Posts by that one in the corner

Old but gold: Paper tape and punched cards still getting the job done – just about

Re: Fascinating stuff

Good thinking.

Do you allow yourself any markup?

My (digital) notes don't go back anywhere near that far but for the last almost three decades (yikes!) I've been using a tiny wiki (many thanks, Mr Cunningham) running on localhost for note keeping about absolutely everything. The formatted version is easier to read (or there wouldn't be much point!) but all the editing is still directly in plain wikitext[0], so it has to be comprehensible in that form as well[1].

[0] in my foolish youth (well, younger than today!) I did plan to "drop in" a Javascript editor to do all the bold etc WYSIWYG but honestly I'm glad I never did - for many reasons.

[1] having tried too pull together a complete, can be presented to the auditor, document from a slew of pages in Confluence[2][3], I've stuck to a simple, if idiosyncratic, internal format for my own stuff. Especially as Confluence changed formats, for no obvious improvement to the end user, so it was obvious anything that worked this week was doomed as soon as somebody decided "we must always use the latest version" without considering, again, whether both our second-party processing and third-party Marketplace addons would still function...

[2] Sorry, this started as a simple enquiry as to whether jake allowed himself the luxury of, say, a bit of Markdown and ended with a rant about software that doesn't work well as an archival format, IMO, but I suppose that is still on topic...

[3] crap 'export' facilities[4], "Print to to PDF" unable to collate pages, let alone manage top'n'tailing a complete document, nothing showing up in the Marketplace (at the time)...

[4] was able to read the mangled, partially XML, export by finding a parser that wasn't totally compliant to start with and slicing off its malformed input error reporting, mwahaha - fun but worrying if you are thinking of project documentation lasting, and growing, over years...

Re: Fascinating stuff

Did you perchance mean Charles Stross's Glasshouse[1]?

(Actually, I hope not, as then there is some more Stross I've not yet read whilst waiting for the next pauperback to be released)

[1] Warning: that link contains spoilers!

Chap claims Atari 2600 'absolutely wrecked' ChatGPT at chess

Re: Why?

> The tricky bit wasn't encoding the rules of chess and the allowed moves, it was writing algorithms to have strategies to win ... virtually unlimited combinations of piece layouts

Curiously enough, back in the day, that was called "AI research": "Can a computer ever play Chess?". Yes[1]. Then it morphed into "Can a computer ever beat a human at Chess?". Yes[1]. Then "Can a computer ever beat a Grandmaster Human at Chess?." Yes[1]

One way that we can tell that these LLMs are still flailing around trying to actually be useful and cost-effective is that they are still being flogged by the salespeople as marvellous toys that all the cool kids have. If the mechanisms are ever entirely subsumed into the day-to-day "100 Algorithms That Every Effective Programmer Must Know" then we will see that they are actually worthwhile - but by then, guess what[1]/

[1] "It it works, it isn't AI"

Re: Why?

> Try training your AI model on chess game records and you would have, I suspect, a very different outcome.

Well, ChatGPT - like the other LLMs - was trained by pulling in everything that could be found on the web, and every other digitised text source, which includes the entire rules of Chess (many times, in book and web form), more books discussing the Great Players and their strategies and, of course, play-by-play records of games. At every level (chess by email, forum post etc).

So it had all the data that you suggest - and yet it failed.

Now, using *sensible* ML techniques on Chess rules *will* create a machine that can play - you can literally even do it with (big enough) matchboxes, in precisely the same way as you can train matchboxes to play tic-tac-toe - although it will take some time for that mechanism to complete its learning phase[1]. But, logically, it will work. What is the MAJOR difference between that ML and ChatGPT? The matchboxes are trained and punished/rewarded against a specific goal and with specific metrics. ChatGPT - nah.

> Why would ChatGPT be any good at chess?

Why would ChatGPT be good at *anything*? Unlike the matchboxes, it has not been trained with any sane goals - it won't be good at designing airplane wings, or baking bread, or writing Python code.

> not a problem to which general knowledge, or even basic understanding of the rules, gives a great solution.

There are NO problems to which general knowledge is the solution - except for winning pub quizzes or beating The Chase[2]. Knowledge plus nous - that'll get the job done. Heck, nous, time and an inquisitive mind will get over the lack of general or even specific knowledge[3].

So what great solution can ChatGPT or its cousins give us?[4]

[1] the Heat Death of The Universe says "hello"

[2] and the prizes you get for either do not compensate for building an entire LLM.

[3] we call this "research"

[4] well, comments on El Reg do say that it works as a super-ELIZA, but there are cheaper ways to achieve that goal.

Japan's latest Moon landing written off as a failure after ispace probe goes dark

Re: Did they get too close to the Nazi moon base?

> The Nazi ship on The Moon in that book is called Wotan.

(Insert at this point "I must preview my Register comments, especially after Midnight" 100 times)

I had almost had pun to go in here, but couldn't drag it out.

Re: Did they get too close to the Nazi moon base?

Was it Robert Heinlein's Rocket Ship Galileo, published in 1947? The Nazi ship on The Moon in that book is called <a href="https://concord.fandom.com/wiki/Wotan>Wotan</a>.

Re: Only the one lazer?

> light amplification by stimulated emission of radiation

zzzzztimulated[1]

(although that usage turns up more in Earth-based documentaries than in serious descriptions of life on The Moon).

[1] alternatively, the Reg Reader yclept "Sampler" hails from Zuummaaaahzet, but that just seems like clutching at straws (if not sucking on a straw).

Schneier tries to rip the rose-colored AI glasses from the eyes of Congress

Re: centi-billionaire Elon Musk

> It's Friday evening

Sorry, wrong continent.

It was very much Saturday morning. Should have just stuck with catching up with "On Call" but no, have to read this one as well....

Re: centi-billionaire Elon Musk

> NOPE. See https://en.wiktionary.org/wiki/centi-

Your comeback is to reference - wiktionary? You think wiktionary is reliable or authoritative?

That is hilarious!

My post-midnight rant has at least borne some fruit.

PS

Best bit is, amongst their litany of twaddle and illiterate neologisms, they do mention the one common word that does use centi- "badly", which is because it predates the (now far more common, hence "correct", and useful) SI prefixes. To save the embarrassment of quoting wiktionary all you had to do was use some nous and just write the single word:

CENTIPEDE

centi-billionaire Elon Musk

Centi-billionaire? FFS where are you people getting this ridiculous usage from?

If you really must try to slam extra syllables onto words, at least try to remember junior school and use hectobillionaire. Easy enough to remember: "he has how much money? Blooming heck(to)".

I know, I know - you heard someone else use it and it sounded a bigly long word, which makes it clever, so it's better to use it than risk pointing and laughing at the idiot who doesn't know what they are saying.

PS listen to Schneier

Techie traced cables from basement to maternity ward and onto a roof, before a car crash revealed the problem

Easy mistake to make - Peregrine was Head Boy and the name tends to stick in the mind.

Re: At my last place....

> Those old Victorian 3 decker primary schools ... not inclusive for disabled kids

They were perfectly adequately designed to be inclusive for the Victorian kids, at least those in a position to go to a three-decker: "Smythe, nip out and round up a couple of guttersnipes, it's time to get Jenkins up to the Art Room".

Of course, that was just primary schools. For secondary, the First Form were all available for fagging duties, including sedan chairs for School Bully and his unwed Filipino women.

Re: At my last place....

That is what you get for hiring Druids to raise your infrastructure.

Re: vlans & other segregation config might have hidden the issue

This was the most unkindest cut of all

Re: vlans & other segregation config might have hidden the issue

You have to know the font characteristics and the precise grapheme: are we talking hyphen, endash, emdash or "horizontal line extension" (U+23AF)?

But if you find it is all tildes then best to seek the help of a professional.

Musk and Trump take slap fight public as bromance ends

> Considering that Musk basically called Trump a 'pedo guy'[1] ... I don't see the bromance ever getting rekindled

What, not even after Elon shows Donald a copy of the US court ruling that calling someone 'pedo' isn't defamatory in any way? So it was just a bit of joshing, like how best buds jokingly insult each other.

[1] it would be a great popcorn moment if Musk actually did use that precise phrase in reference to Trump

Centibillionaire?

That would make him worth what, 10 million dollars in US-style "billions"?

Unless - can it be, El Reg is going back to its roots, using Proper, Full-Fat, High-Protein, British Billions? Then a centibillion would be translated into meagre yank "billions" as ten thousand million, or as they would say "ten billion".

Still a bit short of his reported wealth - maybe "just" what he'll have left after he pays a few of his outstanding bills?

Ex-NASA Admin pick blames Musk ties for pulled nomination

Space, The Final Tantrum

Posting threats to cancel each other's spacecraft on one's own social media.

The US has gone from "We choose to do this ... not because it is easy, but because it is hard" to throwing rattles out of prams.

Really, really, hoping that Japan's attempts have more success, that India, ESA and everyone else - yes, including China - keep their socks pulled up and their missions flying.

Elon Musk pukes over pork-filled budget bill with Tesla subsidies on the line

> Now he is looking for sympathy! Empathy works both ways mate - but then he wouldnt know about that.

Elon would be really insulted if anyone tried to say he had empathy for anyone/anything: “The fundamental weakness of Western civilization is empathy"

Need for speed? CityFibre punts 5.5 Gbps symmetrical broadband at ISPs

Re: I would be happy...

> My local GPON provider

Bully for you.

But whilst there are alternative providers in a few places in the UK, for the rest of us, OpenReach's infrastructure tends to be it. Or Virgin, but let's not go there. So unless someone gives their location, the only safe answer is to reference the ex-BT people.

Re: I would be happy...

How much are you willing to pay?

Openreach launching 1000 Mbps symmetric full fibre service (posted 28th Feb 2025)

> we expect retail pricing to be in the region of £200/month making this a connection for those operating a business.

Microsoft will stop pestering Windows users about Edge in EU

But will they stop Edge

quietly taking taking over every default "Open with" action every Patch Tuesday?[0]

Although, on the bright side, this has been training me never to blindly double-click on anything under Windows, but to first examine it carefully to see if doing so will still be opening it in my preferred program. And we are always told never to just double-click on something but stop and think "could this be about to launch a virus-laden program that will run and trash my PC?"[1]

[0] no, because I'm not in the EU. Close to it, can almost see it, but not actually in it, so this is just a pipedream

[1] awaiting the obvious response, any second now...

X's new 'encrypted' XChat feature seems no more secure than the failure that came before it

Re: Blockchain

More subtle than that, and he isn't keeping it a secret (at least, that will be his excuse - "We told you from the start")

> "Bitcoin style encryption."

As we all know, the closest the blockchain gets to encryption is the work involved in the hashing of the blocks, which is supposed to be a one-way calculation and proof that the copy of the chain you hold hasn't been tampered with. At least, as agreed by a quorum of all the miners*. Hence the issue with the 51% attack: if you hold 51% of whichever mechanism is being used (proof-of-whatever) then you can modify the contents of the blocks, recalculate the chain and everyone else will take your copy as Truth. The Bitcoin (or other) blockchain is an Immutable Ledger, but only for certain values of "immutable".

So we are being told that not only will your chat history be capable of being made public (at Twitter's discretion - oops, I mean when Twitter responds to a legitimate request from "an authority") it will do so with proof that that IS what you said because, look, our crypto-quality hashes can't be wrong, anyone can verify them. As if they were on "the Bitcoin blockchain". Only, cough, a certain someone happens to hold control over this particular "chain" and it might get recalculated once we have corrected your posts to show our Preferred Truth about what you said.

* Description courtesy of Really Quick And Shoddy Explanations Inc, on the basis that if you, gentle reader, do know how blockchain works then you don't need anything better to follow what I'm trying to say. And if you don't know how it works I'm not going to be able to explain it all in one comment. Either way, just go with me here.

Ukrainians smuggle drones hidden in cabins on trucks to strike Russian airfields

Re: “basic cameras can discern between cats and dogs”

Send you child to school wearing an anti-Tesla safety backpack

Re: Unintended (?) consequences

Ah, that explains the excessive use of "terror attacks" to describe targeting purely military bases.

The "terror" part comes from every lorry and van driver now having to bribe their way past even more Russian internal checkpoints.

Atlassian tweaks licenses to reward those who buy more, but gets its sums wrong

Atlassian customers will enjoy that predictability and simplicity

That is not a phrase usually associated with using Atlassian products...

Microsoft's plain text editor gets fancy as Notepad gains formatting options

Re: Pointless

RTF is still useful for sending out unbloated lightly formatted text that can be opened by all of the word processors out there (or that copy of the wordpad exe that accidentally slips onto newer Windows installs).

Just give the file a .doc extension, the program's can figure out the rest for themselves and the recipient's needn't bother themselves with why your attachments are so much less bulky than theirs.

Get out your copy of Petzold

The old Notepad was basically just a menu wrapped around the Windows edit control in multiline mode and one of the examples from Charles Petzold replicated that (IIRC the example was actually about adding menus, as that is the bulk of the code you needed to add around the edit control).

If anyone can lay their hands on their copy, we can get back to the basic simplicity we love.

PS

I agree with others (see above) that Notepad++ is well worth having as well, especially for the computing literati who inhabit The Register comments, but it still isn't as quick to start up from scratch as good old Notepad - and NP++ is a bit OTT for the casual user.

Odd homage to '2001: A Space Odyssey' sees 'Blue Danube' waltz beamed at Voyager 1

Re: Right on Commander!

I hope it did manage to buy the docking computer - without it, my Cobra had about a 2% chance of making it into the station!

Tesla FSD ignores school bus lights and hits 'child' dummy in staged demo

Re: It is problematic.

> Not all trucks are equal either

>> and verified on you vehicle before you get to join the Train Lane

Leaving aside my missing 'r', that comes under verifying that you are capable of doing "follow the leader" along this stretch - if you can't keep up, you ain't capable of following properly and don't get to join.

As implied by this and other comments, there is the problem of how to stop people trying to push their way into the road train lane; that is left as an exercise for the reader*

* as it largely depends upon what the state of the art is when we get to the point of trying to assign said lane; if by then every car has lane assist, that can be activated to prevent crossing across, for example. But if it were done this year then we'd need something cleverer (or something dumber - a physical lane divider except at exit/entry points for starters...)

Re: "There's a million people who die in car accidents every year"

> The figure for the USA in 2022 is reported to have been just over 42,500

Approximately 1.19 million people die each year as a result of road traffic crashes.

You may want to buy a globe, there is a lot more World out there than just the USA. And Tesla et al are trying to flog their wares into every bit of it.

> If you are going to exaggerate on minor points, it tends to undermine trust in whatever claims you make.

If you are ignorant of the existence of 98% of The Planet Earth it tends to undermine trust in whatever point you are trying to make.

Re: It is problematic.

> Controlled driverless zones on motorways might then be implemented to give drivers a break.

Road trains, with a clearly defined controller (just one Casey Jones) and/or clearly defined (and verified on you vehicle before you get to join the Train Lane) "follow the leader" behaviour in each and every vehicle. One protocol to inform the cars fore and aft how you are doing and that that you are about to enter/exit the train.

Not a random pile of fully autonomous vehicles each making all its own decisions in solitude.

(But is does also worry me, when discussing presentations about road trains, that they all seem to like bunching cars up very close, "for efficient road use", but ignoring what happens when one of them snaps a timing belt or a pump seizes or the turbocharger goes phut ... Maybe better wait until we are all electric and have fewer parts, but even then...)

Re: Statistics

> it will sometimes do that with zero warning, so if you weren't following directions and keeping your hands on the wheel

AND maintaining your best level of full situational awareness (and haven't tucked your foot away under the seat, so it is ready to use the pedals).

In other words, doing everything needed to actually drive, except for the awfully tiring jobs of turning the steering wheel or pressing the pedal (assuming you don't use cruise control).

So if you have to be alert all the time, remind me, what is the value in FSD again?

> good auto focus? I see way too many glasses around

Just want to point out, "auto-focus" as opposed to "fineness of focus".

As we're comparing with mechanical cameras, I have a camera with rather naff general focussing (the images are always soft and there is bad chromatic aberration) BUT it has really good auto-focus, in that it always finds the best focus that the lens can manage. Conversely, I have another body with a gorgeously sharp lens, flat field, colour-corrected within an inch of its life and totally unable to auto-focus at all ('cos it ain't got the wibbly bits to connect the focus motor across the lens adapter, if you want to know the reason).

So, as you point out out, the absolute quality of the human eye varies between individuals and across our lifetimes, as the mechanical bits break down, but the auto-focus gets the best out of what is available.

Meta – yep, Facebook Meta – is now a defense contractor

Re: Still waiting...

Still just the boring places where these sorts of things are/can/might be useful, including but not limited to:

* AR as an occasional HUD for warehouse workers (holding a box in both hands, get the QR code turned into text and/or picture of the contents)

* AR as an aide memoire for people whose memories are starting to fail them (e.g. as above, but recognising objects/markers about the house; like super post-it notes)

* VR for immersive walkthroughs of architecture (not necessarily at a human scale - e.g. 3D CAT scan or whilst operating waldoes for keyhole surgery)

Whether any particular use case justifies the cost is another matter (then again, HUDs into specs have been produced through for decades, if they'd stop piddling about they could be affordable by now and al the niches made viable). So the warehouse worker will more likely just be given a code scanner gun and keep a hand free for it, the house is littered with real post-it notes and so forth. Sigh.

HOWEVER

Note that VR or AR does NOT automatically imply wearing a pair of specs/headset (although that may be more comfortable). You could be looking down a binocular eyepiece (one of those with the big rubber light shield) in which case some of those are already exist, the man on the street just doesn't get involved with such niche uses. E.g. the waldoes example exists (they also put the views up on normal monitors at the same time). Yes, you do run the risk of looking like Mr Spock (or Mr Sulu if you have the fancy "rises up to meet you" version).

Finally:

If we take you at your word, "AI" as opposed to the current hype around big expensive LLMs, then lives have been saved by the use of Machine Learning applied to medical scan. Much, much money has been made by applying the same to geological surveys. And older techniques that originated from AI Research Groups are all over the place nowadays[1].

Would you like another game of Chess, Doctor Poole?

[1] "If it works, it isn't AI", just another software tool.

Re: Technomancers

Oops.

Footnotes [3] and [4] got swapped - although the idea that The Shadows were involved with Jiangshi does raise interesting possibilities for crossover cosplay...

Re: Technomancers

Weird term to use in this context, "technomancer"[0] - it only means somebody who practices divination by looking at technology[1], about as accurately as oneiromancy[1a]. You *could* go with the (mis)interpretation of the suffix as used in modern uses of necromancer, but using zombie machines whose limbs keep falling off as they shuffle[2] towards you isn't that useful.

Technomages[4], on the other hand, taking Clarke's Law seriously, would be strategically useful.

[0] yes, I know, Technomancer, Vory, Abundance... still dislike the usage!

[1] i.e. Gartner

[1a] because most of Gartner's work clearly came from the bottle

[2] although the Chinese Jiangshi[3] "zombie" machines hopping around could be menacing, with fewer bits dropping away, although susceptible to post-it notes.

[3] sorry, nope, don't buy the crazy ideas of the tie-in books, only screen-time is canon: Elric and Galen. None of this "they are doing the bidding of The Shadows".

[4] yes, zombies, not vampires! Come on, if you're going to make the comparison, they behave more like Zeds than Code 5s.

Re: XR devices for the military

Black Mirror, series 3, episode 5: "Men Against Fire"

Techie fixed a ‘brown monitor’ by closing a door for a doctor

Re: Does not sound like any kind of fix to me

> How about some kind of anti-reflective coating for the monitor so you don't get the issue?

How about turning the monitor a bit?

How about painting the door white - or pinning up a sheet of paper?

Anti-reflective coating!

Google co-founder Sergey Brin suggests threatening AI for better results

One must go through a rigorous scientific...

> AI security process that adaptively tests and probes the AI security controls of a model

Oh, great.

NOW you are going to be rigorous and scientific about this, trying to figure out what the bleep your so-called "security models" are actually doing.

The "security models" that were just randomly slapped on top of the pile of nadans that resulted from rigourously dumping piles of junk that were scientifically scraped from everywhere and anywhere? The amazing "security model" that turns out to be little more than a screed pumped in ahead of the end-user prompt which can, surprise surprise, be modified by an "anti-screed". No clever side-channel attacks.

Shame the LLMs weren't subjected to such rigorous scientific probing as they were being built from the ground up. Maybe then you'd not be surprised at what they do.

Mutter, mutter. You know, in my day, we only fed our ML systems with carefully curated datasets and ran probes every day as they were growing up. These kids, don't know they're born.

HP Inc. hastens China exit as tariffs kick a hole in its profits

Re: pardon me?

Dunno, but I can you an "Eh? Aye!" PC or an "Ey Oop"* PC from one of our regional suppliers.

* Good for Smalltalk

China to visit Earth’s ‘quasi-moon’ and bring a chunk of it back home

A diameter of just 40 to 100 meters*

Will start counting the days until Trump announces that Musk will send Starship up to collect the entire thing[1] and bring it back to to get one over on China.

[1] yes, I know, even at best this is out by a factor of around 25 to 30 (depending upon how well it is packed) to fit into the payload bay, but perhaps just give it a shove in our direction?

* insert obligatory "gas or electric?" here.

PS

In all seriousness, good to see more science missions, even if there are sneaking suspicions this may(!) have some political motivations.

AI won't replace radiologists anytime soon

Re: There was an AI that was VERY good at identifying cancer

> It trained on the metadata, not the actual image

That is an absolutely classical mistake to make with any learning system that can't/doesn't provide any explanations for its reasoning. It was a well-known phenomena back in the early 1980s and a major part of the push for Expert Systems back then (not that XPS or ML were brand new at that point, but were becoming better known).

And please don't mention "chain of reasoning" within LLMs or we will Have Words - CoR is junk compared to even the most basic XPS dump!

Re: Same day, different story

With the possible exception that the DERM system in your reference *might* refer to a sensible Machine Learning system (that article is incredibly light on, you know, information), built from the bottom up to analyse the sort of images it wil be fed and (please, please) sensibly controlled and "taught" to give a "no idea what this is" response when that is appropriate.

Whereas all of the stuff mentioned in TFA are various way of taking the famously[1] unreliable LLMs and then trying to quickly slap on top a veneer of "medical knowledge". The results can not possibly even be as solid, reliable and trustworthy as using the same underlying LLMs for writing software[2] because there is at least a huge pile of working code out there to train on[3] but, as TFA points out, there is a paucity of good training data for medical use. And you can test the generated software far more ethically, spinning up a throwaway VM to test it on...

[1] to anyone who reads articles on LLMs on El Reg

[2] see recent articles about "vibe coding", typosquatting on hallucinated package names etc etc

Three ways to run Windows apps on a Linux box

Re: I thought...

You can do GPU sharing with a desktop PC.

First, buy an expensive nVidia "corporate" GPU - at which point I'll stop and just slap a second, cheapest I can find, GPU into the case and read up on IOMMU.

Re: Hypervisors

> With regards to networking, if you wanted to SSH into a VirtualBox VM using the default networking set up, you had to do something like "ssh username@localhost -p 2222".

That is - sort of true.

If you leave the networking type at its default of "NAT" - which is "the simplest thing for a newbie to use to just get a 'normal' PC experience inside the guest, so we'll give it as the default", you *can* then do a load of clicks (or a number of CLI options if running VirtualBox that way) to set up port forwarding from localhost through to the guest's ssh port. In the advanced networking settings portion of VirtualBox. By which point saying "default networking set up" starts to feel strained.

Or just make the one network setting change, by using either of the non-NAT options in the combobox, then use ssh (and everything else) as if the guest was real, not in a VM.

End of technical article? Ok, didn't include pictures of where the combobox is, but it is the first thing in network options, way before you find the port forwarding settings.

Not wishing to be rude, but, um, I was so fascinated by your description that I ended up searching around to find out *how* to set that up: I've tested peer to peer protocols by firing up from five to twenty VirtualBox VMs (the lazy way, through the GUI - very crowded host desktop, with all the PuTTY sessions as well!) and had never even thought of looking at how to fiddle with the NAT in VirtualBox. So I guess I just learnt something new today.

PS just for context, I run KVM pretty much every day on the desktop, VirtualBox on a server (I know, wrong way round!) and am banging head on mapping IOMMU to physical ports on the back of the various cases.

AI agents don't care about your pretty website or tempting ads

Your well-honed customer experience is noise to

Anyone looking for useful information, AI, human, even Salesmen.

In fact, is there ANYBODY outside of your own marketing team who find your "customer experience" to be anything more than a bloody nuisance?

> For example, when a travel website shows image-based banners showing a call-to-action (CTA) – a pitch to book a particular hotel, for example

Then WHO is being served by that being an image? Anyone wanting to search for reviews by copy'n'pasting ("you DARE to mistrust our recommendations?"). Anyone with a screen-reader ("Don't worry, we've used special tags with the text for them" - not being able to see the absurdity in making more work for yourself because you cocked it up using an image in the first place!).

Maybe, MAYBE, one good thing could come of the LLM Frenzy: getting rid of all the stupid, over-designed, over-coded* bullshit websites that infest the Internet nowadays.

* I mean, come on: you specifically wrote code to PREVENT me opening details of multiple carparks in new tabs! Why? What do you gain from that? All I want is to have them available so when we find one is full we know where our next choice is! [Story based on Real Life, looking for parking in Coventry - yes, the real one, in Warwickshire** - this morning. And don't me started on Google Maps, who gave as their FIRST pick when walking from the cathedral back to the car park somewhere on RHODE ISLAND! Gaah]

** You heard me, Warwickshire!

Microsoft-backed AI out-forecasts hurricane experts without crunching the physics

Re: Wait a minute

> an expert system basically

Ah, nope. Expert Systems use far more explicit rules than this does and those aren't classically derived from Machine Learning.

But, yes, this does use techniques that are also found in an LLM*, but it isn't an LLM.

* and non-trivially (insert snide comment that LLMs use stdio and so does grep).

Apartment living to get worse in 5 years as 6 GHz Wi-Fi nears ‘exhaustion’

Re: Internet could/should be included in the rent

Electric in the rent?

Whenever we've rented, gas, electric, phone, water - they were (and still are) all the responsibility of the occupant, from bedsit upwards.

Only when living in student dorms and flats have any of those been rolled in (and even the suppliers would be thankful, as the churn rate is high).