* Posts by Paul Kinsler

1051 publicly visible posts • joined 9 Aug 2007

Page:

How to run an LLM on your PC, not in the cloud, in less than 10 minutes

Paul Kinsler

Re: "was only running a 1060"

Having just searched, I note that a 1060 has 6GB as standard... i.e. three times that of mine; but then it does appear to be several years younger ..

... such luxury!

Paul Kinsler

Re: curl -fsSL someurl | sh

On the plus side, at least the install script is Brexit-compatible. On the fifth line, it has

set -eu

:-)

Linux for older phones postmarketOS changes its init system

Paul Kinsler

Re: If systemd is so modular

OK, I did. That page tells me some things, but is headed with "This page has been obsoleted and replaced: https://systemd.io/devdocs/MINIMAL_BUILDS.html"

If I follow the link, it 404s. A bit of searching using some text from the "obsolete" page gets me to

https://systemd.io/MINIMAL_BUILDS/

... which is rather similar to the obsoleted page (ok, so mainly its just been moved. Fine).

I'm not sure the instructions are super-clear to me (although I've probably seen worse) but it looks like they would make more sense if you were sufficiently into compiling systemd... I guess.

Paul Kinsler

Re: GUI frameworks being dependent

OK, so could easily I run it per-user (or just for users), but without it (also) running as a systemwide init thing?

Font security 'still a Helvetica of a problem' says Australian graphics outfit Canva

Paul Kinsler

Re: But isn't that a problem with the font management tools...

Yes, but perhaps there is a communication issue here - most people installing a font aren't going to see themselves as "using a font management tool", they are just going to think "I'm installing a font". Therefore if you lead reporting of this with "font management tools" it probably wont be seen as relevant or applicable - even though it is.

Boffins propose fiber-optic network for the Moon

Paul Kinsler

Re: ... I don't think it's going to get nicked

Heh. I was more thinking about how well the fibre would be coupled to the small ground movements it's being used to detect.

Paul Kinsler

... then trundles off, spooling out fibre behind it.

I suppose there might be an issue hiding here -- is it really sufficient to just leave the unspooled fibre lying on the surface? Or would it be better (or even necessary) to bury it, or (perhaps) to at least fix it securely to the ground at intervals?

Hmm.

Paul Kinsler

Re: Doesnt sound very realistic

100km of fibre doesn't just give you data from 100km away; you can get data from any point along the fibre -- so it's more like 100km of seismometers, which is a lot of instrument.

Not that makes it easier to deploy, mind, ...

(Recently I saw something similar pulled off comms fibre on the west coast of NZ ... you could see/detect cars driving along a nearby road)

Lightweight Windows-like desktop LXQt makes leap to Qt 6 with version 2.0

Paul Kinsler

Re: a forever unchanging UI

For me, fvwm comes pretty close ...

Probably not for everyone, mind. :-)

Developer's default setting created turbulence in the flight simulator

Paul Kinsler

Re: trouble with cron,

I forget the exact details, but I once suffered from a post-articles-to-usenet cronjob interacting badly with an unreliable nfs mount. I think what was happening was that article in out.going could be read and was posted, but then never got deleted due to the nfs issue, so that as a result it got posted over and over again. There was quite the little forest of stuck processes when I eventually found out and looked.

I must say the email from the moderator of sci.physics.research, alerting me to this problem, was remarkably polite.

Fairberry project brings a hardware keyboard to the Fairphone

Paul Kinsler

Re: Sharp Zaurus

I still use mine, although mainly just as an alarm clock - I like the little chirp noise, and unlike a dumb clock, can set different wake times for different days in advance.

40 years since Elite became the most fun you could have with 22 kilobytes

Paul Kinsler

Re: And its such a simple and elegant solution to "How do I generate and store 8 galaxies in 22K?

OUAT, and on the BBC B, I wrote myself a sort of text-only open-ended map-based adventure game where you could wander about in various sorts of terrain and have various sorts of encounter (i.e fighting monsters, and maybe finding food..?); I used a combination of fixed-seed RNG with weird-argument trig functions to make reproducible terrain with contiguous patches of swap/hills/etc/. It worked, and I suppose I would now have to call it "procedurally generated", but can't help feeling that that phrasing makes it sound much more well thought-out than it was...

Could immutability be a Leap too far for openSUSE users?

Paul Kinsler

Re: There are plenty of alternatives to systemd and other common component...

There are indeed.

However, not everyone is a fan of distro-hopping. And if you have been with a distro for a long time, you may well have a body of experience with it that makes your life very easy and familiar.

But then if your distro-of-choice then decides on making a significant change of direction (let us e.g. say swapping to systemd) that you do not find helpful, you have essentially two options, neither very palatable: (a) swap distro and start somewhat from scratch, or (b) learn to live with this new horror imposed upon you.

Fortunately, I'm primarily a slackware user, so so far I have mostly escaped many of the big swerves like systemd, snaps, &etc. But applications I use a lot have sometimes changed under me, and the transition is not always easy - whether I choose to stick-and-adapt, or to jump-ship to something else. And even if the a-vs-b choice is indeed yours, the timing imposed is not.

Atari 400 makes a comeback in miniature form

Paul Kinsler

Re: a bequeathed Atari.

Heh. Never used an Atari, but I started on a bequeathed ZX81. Still, a start's a start.

Eben Upton on Sinclair, Acorn, and the Raspberry Pi

Paul Kinsler

Re: "Suppose you'd taken the Tube out. Would anybody have really cared?"

Really, the statement should be "Would enough people have really cared?" - because yes, I would have cared. I needed my 6502 coprocessor to make my programs run fast enough.

Not that they were especially exciting programs, but they did keep me amused.

British railway system is getting another excuse for delays – solar storms

Paul Kinsler

under the ground

Geomagnetic events are, as the name suggests, magnetic, i.e. not electric (charge). And they can induce currents in the ground itself.

And albeit at a tangent to this subcommenting, it might be worth noting here that solar wind and CME interactions with the magnetosphere/ionosphere/etc are rather excitingly complicated, and all kinds of crazy and/or weird physics goes on up there (and sometimes down here). There is no guarantee that "reasonable assumptions" will be correct, unless perhaps it happens to be ones particular specialism.

Paul Kinsler

Re: Pointless waste of time

No doubt because of my many sins, I have now attended a number of space weather conferences (the attendees are quite wide ranging - academics, space hardware companies, met services, government labs, the occasional from defence or civil service. As I understand it, although it is not always so easy to get rail companies to take space weather seriously, those who run power grids pay more attention. Those who run Met services (eg the UK Met Office, and its various European and US equivalents, and others around the world) tend to offer space weather predictions; and these even start by measuring current solar activity, and using it to predict the eventual terrestrial consequences. As far as I recall hearing, "Doom" is pretty unlikely, but some might have to react quite nimbly if something Carrington-like seemed to be on the way.

In NZ recently the power networks org even plugged one of their power stations into the ground and had a look to see where the current came out - here's a couple of bits of the abstract for a recent conference presentation:

"A government-funded research program in New Zealand is currently examining the impact of extreme space weather on the country's energy infrastructure. Specifically, we are interested in understanding how geomagnetically induced currents (GICs) affect high voltage power transformers. GICs manifest as quasi-DC currents within the power system and can create issues in the electrical network due to transformer saturation." [...]

"In January 2023, with the assistance of the National Grid operator Transpower, we leveraged New Zealand's high voltage DC (HVDC) transmission link to directly inject current into the ground at the Haywards substation near Wellington, New Zealand. We closely monitored the impact on two 216 MVA, 220/110 kV autotransformers, and one 80 MVA 11/110 kV step up transformer, as well as the associated transmission lines. This testing involved conducting six injection tests over a nine-day period, each lasting between one and two hours. The maximum current injected into the ground reached approximately 612A..."

Paul Kinsler

Re: In what world would a solar flare affect a [] train from "shorting" a the section signal out.

I wouldn't imagine it stops the short existing - but presumably it might induce counter- or over-currents across it, in addition to what you normally expect. And so by altering the "short"current, thus also change the measurements, and hence affect the inferences you make regarding whether the short (and by implication the train) is present in some location.

Paul Kinsler

Re: replicate these results using actual equipment?

This isn't anything to do with RF interference: they are talking of significant geomagnetic disturbances, over geographic scales. Space weather events that they consider induce current in all sorts of things, including the ground itself.

FTC wants Microsoft's relationship with OpenAI under the microscope

Paul Kinsler

Re: Non-expert opinion

In passing, I happened to catch about half of a thing on BBC Radio 4 this morning which seem to have an informative summary...

"Artificial Implosion"

https://www.bbc.co.uk/programmes/m001tkbr

Systemd 255 is here with improved UKI support

Paul Kinsler

Pointing a phone camera at the screen is arguably more convenient

Indeed. You could photo a displayed error message, for example. And I have no particular objection to the idea of adding QR codes as well, but really do think there should be at least some human readable information.

Veteran editors Notepad++ and Geany hit milestone versions

Paul Kinsler

Kate. I don't know why.

I use aXe, because, er, habit.

Every couple of years I try geany again, but at some point forget, or get annoyed about something, and drop back temporarily. Or, rather, "temporarily". Ah well.

'Wobbly spacetime' is latest stab at unifying physics

Paul Kinsler

Re: Nature has no more trouble handling four dimensions than one. Mathematics does.

No, mathematics doesn't have trouble with high dimension numbers. But some mathematicians might, depending on their specialty :-)

Digital memories are disappearing and not even AI or Google can help

Paul Kinsler

Re: photos of people where the subjects are unknown are of next to no value,

Are you sure? It might be that some future digital archaeologist will more interested in the clothes worn, ther things carried, or even what might be shown - perhaps by accident - in the background of our photos than the notional "subject".

Tiny bits of space junk reveal their wherabouts when they collide, boffins hope

Paul Kinsler

DSN is also quite highly-tasked right now,

It might be they intend to do some signal processing on whatever the DSN records in its usual operations; in which case it might not require any dedicated time at all.

Six pack of sub-Neptune exoplanets hang tight around nearby star

Paul Kinsler

Re: How do they know that ...

It is unlikely they "know" it in the sense I believe you are implying; it will instead be a reasonable inference based on their expertise and the evidence they have available.

AI threatens to automate away the clergy

Paul Kinsler

Re: insert famous Arthur C Clarke quote

Ah, I know this one:

"Any sufficiently crappy magic is indistinguishable from technology"

Do I win a prize? :-)

Leader of pro-Russia DDoS crew Killnet 'unmasked' by Russian state media

Paul Kinsler

Re: Axis nations

Yeah, that read weird to me as well. I assume the issue is that in the West "Allied" doesn't mean merely allied, but is strongly associated with "us good guys" (because WWII), and Axis likewise doesn't mean just "some cooperating countries", but also "those bad guys" (cf WWII again). How can (or why are) we - as targets of a pro-Russian DDos crew - be an "Axis"?

So if you are talking about Russia and those cooperating with it, "allies" would be correct but might confuse us western readers; but calling the crew's targets the "Allies" would be even worse (why are they targetting allies??). So they (the writer) used "axis" instead, trying to use the "otherness" implication of axis, but starting from a Russian point of view. I.e. that if we might say "The Marvellous UK and her Wonderful Allies" and "That Axis of Problematic Countries Opposed to us"; then *they* might perhaps say "Our Great Russian Nation and our Allies" and "That Inconvenient UK and its Axis of Nuisances".

Bah humbug, I've now spent too long thinking about this, and failed to be as clear as I would like. Why couldn't they just say "target"? Much clearer and without confounding associations.

IT sent the intern to sort out the nasty VP who was too important to bother with backups

Paul Kinsler

Re: every other

To be pedantic, *every other* is not only an ambitious claim, it is also wrong: Notably, xedit, the minimal default X windows editor uses it to delete selections, not windows; as do both xterm and xfce-terminal, and aXe (an old school editor). My (perhaps unreliable) understanding was that Ctrl+W as delete-selection was standard/default X windows behavior (as is somewhat implied by usage xterm and xedit), although naturally that isn't enforced by the X-police.

"Many other" might be correct, I suppose, but given the extraordinary variety of editors and other programs out there it might be hard to be sure. It seems that in your usage, one thing is dominant, but on mine the other is.

Mainly, I am just irritated that since programs I usually use follow X, that I cannot at least set a config option to remap Ctrl+W in firefox, especially given all the other weird things firefox allows me to change.

PS: naturally, as I typed this, a muscle-memory use of Ctrl+W as delete-selection meant that I lost the first draft. :-/

Paul Kinsler

muscle memory

... and then there is what firefox insists on doing in response to Ctrl+W (close tab), but which is -- in very many other editor-type things -- merely "delete selection". Thus an attempt to remove an extraneous phase instead becomes "delete my entire carefully constructed prose/argument/rant", and throw away the tab to boot. I read somewhere online of someone who said they have to go as far as patching and compiling firefox from source itself to avoid this, because that is apparently what it takes (!).

FFmpeg 6.1 drops a Heaviside dose of codec magic

Paul Kinsler

Re: Heaviside brought back memories

And one of the other important things Heaviside gave us was the vector-calculus version of Maxwell's equations...

Paul Kinsler

... the Extraordinary Monsters of the Heaviside Layer.

That's just the "here be dragons" version of space weather. :-)

Gutenberg is quite good for old SF. I rather liked the "Venus Equilateral" series: https://www.gutenberg.org/ebooks/68008

Airbus to test sat-stabilizing 'Detumbler' to simplify astro-garbage disposal

Paul Kinsler

Re: still functional ...thus doesn't need a Detumbler.

But I suppose if you were going to worry about when a satellite might become non-functional in the future (whether by accident or end-of-life), then having a detumbler already installed might be sensible. But perhaps a detumbler might have unwanted side-effects..?

To pay or not to pay for AI's creative 'borrowing' – that is the question

Paul Kinsler

Re: if the AI operator was giving content away for free ...

Perhaps, ... if the giveaway was at sufficient scale.

Paul Kinsler

Re: can't see how it's going to be controlled

I think the distinction between likely human and machine learning is not necessarily about the input stage, but rather the output stage, and the possibility of (non-trivial) monetisation.

If I were to read and remember a lot of Harry Potter trivia, I have spent a great deal of effort, and at best will be able to impress Harry Potter fans, or appear on quiz programs; if I try to make money off it at some point some rights holder's are going to make claims. I might instead program that weird and entirely unreplicable machine I (hypothetically) have in my shed; but again this doesn't change the situation - it can be an amusing curiousity, but little more, or rights holder could quite reasonably come calling.

In contrast the whole *point* of these text-ingesting LLM's is to be reproduced at scale, used at scale, and monetised at scale. So naturally they (should) have to deal with the rights holders whose materials they have ingested, and ingested primarily as a commercial activity in search of profit. Of course, they are perfectly welcome to instead keep their LLM's as amusing and unique curiosities, making little or no money for anybody; and if so rights holders would no doubt be mostly content to simply marvel at all this LLM cleverness. But there seems little likelihood that the "amusing curiosity" route is the point - the intent is replication, use, and significant monetisation.

Just because you're paranoid, doesn't mean AI's not after you

Paul Kinsler

horses can talk.

I suppose if you were in charge of Yahoo, rather than working for BT, this belief might be more understandable...?

Introducing the tech that keeps the lights on

Paul Kinsler

Re: No requests, no retry, that’s why it takes ...

I wonder whether an "almost diode" might be more useful, where the reverse channel is low bandwidth and (eg) only consists of a few bits - so you might be able to say say e.g. "I hear you ok" and "please resend" once a minute (or whatever), but nothing else. So there'd be less to lock down, and complicated exploits would be very slow, even if you could work out what to do.

Wanted: Driver for rocket-powered Bloodhound Land Speed Record car

Paul Kinsler

Up to maybe 1930 or so, anyone with an aptitude [...] could go about constructing an aircraft

Indeed, on Gutenberg there is a book from that sort of period, describing how you might build one yourself; i.e. "Building and flying an aeroplane", C.B. Hayward, 1912.

GPS leading your phone astray? We can just fix that in code, startup claims

Paul Kinsler

Re: doesn't have great GPS accuracy individually but in a network will be better.

You can already use differential GPS if you need accuracy, which is the sort of thing you might do if doing some sort of survey-by-drone arrangement. But I guess this idea could allow you to get improved - but not as good - accuracy, whilst needing to be slightly less organized about how you achieve it: :-)

Microsoft's 11-year itch: The uncelebrated anniversary of Windows 8

Paul Kinsler

Re: It was easy enough to remove some of the nonsense

Indeed. I got W8 pre-installed on a laptop, and at first sight I even thought it looked rather nice. However, I never actually used W8 for anything, ... because straightaway I installed linux on it and never booted into W8 again :-)

To prevent 'lost' nukes, scientists suggest storing them in a hall of mirrors

Paul Kinsler

Re: replacing blue barrel with a red barrel

FWIW, the article says the system uses radio waves, so the visible colour is unlikely to be relevant.

Further, since the claim is system is sensitive to shifts of only a few millimeters, any new (or tampered-with) barrel would have to match the original sizing and positioning closely.

It might indeed not be tamper proof. But it would make any tampering process at least somewhat more complicated than it was before, and therefore arguably more risky. Just because the idea is not a panacea does not mean it is not useful as part of a suite of controls.

Airbus commissions three wind-powered ships to sail the Atlantic

Paul Kinsler

Re: My guess is that it's all a load of greenwashing.

On the other hand, my guess would be that they are trying a range of moderately promising things out, to see what works best; and inevitably some don't work out as well as they hoped.

NASA eyes 3D-printed rocket nozzles for deep space missions

Paul Kinsler

Re: not principal investigators

FWIW, "principal investigator" is in fact a term used in research grant applications in the UK, as in the person who led the submission, and who would (or will) be in charge of the execution of the project. Terms like "lead researcher" do not have those specific in-charge and/or administrative overtones.

Astroboffins spot high-power 8b year old radio burst from pre-Earth event

Paul Kinsler

Re: but it could be anything

Well, yes and no -- it's called "dark matter" because whilst indeed it could be anything, it isn't being seen (dark), and it seems to be have properties consistent with matter. In this it might either be hot or cold dark matter (with different implications), and very much differs from dark energy (doesn't seem to behave like matter).

So it's not "we have no clue" -- there are *some* constraints on it's properties.

Look ma, no fans: Mini PC boasts slimline solid-state active cooling system

Paul Kinsler

Re: Flanders & Swan said it first

Indeed, is the system actually quiet, or just apparently quiet to humans..?

Moonstruck Modi wants lunar Indian crew by 2040

Paul Kinsler

As the saying goes, no Bucks, no Buck Rogers.

Just in passing, anyone who feels the need for more Buck Rogers (steady there, Wilma), might wander off to the original strip:

http://www.rolandanderson.se/comics/buckrogers/buckstrips.php

'Gay furry hackers' brag of second NATO break-in, steal and leak more data

Paul Kinsler

"and have a flare for political publicity stunts"

Well, I suppose if you have no natural talent or aptitude (i.e. "flair") for publicity stunts, then something as crude as lighting a firework will have to do...

PhD student guilty of 3D-printing 'kamikaze' drone for Islamic State terrorists

Paul Kinsler

Re: (that implies) universities are more interested in the income than the quality of research.

However, someone has to agree to supervise the PhD, and most very likely wouldn't want the drag of an insufficiently capable student. It's not like some supervising Professor X gets to pocket the fees.

Perseverance rover sets a Martian speed record with software controls

Paul Kinsler

Re: Doesn't someone realise that "more than a month" is very similar to "weeks" ???

Perhaps that that "weeks" was (or would have been) "weeks only driving and doing no science", but the "month" included both driving and science...?

CERN experiment proves gravity pulls antimatter the way Einstein predicted

Paul Kinsler

‘Antimatter’ is just matter with opposite spin.

"Opposite charge" would be better than "spin" ... at least for sufficiently general notions of charge (e.g. arguably dated usages such as where lepton number <-> lepton charge).

The problem with using "spin" is that - for the symmetry-based "spin", unless the object is spin zero, there is both a total spin and spin components; and the total spin is always positive, and the sign of the component is irrelevant to the particle/antiparticle determination.

Using "charge" in the generalized sense is also confusing, given that you might conflate/confuse it with (only) electrical charge; and not the specific implication of "oppositely charged to the normal (i.e. non-anti) version of the particle".

Page: