Posts by Fonant

Single passenger reportedly survives Air India Boeing 787 crash

There's a useful video showing the take-off. The plane was climbing as normal, but then stopped climbing while the plane's attitude hadn't changed. So almost certainly a double engine failure, at the worst possible moment. A bit earlier and the takeoff could perhaps have been aborted. A bit later and the plane might have had enough altitude and speed to land somewhere more forgiving.

So the question probably is whether it was a double bird strike (there are reports that this airport has a bird strike problem), or a catastrophic failure of some sort that unusually affected both engines simultaneously. These planes have a LOT of redundancy built in, and the engines are extremely reliable. So my bet is that they flew into a flock of large birds, which took out both engines.

Linux Foundation tries to play peacemaker in ongoing WordPress scuffle

Re: There is a YouTube video of the announcement

Link to the actual announcement:

https://www.youtube.com/live/IOsOAsbD1gA?t=9402

(you can post some HTML to this forum, and YouTube links can include start seconds :))

Reddit sues Anthropic for scraping content into the maw of its eternally ravenous AI

Those bloody AI bots keep loading my servers with their excessive page requests. They soon get blocked, though. I don't want data from my machines powering bullshit generated by others.

The "AI" bubble can't burst soon enough!

Please tell us Reg: Why are AI PC sales slower than expected?

Of course ChatGPT designed a logo similar to existing logos. That's how "AI" works - it's all about statistics and pattern matching.

It's really good for generating plausible-looking output. That's all that it does. Be careful not to think there is any "intelligence" involved.

Ex-Meta exec: Copyright consent obligation = end of AI biz

forcing AI companies to ask for the permission of copyright holders before using their content would destroy the AI industry overnight.

Excellent! Do it!!

Everyone's deploying AI, but no one's securing it – what could go wrong?

Conflation of issues?

There seem to be two separate issues caused by "AI" here:

1) The blind-faith use of "AI" in business systems. Clearly a Very Bad Idea to insert automated bullshit-generation into your business. Especially if that bullshit might possibly be modified by external agents to be malicious.

2) The use of "AI" bullshit-generation to find vulnerabilities. Doesn't seem to me to be a huge additional threat, but perhaps AI hallucinations might find vulnerabilities that humans haven't yet spotted? Could well be a Good Thing overall, in that more code vulnerabilities will be discovered sooner.

We'll ignore, for now, the much bigger potential security problems as AI-generated fake information seeps into governments, military, education, and other parts of our lives.

Amazon tested warehouse robots and found they're not ready to replace humans

Re: Capitalism is dying

Pure Capitalism has been broken for ever. Pay under Capitalism is inversely proportional to work effort and value to society: unless we deliberately tax the rich heavily and invest heavily in public services.

Look at the people being paid the most: are they all working long hours at significant personal effort, doing essential tasks to help society? No, they're the ones who are avoiding taxes, and making money simply by having money. Unfortunately they're also the ones who hold significant power over politicians...

Solutions:

- Universal Basic Income.

- Tax rates on highest earners as they were in the 1950s and 1960s. That's 90% top rate on the excess earned by the very few.

- Heavy investment in public services: benefits everyone, drives the economy, no direct effect on inflation because public services aren't profit-making.

Paul McCartney, Elton John, other creatives demand AI comes clean on scraping

Far Too Late

The creative works have already been stolen, not only from published authors but from everyone who has written anything on the internet.

The only hope is that humankind finally realises that AI-generated slop is no more than bullshit, and the whole destructive "AI" bubble bursts with a big bang.

Yolk's on you – eggs break less when they land sideways

Re: "Through hundreds of experiments"

Egg basket panniers for cats, with the eggs in the correct position, obviously!

Nvidia rolls out NeMo microservices to help AI help you help AI

Poishing Turds?

Since AI is generating bullshit, plausible output that may or may not be accurate or true, this looks like an attempt to polish that bullshit.

The AI people need to work on ways to guarantee that AI generated content is accurate, and cannot hallucinate. I'm not holding my breath!

They also need to reward the humans who generated the content they used to train their AI models with.

Microsoft: Why not let our Copilot fly your computer?

Yay, let a bullshitter have a go on your computer!

AI is bullshit generation: generation of plausible but not necessarily correct or true output.

It's great if you need some bullshit.

It's a complete nightmare if you require something that is correct/accurate/true.

It's fun making Studio Ghibli-style images with ChatGPT – but intellectual property is no laughing matter

"AI" is simply bullshit generation

Most people instinctively know that "AI" is just bullshit generation. The sooner the "AI" bubble bursts, the better.

Bullshit, used on an industrial scale, is extremely harmful to society and even, perhaps, humanity.

M365 Family users wake up to notice 'Your subscription expired'

The Cloud FTW!!

"Subscribe to cloud services", they said, "it'll be super-reliable!".

Heh, local copy of LibreOffice here on my Fedora desktop. Haven't used anything Microsoft for years now, and I really don't miss any of it!

Copyright-ignoring AI scraper bots laugh at robots.txt so the IETF is trying to improve it

AI is bullshit-generation. The bubble will crash once a large enough majority realise this.

In the meantime, AI bots get blocked by IP on my servers. Not only to stop them stealing content, but also to stop my servers getting overloaded!

Trump tariffs to make prices great – a gain

Re: "accept the wages offered in less developed nations"

And a "trade deficit" is NOT a Bad Thing. It doesn't mean you're making a loss!

He's not doing it to impose tariffs, he's doing it to stop others from tariffing USA.

Except adding tariffs to imports only encourages other countries to either (1) increase prices to match or (2) stop trading with you or (3) impose retaliatory tariffs on your exports. Which is exactly what is happening.

Either Trump is a complete idiot, destroying the USA by mistake, or this is all deliberate: the main winners of crippling the US economy and trade will be the ultra-rich and Russia.

From an American perspective what is wrong with that?

Increased prices for everything imported by 10% on average. That's going to increase your inflation, on everything you buy, noticeably. Hence the stock market crashes we're seeing.

The Trump government's official explanation of the new tariffs means that they expect prices of imported goods to the USA to increase by 10% on average.

"Explaining the Trump Tariff Equation" by Stand-up Maths: https://www.youtube.com/watch?v=j04IAbWCszg

EU: These are scary times – let's backdoor encryption!

Impossible

You can't un-invent encryption. End of message.

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

Re: If tariffs are so bad why does the UK / EU etc have so many of them?

Tariffs in themselves aren't either "good" or "bad", they're a tool to control trade. Mainly to protect a country's own businesses against being undercut by foreign competitor producers with cheaper labour or fewer safety/health regulations.

Suddenly imposing large tariffs on everything your country imports from anywhere else in the world is, erm, "unusual". Trump is saying to US citizens "if you want to import anything, from anywhere, you're now going to have to pay the US federal government lots of extra tax on it".

Re: Trump is easy to model

To get into the top bracket you'd need to be on the equivalent of about $4.3 million today. Back then this accounted for VERY VERY few people.

Which is exactly how taxes should be organised. Zero tax for the lowest-paid, and exponentially increasing rates for the ultra-wealthy.

The ultra-rich don't need their $billions to have a happy life, heck even $millions is way more than enough. The very-few ultra-rich are simply hoarding money that many other people could be using to do good things. Hence the sense of 90% tax rates for top earners.

Wealth has a natural positive feedback mechanism (especially when charging interest on loans is permitted). The more money you have, the easier it is to make yet more money. Also the more debt you have, the more difficult it is to get out of your debt. To counteract this, civilised nations have taxation. To counteract obscene inequalities in wealth, you need very high tax rates for the very wealthiest.

Taxes are a Good Thing if you want a pleasant society.

Microsoft's many Outlooks are confusing users – including its own employees

Re: What about Outlook (less shitty)?

Like, perhaps, allowing the account username to be different to the email address for POP3/IMAP/SMTP servers. Without needing to find the route to get to the old settings UI?

Vivaldi 7.2 browser wants to topple tech's feudal lords

Re: Vivaldi uses Chromium

So far, so good. It's my main browser, not least because of the integrated mail (much like the good old M2 in Opera).

AI crawlers haven't learned to play nice with websites

Yes, AI scraper bots have been hammering my servers too, on and off for many months now. I've managed to block the majority with Apache based on their UserAgent names, and by firewall for the worst offenders.

Go away, bullshit generators are not welcome here!

Apple's alleged UK encryption battle sparks political and privacy backlash

Re: If I remember right

That's part of the point of E2EE for your iCloud data. Even if there is a break-in, or rogue employee, no-one can read your data.

The UK Government seem to think that it's worth the potential exposure of everyone's personal data just to be able to read the personal data of "bad people". Where the definition of "bad people" is a function of time.

AI running out of juice despite Microsoft's hard squeezing

Re: As much as I agree with the sentiment that AI is overhyped...

perhaps AI isn't all that it has been cracked up to be.

Spolier: it isn't. Nowhere near.

LLM "AI" is simply bullshit generation

"AI" is very useful for everyone who needs some plausible (might not be true) text or images or video.

"AI" is completely useless for anyone who wants outputs to be accurate and/or true.

"ChatGPT iS Bullshit": https://link.springer.com/article/10.1007/s10676-024-09775-5

Dash to Panel maintainer quits after donations drive becomes dash to disaster

I've not had many problems with it over the last several years. Running with Gnome on Fedora. But then my anecdote ≠ data.

I decided that Dash to Panel was worth a tenner, so donated. The donation button was intrusive, but very easy to dismiss.

Microsoft goes native with Copilot. Again

Linux

See title.

It begins: Pentagon to give AI agents a role in decision making, ops planning

AI?

Are they talking about some different type of "AI" than the currently-overhyped LLMs?

"making Thunderforge’s reasoning process legible, so users can trace its logic"

I don't think anything LLMs do can be classed as being "reasoning" or "logic". They're statistical models based on words. Bullshit generators.

Payday from hell as several British banks report major outages

Websites being hammered by AI web crawlers?

As title. Lots of AI web crawlers out and about at the moment, hammering websites as fast as they can.

Does terrible code drive you mad? Wait until you see what it does to OpenAI's GPT-4o

LLM = Bullshit-generator

LMM generative AI is merely bullshit generation. Plausible outputs that may or may not be accurate or true.

https://link.springer.com/article/10.1007/s10676-024-09775-5

Signal will withdraw from Sweden if encryption-busting laws take effect

Re: One of many ironies

PGP.

Encrypt your message (or one-time pad) using your recipient's public key, and sign with your private key. Only the recipient can decrypt the message, and they can verify that it could only have come from you.

Humanity cannot un-invent encryption, however much authorities try to do so with legislation. It's just maths.

The software UK techies need to protect themselves now Apple's ADP won’t

Re: What Apple should have done

Or, perhaps, encourage a third-party app to add a shim/proxy layer that locally encrypts the data before sending to the iCloud, and transparently decrypts the data again when it's restored. ADP but provided by someone who isn't troubled by the Snoopers Charter.

Security Theatre at best

Since it's impossible for humankind to unlearn how to do secure encryption, it's impossible for anyone to ban encryption.

Encrypt your files locally, before uploading them to iCloud. Heck you could even use steganography to make them look innocent. Let Apple give the UK/USA/RU government access to the files you uploaded: they won't be able to decrypt them without a lot of trouble.

AI summaries turn real news into nonsense, BBC finds

Generative AI is just a bullshit generator

Generative AI can only create bullshit: content that looks plausible but may or may not be correct, accurate or truthful.

https://link.springer.com/article/10.1007/s10676-024-09775-5

UK government insiders say AI datacenters may be a pricey white elephant

Generative AI, from LLMs, is just bullshit: no accuracy or truthfulness guaranteed. Great for generating bullshit where the content doesn't matter, but mostly useless for anything else.

"AI" already has a bad reputation amongst the general public, who have quite sophisticated bullshit-sense. When will government realise they're being conned by Big Business?

DeepMind working on distributed training of large AI models

Still Bullshit

More efficient bullshit generation? Excellent news!! </sarcasm>

Why UK Online Safety Act may not be safe for bloggers

Re: VPN to connect the family "intranet"[1]

Update: Schedule 1 Part 1 Paragraph 7(3) says:

“business” includes trade, profession, educational institution or other concern (whether or not carried on for profit);

So it depends on what "other concern" might include. Clubs? Family groups?

Re: Storm in a Teacup

Speaking as a forum admin, I agree on a top-level basis. I am not shutting down my 18 year old cycling forum, at least not yet.

I've got very close to deciding to close it, though, while listening to the series of Ofcom presentations on the OSA earlier this week. The more you delve into the details, the more vague the whole thing is.

This Act creates a whole new set of legal duties that I have to comply with. With legally-required risk assessments that are entirely my responsibility.

It has taken weeks to get a basic understanding of what the Act says, and the duties that Ofcom is imposing on all user-to-user service providers. We still don't have answers to several fundamental questions, and the whole thing is written from the point of view that all online services are run by businesses (The "nominated person responsible for risk assessments" has to officially report their findings to a more senior person, for example. While "internal business user-to-user services" are exempt, internal services used by any other group of people such as families or clubs, are not exempt).

The GDPR was quite different. It did not cover data for personal or organisational use: we didn't have to register our address books, or our club membership databases, with the government.

The OSA most definitely does cover online user-to-user services, even if restricted to family groups, or clubs and societies. Run a small club-only forum? Comments on your website? You need to read all the reams of documentation, try to answer the questions in their "tool", carry out a risk assessment and a child access assessment, document it all, update your terms and conditions, create new procedures for people to report content and complain, add your name as the person responsible: then cross your fingers that your understanding of your new legal duties is correct.

Re: is the existing act fit for purpose?

From what I've read of the Act, I don't think "running a service which has had illegal content posted on it" is a criminal offence.

The act of posting illegal content is quite probably a criminal offence.

The offences in the OSA are all concerned with service providers refusing to assist Ofcom with their enquiries.

Re: Run and hide

Good Point.

If the presence of off-topic comments, and comments replying to other users, make a comment section "in scope", then El Reg will need to comply with the OSA.

Along with trillions of other websites around the world. <jazz voice>Nice!</jazz voice>

Re: Does anyone know how this apples to IRC?

* For any non-business use? Probably in scope.

Re: ...chances of Ofcom even discovering...

I hope Ofcom are ready for a rush of vexatious complaints, and have the staff, capacity and incentive to investigate them all. Spoiler: they don't.

The Ofcom message to small services is: "show willing, and we'll work with you if it ever becomes necessary".

See also: federated user-to-user services like Mastodon.

Ofcom are currently "thinking about" how the OSA applies to mastodon server providers.

With any luck the excitement will die down soon, and we can quietly carry on as before. If not, a lot of benign stuff will be turned off. Or will go underground, well out of reach of government view.

Re: VPN to connect the family "intranet"[1]

The same way you define "significant number of UK users", and "low risk of illegal content being posted", and even what is "email" (since for some reason "email-only" user-to-user communication services are exempt from the Online Safety Act).

This is the problem with laws. The only way to find out what the law means is to ask a judge in a court of law.

Re: VPN to connect the family "intranet"[1]

Other examples mentioned at the Ofcom presentations were things like local WI group discussion boards, forums for members of a club or interest group.

Even if you hide everything behind a member login system, and restrict who can sign up, it's covered by the OSA unless a "business" runs it for its own internal use.

Re: VPN to connect the family "intranet"[1]

Yes, the "business intranet" exemption should really be extended to "services limited to a small specific group of people" or something like that. My family NextCloud instance is technically covered by the Online Safety Act, it's a file sharing and messaging user-to-user service and not internal to a business.

For now I think the best thing is to carry on as before, the chances of Ofcom even discovering small user-to-user services (within the entire internet!) is minuscule. And they promise to "work with" services and to "be constructive". Fines and criminal sanctions only follow if they think you're obstructing their aims. They do like saying "£18 million maximum fine!" though - I think it makes them feel they have some power that doesn't exist in the real world.

Re: Brown said he intends to put this question directly to Ofcom

Thank you for the time you've spent trying to make sense of this legislation and Ofcom's "guidance" from the point of view of small service providers, Neil.

It's all so vague, and reliant on the service provider's own personal assessment of risk. Risks that cannot ever be quantified.

Copilot+ PCs? Customers just aren't buying it – yet

Re: confusion

Generative "AI" is just bullshit-generation. It produces something that, by pattern matching and statistics, looks very plausible. But no-one can guarantee that the output is either "correct" or "true". It's just "plausible".

I think most people have quite sensitive anti-bullshit senses, and they've already spotted AI-generated rubbish. They have no reason to want to produce their own bullshit: leave that to the advertisers and spammers.