned Wyse-30 terminals [VT-100s... VT-220s... whatever] but at IPL it would only be the console port that came up with an admin login... and it would be the console port that had to be used in the event that general terminal support failed to start.

It's been a minute since I worked in that area of infrastructure, but I have this idea that the distinction between a console and a terminal really did boil down to "hardware addressability".

Do I have that right?

Ex-FTC boss Khan urges Uncle Sam to break out the handcuffs for AI CEOs, citing 1934 precedent

Re: She is right

Not so sure...

I think there are two reasons that President Trump took an interest in Greenland - and Canada...

First, those two nations are excellent sources of Rare Earth Metals - REMs - that are critical to US semi-conductor interests. If you look at the timing of the President's sudden interest in Greenland, you'll see it came directly after he met with the "Tech Bro's", shortly after becoming President for the second time.

Second, if you look at his broader expansionist ambitions, there seem to be connections back to the pitch that President Putin put to him during their Alaska summit... where Putin described "Spheres of Influence" and portrayed Russia's incremental annexation of Ukraine as being a "necessary buffer" to hold back NATO - which is complete and utter tosh, of course... it's all part of his tsarist tendencies ... but my sense is that Putin sold President Trump on this idea of imperialist expansion, maybe going so far as to point out that expanding America's borders by acquiring new territory. President Trump has made similar references to the land of the Panama Canal... he's not only talked, he's made social media posts suggesting that "Cuba's next".

My personal hunch is that Putin used President Trump's naked ambition to be "the greatest US President ever" and suggested some ways that President Trump could achieve that goal - ways which, of course, would directly serve Russia's interests...

... a lot like the way that Elon Musk sold President Trump on the idea of saving $2 Trillion from the US budget - got himself appointed "Head of DOGE" - and then seems to have immediately terminated the employment of all the staff at federal agencies that had been involved in cases against any of his companies. President Trump got suckered by them both - Putin and Musk ... far from being the greatest deal-maker, he seems to be something of a patsy...

UK's Digital ID walks into a bar... two months after being killed off

Red October...

In the 1990 movie from John McTiernan, based on the book of the same name by Tom Clancy, there is a brief and lighter interlude, a conversation between the Red October's Captain - Borodin, played by Sam Neill... and his superior, Marko Ramius, played by Sean Connery. These two, along with a small contingent of some of their officers, have just set in motion plans to not only defect from the Soviet Union, but to take their entire submarine, the "Red October" of the movie's title, with them...

Ramius is relaxing in his cabin when Borodin pays a visit:-

Borodin: "Do you think they will let me live in Montana?"

Ramius: "I think they will let you live wherever you want."

Borodin: "Then I will live in Montana... and I will marry a round American woman, and raise rabbits, and she will cook them for me. And I will have... a pickup truck... Or possibly, even, um, a recreational vehicle.... and drive from state to state... ... Do they let you do that?"

Ramius: "Oh yes..."

Borodin: "No papers?"

Ramius: "No papers - state to state..."

On November 9, 1989, the Berlin wall came down, with citizens using hammers and chisels to literally tear down the concrete. This event was the spark that ignited the flame that led directly to the reunification of Germany in October 1990 and marked the end of the Cold War in Europe. After this, we came to learn just how oppressive the East German regime, under the Stazi [Secret Police] had really been.

Back at that time, of course, the propaganda being pushed by all Western governments was how oppressive and cruel Communism was. How people had no choices, no freedoms. How so many in Communist states lived below the poverty line and struggled for food and shelter. How it was impossible to move freely; how "papers" could be demanded at any time and the "freedoms" the west was so proud of could be taken back at a moments' notice. How even citizens could be "snatched away" by goon squads - and "disappeared", never to be seen again.

And here we are, 36 years later.

Despite western nations being among the wealthiest "per-capita" in history, we have unprecedented levels of poverty, we're now also being tracked in terms of everywhere we go, everything we say and everything we do:-

- Our [LG] televisions spy on us - literally "telescreens", right out of George Orwell's "1984" dystopian warning...

- Our phones track everything we say and do - monitor where we go, when and for how long...

- Every call we make and receive is recorded...

- The metadata for every email we send is recorded by ISPs... assuming they weren't keen to spy on us in the first place...

- Email providers parse the text of emails in order to serve tailored advertisements...

- We're being told that we should give up "wasteful, expensive" physical money and instead move to all-digital cash - which could be "switched off" by a government, at their whim...

- Every company we deal with either harvests, sells, buys, abuses or otherwise manipulates our personal information*

- In the US at least - certainly at the moment - anyone can be snatched off the streets and thrown in to detention centers - or shot and killed, for that matter...

Yet, apparently, we live in a "free country".

Snort.

* Any time you would like to get an understanding of just how invasive company surveillance of your life has become, take a look at the Terms and Conditions from your bank. You'll find that they grant themselves the right to monitor your private lives through social media and "other data sources" in order to ensure that you remain a "responsible, creditworthy borrower". In other words, they will absolutely scan your Facebook content to look for details of how and where you spend money. If they think you spend beyond your means, they will still let you get in financial trouble, because they want you in debt, but they will control the amount of debt to a rate where you can afford to pay their high rates of interest... Or your car insurance company... on and on... And remember, any information about you captured and held by a private company can be demanded by a government, often without you ever being told.

Europe's right-to-repair rules are broken, not beaten

It's Getting Worse

Car manufacturers are now taking a trick from Tesla... They are building cars where a number of optional extras are built in at the factory and are then "activated" at the dealer when a buyer asks for that particular option on their new car. However, when that car is part-exchanged to the dealer, the first thing the dealer can now do is deactivate all those optional extras and make any new buyer pay for the same features all over again, even if they were installed on the car at the factory.

Every industry is furiously working out how to maximize their profits at the expense of everyone and everything else.

And remember, repairing existing technology will generate a fraction of the revenue that a "new sale" will make the same corporation. It is literally in their financial best interests to convince you to throw away your existing product [like a set of earbuds which no longer hold a charge in their batteries] and buy a new set, even when you could buy replacement batteries for under a fiver.

If we do make a push for "right to repair", the regulation has to include specific references to components that have a "timed life" - things like integrated batteries - and the products that use them must offer the means to replace parts like these as and when they wear out.

We're all going to need to brace ourselves for pearl-clutching and shrieking from the manufacturers, who will fret about "sub-standard repairs causing them brand reputation". Unfortunately, trying to organize a mass boycott of nasty companies like this is just going to be an exercise in futility...

Re: I'm glad somebody else sees it that way

"And I don't mean some arbitrary small percent limit, which then becomes a cost entry on an accountancy spreadsheet. Fine them first pour encourager les autres, then fine them 25%. Word would very quickly go round. This is not being harsh to vendors, it is being fair to consumers.!

I like the basic premise of your proposal, but can I respectfully suggest a slightly different inducement?

If a product can be shown that it cannot or does not support "right to repair" - and we'll need to come up with some terms to define what we mean by that - then how about an outright ####ing ban on all sales in the EU until the product design is changed to allow repairs?

The only way, the only way that these huge, global corporations are ever going to act in a responsible, ethical way is when they are left with no other choice.

As the old saying goes: " when you've got them by the balls, their hearts and minds will follow" ...

The myth of killer AI is a self-serving attempt at regulatory capture

A good example...

... of regulatory capture can be found in the aircraft manufacturing business.

At the time of the 737Max air disasters, Boeing - like all other US aircraft manufacturers - subject to federal oversight through the on-site working and scrutiny of FAA Inspectors.

But.

Those inspectors were recruited by Boeing, paid for by Boeing... and only sent off to "FAA Headquarters" for their training.

If you haven't watched it and have a Netflix account, watch, "Downfall: The Case Against Boeing" - and be literally stunned as you watch first-hand witness testimony from workers and Quality Managers at Boeing who describe how serious, safety-relevant parts had been removed from the production line for failing to meet defined quality standards... and were subsequently retrieved from the "defective" storage area and *put back on customer aircraft*. There are literally interviews in that documentary with shop-floor workers freely admitting they have told friends and family not to fly in 737 Max aircraft...

When your "regulators" are hired and paid for by the companies being regulated, that is what you can expect.

Best guess as to what is happening in the AI space is that the leading companies have woken up to the fact that regulation is going to happen whether they like it or not and this is very simply their attempt to get out in front. For example, they're going to tell Congress that the field is so complex and moving so quickly that the best way to regulate it is to have the regulators trained by the specialist companies themselves. Then they'll suggest that they should select and hire the specialists, because the skills required are so demanding.

Meanwhile, they're going to tell the regulators that, in fact, AI is so incredibly dangerous that really, no new AI companies should be allowed, only the mature, advanced ones with all the expertise. They're going to argue against open source, open weight models that can be run at home on your personal hardware, arguing that such models would be far, far too dangerous.

And, of course, at the same time we're seeing search companies and other data-providing web sites rush to implement captcha and other mechanisms, as more and more end users start to host and run their own models - and the models are using e.g. Google Search to expand their own knowledge base.

OpenAI, Anthropic and X/AI see all this as a way to cement themselves at the center of an industry while preventing new entrants.

Their technology is so dangerous, apparently, that it should not be entrusted to anyone else, only to them. Only to be used in the most extreme, dire emergency.

It's like they're all challenging Denethor from "The Lord of the Rings". The Steward of Gondor tried to seize the One Ring from Frodo. When Gandalf asks Denethor what he would have done with the ring, he replies,

“It should have been kept, hidden, hidden dark and deep. Not used, I say, unless at the uttermost end of need, but set beyond his grasp, save by a victory so final that what befell would not trouble us, being dead.”

Except, of course, Denethor judges that he, above all others, has the right, the authority, to determine when that time has come.

Yeah, right, these companies have moved from humble-bragging in to full-blown teenage angst. Maybe someone fed their models copies of the Twilight Saga?

Snort.

Security through obscurity is dead, and AI delivered the fatal blow

Re: Only Scratching The Surface

Can't - sorry - all done for an employer - it's their IP as they paid for me to do the work.

Having said that - there are lots of good guides on line, for all the popular OS and Distros.

Only Scratching The Surface

As we continue to see more and more vulnerability patches [and zero days and/or exploits] come to light, it would be nice if we could have a sensible conversation around a few more basic, fundamental principles of keeping technology secure...

For example, when in the past I've documented "Standard Builds" [for example for Windows Server 2025 - or it's ancestor versions ] then the build would *always* be stripped down to the bare minimum needed to "get the job done". There are several reasons for this:-

1. Less Bloat == Faster Machine - you typically find that installed software wants to run some kind of housekeeper thread or background service on a server, so even if the package is not being used for anything useful, it's still taking up a bit of RAM and some CPU cycles...

2. Less Code == Less Patching - if you don't have software installed on your machine, you're not going to have to patch it when the next vulnerability is found....

3. Less Code == Less Attack Surface Area - if you don't have software installed on your machine [with a caveat - see below], you're typically presenting an attacker with a smaller attack surface that makes it harder for them to breach your host's perimeter and gain access. The exception to this rule is, of course, security software. Sometimes you just have to bite the bullet and install e.g. a local iptables based firewall, or, for *nix families of hosts, the sort of password synchronization functionality that you get from freeIPA or Red Hat idM...

Unfortunately, we seem to be a long, long, *long* way from there. For example, Microsoft Office 2024 will typically demand at least 4Gb of disk space at installation - and that's before you start loading in optional features. A typical installation of Office 95 used just 55Mb. That's nearly 75 times more disc space for a later-generation version of something that, for 95% or more of users, the original '95 version would deliver everything they need. I make the point not to rail against bloat for the sake of storage [though given the current prices for SSDs, that's probably fair] but for the fact that those 4Gb contain many, *many* more lines of code, each of which could potentially contain a vulnerability.

We should give the same advice to developers for writing software as we should be giving them to handling audit interviews. ("When asked a question by an auditor, give precisely the shortest, simplest answer - no more, no less - and then *shut up*!")

I'm not convinced that telling people to move past "security through obscurity" is going to be enough.

I think a complete rethink of the relationship between "features vs. risk" is required.

LG accused of 'egregious invasion of privacy' over TV data collection

Re: One Moment Please - the Law would like a Word...

I suppose - theoretically - that since Samsung have only a limited [relatively small] number of supported "apps" in the TV's "app store", that it would be relatively simple for a nanny process running on the TV OS itself to scan all local app versions against server/cloud versions and then trigger updates if it sees the server/cloud versions increment.

But that's not necessarily an acceptable approach for a piece of technology - unilaterally taking decisions about what to update and when to update - because the TV - or any other appliance, for that matter - would have no way of knowing whether my internet connection was high or low speed, whether it had failed over to an expensive 5G backup solution because my physical connection had just been lost, etc., etc.

If we had an appropriately informed/authorized executive from Samsung in this discussion, I suspect they would claim that the approach I witnessed helps to eliminate "user inconvenience" of being in a situation where I switch the set on and then the first thing it wants to do is run updates.

But this is a more sophisticated community than that; we understand that updates can have priorities and severities assigned and unless an update to a package I never use is "critical" - i.e. puts the rest of the set at risk of being compromised with e.g. a firmware rewrite attack - then the set should damn well mind it's own business and let me run updates if/when I want to.

The reason that appliances like this *don't* behave like that is because they don't want you to have any inkling of what is happening in the background when you're not looking. Especially when e.g. Steve at Gamers Nexus shows how LG TVs are literally listening to conversations and uploading text transcripts to a remote server.

Pretty much the only option left is to turn the thing off at the wall via a physical switch.

Mars astronauts could live in houses made of yeast and jello, say scientists

Re: Just add water?

Mars does have water - just not in the "conveniently available" form or location...

There is an abundance of water ice at the poles... but that would be an sub-optimum location for a base, because even though the atmosphere on Mars is very thin, it's still enough to reduce the effect of the solar panels that would be needed to power aforementioned base. Note, of course, that with Mars being a lot further from the sun than Earth, the effectiveness of solar panels is also substantially reduced [because fewer photons-per-square-metre-per-unit-time will arrive on a panel on Mars].

There are also millions of cubic kilometres of ice buried beneath the Martian soil at mid-latitudes... and even near the equator. This would be much more practical as a source of water for a human outpost, but of course it does mean that there will be effort involved in getting it extracted for use.

Bonus nugget for the nerdy... the atmosphere of Mars, thin as it is, is made up mostly of carbon dioxide. There is a chemical process - a two stage process, in effect - by which water can be combined with carbon dioxide to create methane:-

First, you split water through electrolysis, to produce hydrogen and oxygen - very straightfoward, well understood, and achievable with little more than water and electricity.

Second is methanation, where you react the collected hydrogen gas from the first stage [conveniently leaving you the oxygen from the first stage as a by-product], combined with carbon-dioxide, to produce methane. This second step uses the Sebatier Process [it requires temperatures of 300-400 centigrade, a pressure of perhaps 440psi or 30 bar and a nickel catalyst,

If you perform all the above steps and then go on to liquify your methane and oxygen, then you have created the two components [fuel and oxidizer] used by Musk's Starship rocket...

Reality check: on Earth (one Standard Earth Gravity), a fully stacked SpaceX Starship uses a total propellant mass of between 4,600 and 5,250 metric tons, depending on what they're testing at the time. The nominal ratio is 3.6:1 of oxidant to propellant - 78% of the total mass is oxidant and 22% is the methane. OK, the gravity on Mars is roughly 1/3 that of Earth; Mars has a *much* thinner atmosphere, so atmospheric drag is "much less of a thing" on Mars than on Earth... All that works in the favour of a Starship rocket attempting to leave Mars to return to Earth. On the down-side, however, Mars rotates at roughly half the approx 1,000 miles per hour that Earth rotates at the equator... which means that in order to achieve orbit, more of your orbital velocity [and thus escape velocity] has to be achieved from fuel burn.

I add all this off-topic detail to give an illustration of what it would take to achieve viable two-way traffic between Earth and Mars... and the fact that your transportation system is going to place *much* bigger demands on Mars' resources than a human colony. It also means, of course, that it would be necessary to deliver quite a lot of equipment to Mars as well, to extract, process and store it all safely.

Re: Porous"

Yes, Mars has an atmosphere.

On average, however, Martian atmospheric pressure is about 0.088psi or 0.006 atmospheres.

There are a few reasons why this is the case - and not all could be described as having met a consensus threshold, but in simple terms, for a planetary body to keep an atmosphere, it seems like there are a bunch of contributing factors:-

- A magnetosphere - if the planet has magnetic fields - plays a huge role in helping to retain an atmosphere, because the charged particles from the sun's solar wind can strip away an atmosphere that isn't protected by one.

- Planetary Mass - if the planet has sufficient mass, that can help anchor the gaseous atmosphere to the world...

- Vulcanism - one of the key contributors to atmospheric retention over the longer term, volcanoes are responsible for disgorging gaseous content in to the atmopshere that can then be processed by any organisms - plants, animals, microbes - living on the surface.

The reason for the word porous in this context is most likely because even a well-made, 3D printed habitation unit or structure is likely to be "gas porous" in that it is unlikely to achieve a perfectly gas-tight seal across the entire inner surface. As a result it is likely that any mission that intended to occupy 3D printed habitats on Mars would need to line them first - likely with a spray-on material of some kind and almost certainly [initially at least] something carried to Mars from Earth.

If you want to be a bit more nerdy about it... and why not... Mars does have a plentiful supply of sulphur - and this could be used, in conjunction with pulverized local rock, to produce a passable concrete. In fact, it would likely be quite a bit simpler to base a structure on the sulphur-driven approach [assuming - big assumption - a plentiful supply of that element near your proposed base camp] than trying anything more complex.

Also note that no amount of wishful thinking is ever likely to be able to "terraform" Mars - there's a lot of [fictional] discussion around generating a large amount of carbon dioxide, for example, to help form enough atmosphere to hold in solar warmth. That's because of the lack of 1) a magnetosphere; and 2) enough mass to hold the atmosphere in place.

OpenAI's website-hijacking swarm reached far further than we thought

Has Enyone Else Noticed...

... that in the last 48 hours or so, Google Search has started to throw up Captcha challenges in response to sentence-like search terms?

Color me cynical, but I have a hunch that this is in response to a bit of a sneaky change in the end-user-scale clients deployed by the big AI companies.

I've had updates to my Claude client pushed on September 9th, 6th, 3rd; August 19th, 18th, 14th and 12th... but starting around September 3rd, I began to see general queries to Google Search being met with Captcha challenges.

I can't help but wonder if these client packages are now using "bundled browser elements", so that when the client tells you that it is performing a search, it is in fact using browser libraries on your workstation to perform the search on the front end rather than the back end, so that those searches don't get blocked by the likes of Google and other web sites.

After all, there is no legal requirement for e.g. Google to make their search engine available to OpenAI or Anthropic... and all Google would have to do would be to track down the registered IP addresses of those companies and simply block or throttle requests. And in a sense, that might not be unreasonable - if you were Google and you saw searches spike 400% but concluded that most or all of that came from AI agents rather than humans, maybe you'd want to either stop the abuse or get paid for your services.

Suspect this is going to get worse...

Re: Under Oath

Footnote:

https://www.motherjones.com/politics/2026/09/openai-congress-senate-hawley-casar-committee-probe-investigation/

Joke. Total and utter joke.