Posts by Paul Kinsler

The first real robot war is coming: Machine versus lawyer

Re: then it's a copy even if it's not a direct copy.

Hmm. I've heard these LLMs described as "stochastic parrots", i.e. they are like the proverbial infinite monkeys, but ones with language-appropriate statistical biases.

So they *might* reconstruct a source within some finite sample of runs; but they would not do so reliably, and could not do so to order.

In which case, what level of reproduction fidelity - and what reliability of obtaining such fidelity - would (or should, or might, ...) constitute legal grounds?

Modular finds its Mojo, a Python superset with C-level speed

Re: "trialed" means "tested."

Are you sure? To me they have slightly different implications. I might test something in the lab, or on the production line, to see if (e.g.) it met the specs; but "trialed" to me implied that you have actually used it in its intended working environment, and measured performance there (e.g. "sea trials"). Thus a "trial" is a "test", but a test is not necessarily a trial.

I'm not sure about "Architected", however. I think it hints (or tries to) at perhaps a higher-level or more sophisticated attitude to design; but then it also suggests that its user it trying to use fancy-sounding words to impress. For me, I think this latter meaning would usually win out. Especially if "carefully architected foundations" and little else is said, as in the quote.

Re: The argument for quick hack prototypes in science is very, very weak.

Just because you shouldn't *publish* based on a "quick hack prototype" doesn't mean that it wasn't a useful step in the process or working out the scope of the working version (or whatever you want to call it). Indeed, I have found that trying to write some initial code can even bring unforseen aspects of a problem to light.

And on the more general point, we need to remember that scientific codes cover a very wide range of approaches, from the short to the long, from stand-alone codes to ones that rely heavily on library functions, and ones that vary greatly in the rigour of their design.

One particular thing that does, in my experience, need to be borne in mind when tempted to criticise scientific code is that for a reasonable fraction (I might be inclined to suggest even most), their usage needs to be treated somewhat like chainsaw juggling. To use them, you need to understand what it does, why it does it, what it is intended to represent; and benchmark where possible against non-computational results; and in particular to always be ready stop and to question any and all output for signs of misbehaviour. These are not artifacts that are in anyway designed or intended to be used by the untrained, the uncritical, or the non-specialist.

Otherwise, for what it is worth, and also as a research scientist, I am broadly in agreement with LionelB's comments here.

A lone Nvidia GPU speeds past the physics-straining might of a quantum computer – in these apps at least

Re: Um, sorry, but isn't quantum computing supposed to be instantaneous ?

No, it isn't instantaneous. I might imagine, I suppose, that some reporting might have given that impression, but none of the actual physics ever has.

Re: Weather forecasting is a non-starter application for quantum computers.

In general, yes. But there may be some parts of the many various types of computation going on in a weather simulation that could benefit, and be done faster - e.g. Fourier transforms. But it's not so clear to me how much of an overall speedup this could give, ... you'd have to ask a weather simulator :-)

How prompt injection attacks hijack today's top-end AI – and it's tough to fix

Re: AI's reality is, effectively, whatever ...

Just to repeat myself from yesterday on the subject of chatbots, this week's "Word of Mouth" has quite a good summary for the average listener...

https://www.bbc.co.uk/programmes/m001l97m

UK becomes Unicorn Kingdom, where AI fairy dust earns King's ransom

"AI"

On the subject of chatbots, this week's "Word of Mouth" has quite a good summary for the average listener...

https://www.bbc.co.uk/programmes/m001l97m

Let's take a closer look at these claims of anti-ransomware SSDs

Re: I do not change them so if they ever change something bad has happened.

It might be sneakier to change some of them, but only in carefully controlled ways. What if a ransomware writer has thought about such canaries, and tries to avoid touching them..?

Boffins think they've decoded mysterious 819-day Mayan calendar

Re: I'm sure this has been known for a long time - perhaps its not on the web though.

... for example, during the Mayan Civilization itself, perhaps? :-)

Deplatforming hate forums doesn't work, British boffins warn

Re: It's hardly censorship if ...

Perhaps "attempts at censorship", or some similar wording, then.

Double BSD birthday bash beckons – or triple, if you count MidnightBSD 3.0

Re: suspect it was a fvwm 2.2.5 session that you booted into

I agree, it looks like that to me as well...

Goddard Space Flight Center's new boss swears in on holy Pale Blue Dot

Plenty of good options.

Perhaps "An inquiry into meaning and truth", by Bertrand Russell?

Scientists speak their brains: Please don’t call us boffins

Re: So funding this survey is a good use of membership fees?

Well, FWIW, this particular member is a bit ambivalent about the whole thing, but I'm not going to get in a froth about it.

Even if I personally have no objection, if underrepresented groups really /are/ put off by the whole "boffin" thing, maybe it is best avoided.

Re: Mayhap the IoP should concern itself with funding useful research ...

The IoP is a membership organization, not a research council.

Re: they don't know what it means, and they think it's an insult.

Sometimes it is things like tone of voice, body language, and context that let you know whether something is an insult; in such cases the dictionary meaning of the word can be secondary or irrelevant. And words whose dictionary meaning is insulting can be used affectionately.

Somewhat at a tangent, but not entirely irrelevant: I recall being present at a dispute where in one phrase a swear word was deliberately omitted, but with sufficient emphasis, so that the target not only knew what had been meant, but even insisted that the missing word was actually used.

Re: The current terminology is 'Geek' or 'Nerd', I believe.

Of course, different people will have different opinions on these names. For myself, although "boffin" seems a bit old fashioned, it has never had particularly negative associations, which were largely along the lines of "some sort of sciencey type, who we don't really understand, but has done something clever" - and that may have just saved the day in some old war film or 50s scifi movie - and so I would not mind particularly being called a boffin.

On the other hand, I consider "geek" or "nerd" to be essentially terms of insult or abuse, specifically because of how those words were being used when I first heard them. I therefore am not particularly keen on their current widespread usage, but seeing as nowadays - at least in most media output - the implication seems not to be abusive in intent, I see little point in making a big issue of it. And I certainly meet people who say that for them "geek" or "nerd" have never had any negative associations, and who will happily identify as one.

How the Internet Archive faces potential destruction at the hands of Big Four publishers

Re: i.e. you can study for a Doctorate in Clinical Psychology

Interesting. I notice that the one at KCL is a postgraduate degree (as it seems most professional doctorates are), and in the KCL case "Trainees spend three days a week on supervised clinical practice placements and two days a week are dedicated to teaching, study and research"; so apparently (there) it does seem to involve some research. On that basis, I'd be surprised if cohort in a given course numbered into the hundreds of students, as is typical of undergraduate courses, but I suppose it is possible.

Re: For "PhD", read "degree".

By all means complain about publishers all you like; I am not defending them.

The author of the post wrote "doctorate" (which means PhD[1]), and they did not write (the generic) "degree", even if that was perhaps what they meant. I was replying to a post saying "doctorate"; and was not making a comment on the article, which I would have done in a new thread. Unfortunately, whilst at university, I studied physics rather than psychics, and so in this case was unable to guess that was *written* was not what was *meant*.

As regards "Electronic loans are limited to single-digit copies too" - please do note that I was careful to say "the availability of hardcopy texts is not the constraint".

-

[1] As it happens, there are, in some institutions, the possibility of qualifications such as a "Doctor of Science" or the like, which might also reasonably be considered a "doctorate", but such tend to be quite exotic, and to be even more rarified than even a PhD. Not to mention those which have their own names for what is their equivalent to the PhD.

Re: There are 200 people taking the course

First, note that I was replying to a post which explicitly said "that you need to get your doctorate".

"200 people" do not take the same phd ("doctorate") course in any university context that I have seen. Do you even know what a doctorate is?

And even so, there is now this thing called electronic loans, so the availability of hardcopy texts is not the constraint.

Because there is an alternative to purchasing that textbook

Indeed there is: borrowing it from the university library is usually a workable solution.

NASA wants a telescope on the far side of the Moon

... with cables (fiber-optic?) to the near side

That's going to be quite a long cable... and so - presumably - quite some mass to have to lift off from the Earth. Let alone managing to lay it out.

British industry calls for regulation of autonomous vehicles

Re: Does the UK make autonomous driving cars?

There are at least a range of University groups which do research on such things ...

Wannabe space 'superpower' UK tosses £1.6M at eight research projects

Ouch, 100-120% these days? I

Nowadays the overheads are much larger because they are FEC ( full economic costings), and so cover a lot of things that didn't used to be, such as buildings, infrastructure, maintenance, uni services, the vice-chancellor's new carpets :-), and suchlike.

A whole £1.6m, that's the coffee and biscuit budget for other nations

I would not think that 1.6M£ is some sort of national coffee and biscuit budget, unless the nation is rather small, hates coffee and biscuits, or both.

E.g. if everyone spends an average of £1 a week on c+b, that's approx £50/year each, and so ...

I'm not sure it's possible to pay one researcher for one year with that amount of money

It likely isn't enough, once you take into account non-salary employment expenses. But MAC SciTech is not a university, and likely they're already paying somebody to do a related, similar, or even exactly the same thing, and this extra support will part-fund that salary and/or pay for equipment or other expenses.

Defense boffins take notes from sci-fi writers on the future of warfare

Re: Don't bomb it, buy it.

On the other hand, Germany thought that increasing economic interdependence with Russia would help encourage mutual security.

Which isn't necessarily untrue, but the idea only works if both sides share that same viewpoint.

PC tech turns doctor to diagnose PC's constant crashes as a case of arthritis

Re: people say that Charles III will be coronated.

Well, every new Sun-King needs a proper corona, surely?

Twitter algorithm to be open sourced 'next week,' says Musk

( . added just to keep the indents )

Thank goodness you weren't trying to write (pseudo) python ...

:-)

Research raises questions: Are instruments taken to Mars sensitive enough to find life?

or were they just being optimistic?

Well, I do not think it that uncommon for scientists to be optimistic when trying to make measurements - you have to start somewhere in an investigation, and do the best you can with the available (or affordable) technology. I'm reminded of early gravitational wave detectors (of the Weber type; see https://en.wikipedia.org/wiki/Gravitational-wave_observatory) which - as we now know - not even close to being capable of detecting any typical signals. But in trying to make the early ideas work, you learn what need to be improved.

Boeing bids the 747 a final, ultimate, conclusive farewell

Re: Total distance?

No idea, but apparently a 747 has a cycle limit of 35k. So if the average trip is (er, let me make up a number) 1000k, and there are 1.5k airframes, that makes a distance of 52 billion km; so, unless I've made some sort of error in assumption or calculation, is approximately twice as far a Vger 1 is from the Sun.

This is likely wildly inaccurate, but indicates, at least, something of the order of "quite a long way".

Universities offered software to sniff out ChatGPT-written essays

read the output and put in a few edits.

I suspect that to reduce the match to completely non-suspicious you'd have to do more than a few edits; you'd have to reword a substantial fraction of the sentences. At which point, assuming the GPT-generated result was not full of wrong, your learning objective may have been achieved. But are students using GPT output likely to check it for correctness?

Publisher halts AI article assembly line after probe

Re: you must credit it as an author

I think I would disagree with "as an author", but it would be entirely reasonable to credit it in the Acknowledgments. In any case, giving credit to machines and software is hardly unusual in academic work (e.g. "We took the measurements with a Oscillo 9000x spectrometer and processed and analyzed the the data with the SoxLab PI software package.")

Mixing an invisible laser and a fire alarm made for a disastrous demo

Re: And a sign ....

There's few versions of this around - I rather like the "Please do not look into the laser with remaining eye" one, for its additional sense of danger.

Cleaner ignored 'do not use tap' sign, destroyed phone systems ... and the entire building

Re: standing upright and keeping still

I can't help feeling that installing a bright red "Stop!!" button inside the cavity would have been more human-compatible.

Oh, no: The electric cars at CES are getting all emotional

Re: a register anywhere of completely pointless ideas?

I thought that we had one right here :-)

NASA's latest AI will navigate the Moon using landmarks

Re: existing GPS satellite signals be usable at the moon?

Interesting idea -- but when you are trying to n-angulate, it helps if the signal origins are not clustered in one particular direction, as they would be when you are on the moon listening for earth-orbitting satellites. So even if you could hear them, the errors would be larger - much larger, I assume.

Here - read the theory (Introduction) in the start of Vol. I of the ESA GNSS book at

https://gssc.esa.int/navipedia/index.php/GNSS:Tools

and work it out yourself. Or glance carelessly at fig 1.6, go "hmm" and give up :-)

You get the internet you deserve

Re: AI Wars

Isn't that more or less just the generative adversarial network thing already in use?

But I guess we are about to find out what happens when the process gets into the wild and tries to out-adversary itself (or other remangled copies of itself).

Square Kilometre Array Observatory construction commences

Re: Indigenous land

To be fair, that sentence clarified the point later, with "with native title holders". And in some cases, a more accurate statement might be "land that the colonialists had stolen, but have now returned to indigenous organizations (with various degrees of enthusiasm/reluctance).

How native titles are managed in Australia I do not know, but if you want to find out about the NZ situation there is exhaustive information at https://maorilandcourt.govt.nz/. I found it a quite interesting diversion several months ago, but do not expect to find the information useful anytime soon.

Boffins' beam forming kit opens the door to more realistic holograms

So how does a cavity ...?

A "cavity" here means an optical cavity, i.e. something which traps light (e.g. inside a mirrored box, if you like). But since mirrors are never perfect, gradually the light leaks out; and the here timescale on which the leakage occurs is around half a nanosecond.

ISS resupply drops off experiments for life in deep space

Re: Couldn't bring themselves to say "spring loaded" could they?

I think if I was going to say "stored kinetic energy", I'd have to be referring to something like a spinning flywheel. Spring loaded would mean (stored) elastic potential energy.

Spooky entanglement revealed between quantum AI and the BBC

physics [is] comprehensible; it follows patterns we are configured to exploit.

Well, at least the physics we understand is comprehensible [1].

But it's going to be a bit tricky to check if the physics we don't understand has any links with our languages. And since we can't check that, we may need to take care to avoid cart/horse ordering problems, if claiming some (comprehensible) physics is surprisingly good at understanding human language.

AKA: physics as understood by a language using species apparently has deep links to the languages they use. Surprised?

.

[1] Tautology alert!

Z-Library operators arrested, charged with criminal copyright infringement

Re: make their very own books required reading.

In some cases the textbooks are a result of a lecturer polishing [1] their existing lecture notes, and are thus designed closely around the course they are presenting -- other texts simply wouldn't have the same degree of match. Other texts might do, but not infrequently a given lecturer's course is matched to their preferred emphasis(es), and so you might need to refer to two or more alternate textbooks, and even then find the notation, conventions, or preferred terminology differ.

[1] Although I say "polishing" -- but in fact there is an astonishing amount of work required to turn even quite comprehensive notes into an actual textbook.

AI analysis of dinosaur tracks suggests 'predator' may have been a herbivore

Re: So what was the "expert" ..?

That's a good question. But if I had to guess, I'd guess that footprints are (often?) found in multiples, some of which may be clearer than others; thus enabling unclear footprints to be identified rather more reliably due to their association with clearer ones.

But here they just test on identifying *individual* footprints, i.e. without the context that makes it easier; so although a footprint id might reliably known, it is not necessarily easy to id it without that wider context.

Another warning...

Also, be careful not to discover the truth about the AI. Otherwise it might be replaced instantaneously with a new AI even more bizarre and inexplicable than the current one.

NASA's Artemis mission finally launches after faulty Ethernet switch delayed countdown

== Bring us Dabbsy back! ==

Eh? Is the capsule going to rendezvous with him in lunar orbit? That would be really be something to happen on a weekend... :-)

Twitter engineer calls out Elon Musk for technical BS in unusual career move

Re: Made up Words

Quite some time back now, some physicists tried to sneak their new unit, the "hoover" into a paper. They wanted a unit of vacuum noise, a colloquial way of expressing the random-noise analog to the quantum uncertainty in a single mode.

Unfortunately, the journal spotted this and made them change it.

India's – and Infosys's – favorite son-in-law Rishi Sunak is next UK PM

Re:at least he's somebody who understands

I agree those are useful things to understand.

But, on the other hand, his trajectory so far has only ever been upwards, and with parents being a GP and a pharmacist, he didn't exactly start at the bottom. Most people are not as fortunate (or as talented, either, if you like), and it might be a valid question to ask as to how much he understands how most peoples lives are actually lived and worked, and what they need to be done.

We shall see how he works out as PM.

Water pipes hold flood of untapped electricity potential

Re: Back to the Future

I would suppose there are some downhill pipes/conduits/whatever where the water actively needs to be slowed down, so managing that process with hydro generation as opposed to passive baffles or whatever would make sense.

Tetchy trainee turned the lights down low to teach turgid lecturer a lesson

Readable, fuck no !

I ended up inventing my own shorthand, some of which I can still remember...

... should have taught myself some *actual* shorthand, mind, it would probably have been better all round.

Datacenter migration plan missed one vital detail: The leaky roof

Re: Office was located right along the coast in a city known for their yearly hurricanes!

... but perhaps there were also shutters, but, being open, you couldn't see them...?