Posts by that one in the corner

DARPA tasks Northrop Grumman with drafting lunar train blueprints

Re: Railways? Nasty Commie idea.

> what's the difference between a Mag-lev railway and a rail gun *really*?

Not wanting to be pedantic[1] apparently[2] you are thinking of a "coilgun", as a "railgun" has the projectile as part of the circuit, meaning it is in contact with the rails during the entire acceleration phase. And has potential (ho ho) for lovely sparks and plasma, so pretty.

[1] oh, who am I kidding?

[2] personally, I reckon they changed the names on us, 'cos I've been using "railgun" instead of "coilgun" since the 1970's! Anyway, that first link is to a PDF on a dot-mil site, just to reassure you that I'm not just relying on Wikipedia (the next two links); plus the PDF also discusses cool stuff like giant lasers - sadly, without sharks.

Garlic chicken without garlic? Critics think Amazon recipe book was cooked up by AI

Publish only up to three books every day

That would have made Isaac Asimov slow down.

Recipes that stretch out over 2,000 days

Rabbit pie (feeds 15)

First, breed your rabbits.

Britain enters period of mourning as Greggs unable to process payments

Greggs may only be quick'n'cheerful

But they definitely get one thing right: they know what is in their products (and will cheerfully tell you)[1] and they do not randomly change the recipe from one week to the other.

This is an absolute godsend for anyone with food intolerances[2]: SWMBO knows that she can always get a mushroom bake or a sossie roll and know that we won't have to worry about unpleasant-but-not-life-threatening repercussions.

[1] and do NOT mention "oh, the law says everyone has to tell you about ingredients now": what that has done is replace having a list of the actual ingredients, *all* of them, with a laminated card that only lists the items specifically mentioned in the legislation! That does cover the bulk of the life-threatening allergies, which is good, but most places can no longer tell you anything about the presence of, say, alliums.

[2] hopefully also for those with proper food allergies, but I'll not risk making claims on their behalf

ChatGPT side-channel attack has easy fix: Token obfuscation

Specially trained LLMs designed to examine the packets

Training a Neural Net to examine packets is an approach that can be applied to any data stream that exhibits a bursty nature related to its operation, such as the route through the site that a user has to follow in order to do a specific action - the two-way "conversation" as the site sends new page contents, the user responds...

Capture a load of interactions with, say, a banking website: you won't be able to read the encrypted data[1] within the stream, but you may be able to figure out that there is a 79% probability this user is trying to get a new credit card. What you then do with that information - send them more adverts about cheap credit from Kickbacks-Be-Us?

Of course, that 'Net won't necessarily be an LLM, so won't be all sexy and newsworthy.

[1] if you can them just use that, stop wasting time with training a 'Net!

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

Looking at at ollama site, my interest[1] was immediately piqued by their blog post on comparing censored versus uncensored versions of a model.

This may not be something that you want to run day after day, but it is certainly good to have the opportunity to make such a comparison yourself - and it may be even more interesting to show to your local "I know all about LLMs, I have a ChatGPT account" guy (especially if you can get more examples like the blog's demo of asking about what went on in the movie Titanic). In that case, the use of a locally hosted LLM isn't about any argument of local-over-cloud but more about "I actually get to try these non-mainstream models".

> Given that actual uses for the cloudy ones still seem elusive...

That blog - and hence the experiments you[2] could run yourself - actually seem to show just how much *more* useless the censored cloudy ones actually are than they used to be! Again, demoing that to the "right" people may be useful.

[1] Of course, your own level of interest in this may be sated just by reading that blog post and you don't feel much need to install and run the thing yourself - but wouldn't you like to be able to *fail* to replicate their results and show up these LLM-loving loons?

[2] or I could, I guess. Is it interesting enough to get to grips with Docker in order to use the image they provide, instead running one of those scary install methods?

Third time is almost the charm for SpaceX's Starship

> chief being the enormous amounts of land that has to be cleared for them

Wind farms do not take vast swathes of land - first, being giant lollipops, the big part is suspended well above the ground (this is sort of important to how they work), leaving space beneath for grazing, cropping, chasing your beloved through the spring meadow flowers.

Second, the vast majority of UK wind farms are (being) built out to sea.

Yes, they are taking seriously any concerns about the effects on sea life (e.g. marine vegetation growth on the bases leading to increased biodiversity and then to sea mammals gaining new feeding grounds).

Rancher faces prison for trying to breed absolute unit of a sheep

Big Bold Sheep Hunters

The bold hunter cradles the rifle, wipes clean the telescopic sight, kneeling on a platform at the top of a 20 foot metal tower, his blood chilled at the hideous sound of bleeting beneath him. To his left, a line of similarly equipped men, canvas hats darkening with nervous sweat. To his right, nothing; his flank unprotected, he knows his is the honour to guard the line.

Then comes the signal they have all been waiting for, to start the running and their bloody sport:

Phweeet. Phwet. Phwet. "Away lad, right, right". Phwee-eee-eet. " Arbuckle, git bloody gate open, useless boy".

Attacks on UK fiber networks mount: Operators beg govt to step in

What about that iron pillar in Delhi, the one that doesn't rust? Clearly, that is the prototype 5G mast, what else could explain it?

By the time 5G got to Northern Europe, a couple of hundred years later (a trend OpenReach proudly continues), we had a slimmer, more cellulose-based, design which was sadly subject to damage because "it was clearly the work of the devil": a misunderstanding of the purpose of the dual helical cone antennae arrangement.

Your PC can probably run inferencing just fine – so it's already an AI PC

> Does this actually make a difference to how the model behaves?

Apparently.

There was a (brief?) item about this here on The Register not that long ago (sorry, failed to find it - can't recall exact enough wording) which reported claims that you get better results by telling the model that it really likes to do something otherwise tedious, or that it is a starship captain and knows all about calculating orbits.

(I really should have made a note of that article; hopefully someone can provide the URL)

Re: No wonder hardware vendors are on board

Now there is an idea for something that might be interesting[1] to actually try out[2]: could you feed a pre-trained LLM with all the "Arduino for Beginners" articles, plug in a few I2C sensors, then ask it to read 'em and display the values on an attached OLED? Doing all of that on one's home PC, to keep with the theme of the article.

Hmm, best to keep a fire extinguisher on hand; it shouldn't be possible to make a servo running off 5V explode, but...

[1] the assumption being made that there are one or two tech nerds around here

[2] probably already been done, but so has "blink an LED" and we keep repeating that one as well

Neural Nets are good at visual object recognition - it is pretty much the original use case for them.

Whizkids jimmy OpenAI, Google's closed models

Re: If you have the weights, you have the model.

> the weighting of the data could be the model if it's just a set of probabilities, that say the word "course" is more likely to follow the word "of"

That is very much what it is. The extra complication is that the layering and simply vast number of interconnects (which "the weightings" describe) allows a bit more subtlety (the word "course" is more likely to follow the word "of" after I've already seen at least the words "opinion", "my" and "humble" in any order; more so if I've seen "you fool").

The trained model, the thing that we interact with when talking to ChatGPT, is a fixed set of weightings. Lots of them, but fixed. That incudes the results of the "carefully curated" training plus the huger bunch (the "post carefully curated input" thingy is called fancy things, like "pre-trained model" but all it means is we've taken a core dump after treating it carefully, so we can go back and start again from that half-way point).

> then surely in that use-case the models include the data?

The weightings don't need to include the input prompt (e.g. the two legal opinions) - the prompts cause the LLM interpreter to traverse the network of weightings rather than change the weightings within the stored model[1].

Crudely (very crudely) consider a simple lexical analyzer: as you read the input characters you move from state to state through a simple graph, every now and again a transition triggers an action. In a real lexer, a common action is just "start from the beginning again". The larger that graph, the more input characters you can read before having to "start from the beginning again". If you want to, you can take a real lexer and just keep gluing copies of itself into the graph, until you get something that can take in a whole 10,000 line Pascal program without outputting a single lexeme, until it sees the final full-stop when, POW, it can spit out the whole string of lexemes in one go. Sort of sound familiar?

> I'm sure OpenAI are using derivatives of that model to process more data

Exactly; they can release v3 for us to play with, and in the meantime keep on running the training process. When they've exhausted that pile of input, ta-da, V4 is open for business. While we gawk at that, in the background v5 is being slowly built up.

> and maybe will build a model in future that can keep taking in training data - but I didn't think that had happened already

AFAIK they aren't doing that (at least, not the Big Name LLMs). The experiments with letting them send out web requests and adding the results into the overall prompt are sort of a way of sidestepping the need.

[1] bit (bit!) of hand waving here, as - depending upon what sort of 'Net walker they are using - you can have some values modified as the prompt is processed, *but* you treat those like local variables: next session you clear them, otherwise the 'Net will be totally fixated on those two legal opinions. It hasn't learnt but become more limited in scope.

Re: My, what drama

> Artificial neural networks are based upon simulations of those neurons within biological intelligences, such as your brain

Not really "based upon".

Try sort-of-inspired-by-how-we-thought-they-worked-back-in-the-1950s-and-could-fit-into-our-computers-and-hand-run-models-back-then.

And then simplified even more, because then we can really ramp up the speeds of our simulations, using smaller numerical values, varying the number of layers and interconnects (in a gloriously ad-hoc fashion) until it gets faster again. 'Cos the faster it goes, the bigger we can get and the more we can shovel in, without caring if this is how real organisms actually work. And no need whatsoever to care about that anymore, because around approx. 2005 (bluntly) the hardware was available to run experiments based upon computing models, not needing to be fed from the life sciences: for example, we can randomise the bulk of the weightings, starting with noise instead of a blank white slate, does that help the systems we're building to recognise visual input ('cos that was the interesting case)? Yes, yes, it did. Jolly good. No need for iput from the squishy-stuff people.[1]

I don't doubt that there are people out there making carefully researched, beautifully measured, genuine models of signal transport within and between nerves (they probably have a terrific model of the workings of a Giant Squid neuron by now, they've have been examining those for decades now). But I do doubt that that is guiding the creation of the current big-name 'Nets, such as the LLMs. Maybe (hopefully) some proper researcher is looking at it, one whose research goal is "more human knowledge" more than "more money for the shareholders"[2].

> I have been studying biological/synthetic intelligence for more than seventeen years now

But you can, of course, dispel all such doubt simply by providing us with references (don't worry, we can read proper science publications).

[1] Hey, maybe that it is how biological organisms work! To show that is the case, all we need to do is to measure the initial conditions of all the neurons in a brain before it has been fed any stimuli and log how they change as the stimuli are applied - anyone got a brain in a jar, a hundred thousand really thin wires and a steady hand? Good grief, you biology people are so *slow*, I'll just try out another random idea on my computer - done!

[2] No beef with those guys, they need a job just like we do, just - be honest about the reasons for it all.

Can AI shorten PC replacement cycles? Dell seems to think so

We have no idea what "AI" apps will do for the user

but we are going to convince you to buy hardware to run programs that may never even see the light of day.

(Wanted to put in a good last line, like "Yay, progress" or "Won't anybody think of the salesman's bonus?" but this attitude from Dell, compared to the old days and the heady rush of new kit that sliced the runtime of your real workloads, is just too disheartening)

GPT-4 won't run Doom but will play the game poorly

LLMs can never explain themselves

> When asked to explain its actions that were generally correct in context, its explanations were poor and often included hallucinations (aka incorrect information).

(Bangs head on desk) No, of course they can't. We know this. We've known this since Neural Nets were becoming widely discussed in the 1960's!

No form of Neural Net, even if renamed to LLM, is capable of explaining its reasoning or any other part of its behaviour. We *might* get somewhere by bolting something onto/into the 'Net, but we are not there.

If you want a machine that'll explain its line of reasoning, they are available (e.g. planning systems! The clue is in the name!). Shame they are harder to scale than just shovelling random junk into an LLM trainer.

Microsoft calls AI privacy complaint 'doomsday hyperbole'

Re: Nowhere do they say what of their private information Microsoft ever improperly collected

> you need a subscription

Again, that is an issue around copyright.

For example, the majority of information (as opposed to entertainment) I acquire comes from paying a fee to get a copy of the writng that contains the information[1]: once I have learned that "'for' loops in C/C++ are just semantic sugar over the humble 'while' loop" I may be restricted from repeating that precise sentence[2] but I am in no way restricted in using that information nor in passing it on to all and sundry.

> I suspect there's a Terms breach

Putting in "you are limited as to how you use/replicate the information" is going to be really tricky and unenforceable - probably impossible: including all sorts of proofs that they "own" the information (which for a "news report" would mean that they made it up out of whole cloth). That sort of thing is closer to patent than copyright...

[1] I am old and still very much attached to reading physical books; easier and so much more enjoyable

[2] just go with me here; I know that specific example is too short and too generic to make a copyright claim stand up in any reasonable court.

Re: Nowhere do they say what of their private information Microsoft ever improperly collected

There is a big difference between the information contained in an NYT story, which is (now) public and can be freely repeated[1], and the expression of that information - the specific words and word order - which is what is copyrightable.

Spitting out verbatim (too large a chunk of) an article can be a copyright violation.

Repeating information from that article, using a different form of words - and especially if combined (even badly) with information gleaned from another source, is not.

In terms of a human, one is plagiarism, the other is research.

Not that any of this is supporting - or denigrating - LLMs, btw. Just - be accurate in thine attacks, lest ye be taken for an LLM thyself.

[1] until we get to questions of "right to removal"/"right to be forgotten", as appears to be the actual point of the complaint. Which, unless someone can provide a reference, seems to be aimed clearly at the LLM but not at the source of the information, the NYT articles, even though the latter are easier to get the info. Strange, that[2]

[2] cos MS is the Big Bad with deep pockets?

Grab a helmet because retired ISS batteries are hurtling back to Earth

The Master of understatement

> "The re-entry will occur between -51.6 degrees South and 51.6 degrees North." This is quite a large area,

Yes, yes it is.

As you don't give any range of longitude, that is pretty much - everywhere[1] that we are on the planet.

[1] not good enough with the old mental arithmetic to calculate the percentage of the Earth's surface that lies between those two, but "most" is not inaccurate.

Logitech MX Brio 705 – where Ultra HD meets Ultra AI

Re: 4K?

One good use of a higher res camera, especially attached to a (relatively) fixed location: you can clip to a small region of interest and still get a reasonable result for your viewers. You may need to run some extra software (e.g. OBS) if the Logitech stuff isn't up to the job.

Using an ROI is good for cutting out the wide angle shot that includes the window, the annoying coworker, the stoopid "motivational" poster. Or, when using it for "old fashioned"[1] sharing of sketches, just that bit of your desk.

And, if even the ROI is a bit large, it can always be shrunk down before going on the wire.

[1] no idea why sharing an image of a sketch is so derided - no matter how much colab software you are running, doodling on paper is still going to be a lot faster, easier and with generally better results for the majority of people than trying to whip up something in Inkscape or the equivalent.

Microsoft drags Windows Subsystem for Android into the trash

Any particular Android apps?

Not trying to be snarky, just genuinely curious.

For those who do like the idea of running Android apps under Windows, what apps are considered worth it? Assuming that you're not limited to just the Amazon Store selection.

World-plus-dog booted out of Facebook, Instagram, Threads

Re: Those trying to log into Meta's Facebook, Instagram, and Threads............Can't log in........

> Garbage, nothing but garbage I say!

I'll agree that Sturgeon's law works overtime for Facebook, but given that volunteer run organisations[1] find it a convenient and accessible way to contact members and disseminate news, without needing to be even vaguely technical, it is allowed to get past the PiHole[2][3]

[1] techies and commercial organisations have no excuse for using it

[2] although there sem to be sort of holes in the FB pages, almost as if something is being cut out of them

[3] PiHole, so good I had to name it twice

Odysseus probe moonwalking on the edge of battery life after landing on its side

Re: Failure is an option

Just seal the balloon's feed tube (automatic cable/jubilee clip) and cut it on the balloon's side (sharpen the clip on that side).

The balloon will then (literally) rocket away, spinning merrily.

"In Space, no-one can hear you making whoopee cushion noises"

Starting over: Rebooting the OS stack for fun and profit

Re: In the absence of files...

From the first message, I was having a hard time understanding just what problem this is trying to solve.

Now you have given two use-cases, neither of which seem to warrant such a complicated scheme:

1) " if you uninstall an application you can lose access to data without realising it."

Is that something that happens a lot and is a real killer, so much so it needs this complexity? The file is still there, you can see it is still there; presumably you recall what application you uninstalled and can just re-install it? Or just look the name and type of the file and install anything that claims to be able to read it (if you have no idea what was uninstalled, this may require a bit of searching, or just asking your friendly IT guy) but is it something that occurs ofen)? Indeed, if you are as totally application-oriented as this whole idea suggests, your thoughts would more likely be "Ah, I need this data that I usually access via SuperApp, oh dear, SuperApp is not installed, let's just install it then".

2) an idea that apparently "improves security because it prevents applications from opening data objects willy-nilly"

> An attempt to open secret_db.sql with - say - a hex editor will be impossible because you can't navigate to it.

Security for whom? We already have (unless you are suggesting throwing them away) mechanisms for providing security for the data based on the access rights granted to the User (many such mechanisms, of varying strengths); e.g. if your User does not have read-access to a directory containing secret_db.sql then you can't even see it exists, let alone navigate to it.

If an application is able to prevent data being opened then the only thing you are going to protect is the implementation details of that application's data files! That is adding a lot of complexity to the User's system to add a feature that is doing absolutely nothing for the User.

I'd also add that those two ideas can clash, badly. If you have uninstalled an application that can read a particular file - or never installed one in the first place - then methods to decide what type the file is include running the 'file' command and having a look inside with a hex dump. If the file is prevented from being opened because the correct application is missing and you can't identify the file type (no, file extensions are not always reliable) then - ooops. You really *have* lost access to that data!

> or as now when you try to open the file in your mail client the operating system suggests a compatible application

Once you have made your choice from the list of all the compatible applications that the OS has suggested (your OS only suggests one?), that isn't making the selected application "responsible" for the data; the email client just wants to let you view the data it can't render. In fact, if it is anything like my email client, the attachment as received is still in the email store, all you have is a temp file: if you want to keep that as a separate file, save it somewhere sensible now, as otherwise it'll get cleaned up along with the rest of the temp files at some arbitray time after you've stopped viewing it.

> you open a compatible application and import it

Ugh, I absolutely loathe applications that "import" your data: either they just mean they'll open it for you (in which, just say "open") or are making a note of its location for future reference (in which case, just say "noted") *or* they are moving the file, renaming it, trying to take it over completely and prevent you from ever daring to make use of it in any other program. I keep as far away from that kind of program as I possibly can. For example, having tried it, twice, I'll manage all my ebooks (of which I have many - thanks to Humble Bundle) in a simple dir tree rather than *ever* install Calibre again (unless they stop "importing" and just start "taking note")!

If that "import" behaviour is the sort of thing that you like then I can see why I'm having a hard time with this concept.

> The browser would talk to Inkscape to render the SVG

Doing things like providing a rendering service for SVG to any program that wants it is a reasonable thing to do - and, as you mentioned, there are and have been ways of doing that (OLE being one such attempt). But they are totally divorced from any idea of "being responsible" for the data.

Clearly I am not understanding what this concept is supposed to provide to the User and how it warrants such complexity - and, as far as I can see, such ambiguity over "what application is responsible for the data" and the potential for duplicating data with clones for no good reason. Ah well. If it is a good idea that I'm just not understanding then I'm sure it'll appear somewhere and, seing it in action, the penny will drop.

Re: In the absence of files...

> I observe that a number of music applications seem to have given up on the idea of, for example, selecting a music album and playing it from start to finish

And applications that take your carefully arranged files, one directory per album, some of the (newer) files even containing all the MP3 metadata (so even the files know which album they belong to) and carefully ignore all of it.

Our car radio does precisely this - it'll find all the files, put them in alphabetical order and then play them. So the Christmas albums USB memory stick become unbearable - we get every. single. version. of. "Come All Ye Faithful" one. after. another. And end the trip with "We Three Times 11 Kings".

Even worse, trying to play one story from a USB stick with a collection of Blake's 7 audio plays? All the "01 - Title Music" together, then all the "02 - Credits" and so on. OTOH if you find the plots a little simplistic then you'll probably enjoy trying to follow every single one of them at the same time!

> surely, the producer of the album has already curated it for you

Yes, and, as with the audiobooks, made sure that the Concept of the album is made clear by the order of the tracks.

Clearly, this is all our fault for not finding out what format of playlist this player can use (including whether it allows dir names within the playlists) and taking the time to write a program to create them all, because that is something everybody can do.

/rant

PS

This is relevant to the bigger picture, as highlights the point: ok, you want us to label our data - but does everything that insists on reading those labels use the same format as everything else? Or do we have to keep on adding fields when trying to move from one application to the next?

Re: In the absence of files...

> they prefer categorisation by labels.

> Witness some of the comments here

What I notice, particularly in the Reddit comments example, is that they appear to be largely relying on labels that have been attached by a third party to widely-disseminated data (in that particular case, TV and Film, using things like the age ratings).

Other sources of data can also add labels (aka metadata) on behalf of the user: a photo on a mobile 'phone can[1] be GPS tagged, then looked up on an online map to add the venue name, and the shot sent to Facebook to have face recognition applied, tagging all of your friends; whilst my boring old camera will write a UTC timestamp and then anything else is left to me. So my photos are all stored in subdirs by date, with manual dir names added to give a brief description (venue, event, whatever is meaningful to me), if for no other reason than that is consistent over *all* my photos, no matter how old (including transfers from film), and I'm used to it by now.

Similarly, I have a load of MP3s that don't contain any metadata 'cos when they were ripped from the CD we were just glad to get an audio file. So now all my music is sorted using a directory structure that suits my needs, I'm happy with it - and don't really care whether or not the files that make up the latest album that I bought, directly in digital form, have any extra metadata in them.

Perhaps a major difference between the beardies and the youngsters is that they were never faced with doing all of the organising and labelling by hand, so never *needed* to learn any other organisational skills, such as creating their own hierarchic taxonomy of subdirectories, just out of necessity?[1][2]

[1] and, if the scenario given above is realistic, the youngsters would be less wary of just throwing *all* the photos at places like Facebook?

[2] although I do admit to also using a wiki full of notes (which can be searched by text, have category tags attached to pages, ...) where I also keep references to specific locations in the massive archive dir tree - but the amount of material that has been "tagged" in that way is miniscule, mainly 'cos the effort to do that is far too high - and it would lose the metametadata that those items are *so* super-special that they are worth the extra time to add them to the notes.

Re: Hit-and-Miss

> I mean, seriously, if it couldn’t, how could so many important ideas have been originated in it?

Well, version control is all about letting you make a complete mess of things and then have some hope that you can recover

You don't *need* a VCS to create an amazing piece of software and within that originate many marvellous concepts.

It is just a *lot* safer and less nerve wracking to do that with the VCS, especially if there is potentially more than one person working on a section of code[1]. DVCS extends that, making it easier (understatement) to spread your team around; it still isnt *necessary*[2], just helps maintain the sanity of those involved.

[1] The first commercially shipped, shrink-wrap software I worked on was written by a team of half a dozen and we had nothing that could be called "version control" - we each worked on dual-floppy PCs and it was only *just* possible to compile and link the entire program on the single hard drive equipped PC that we all shared. The closest we got to version control was someone running grandfather-father-son backups. We were careful. And delivered. /macho

[2] Barring things like management refusing to allow the longer timescales due to posting floppy discs around the globe - those sorts of things control whether you'll get the resources to write the software, not whether it is possible to ever it under those conditions.

Re: In the absence of files...

> Whether you call something a directory or a folder it's still an abstraction of a place in which to store other things, some of which may also be files or folders depending on the vocabulary you choose to use. You're confusing the abstraction with its implementation.

No - a directory is *not* "a place to store something" - it is an indirection, a reference to where the thing is stored.

Joe Bloggs is not "stored" in any of the paper directories - finding a reference to Joe then points to his physical location (or to another reference that you can then follow to finally reach him physically - e.g. a 'phone number).

>> Folders are a physical box to put things in. I put the thing in box A, it's still in that box later. I move it to box B, it's not in box A anymore.

That is why adding or removing a reference to Joe does not change the way he is stored, nor does it change any of the other references.

Similarly[1] a file system's directory entry is *not* the file itself, it is a reference to where the file can be found. You will often "move" a file from one directory to another, but that is conflating two directory operations into a single command: add the new dir entry, then delete the old dir entry. At no point[1] does the file itself change its location during this move.

The semantics of a "folder" and a "directory" are very different, but everyone continuing to use the UI term "folder" is clearly confusing/limiting people's thinking.

[1] ignoring later optimisations that were added to the file systems, like storing the contents of a tiny file inside its directory node - which then causes all sorts of fun when you create another link to reference that data

Re: Well, at least this reminded me to have a play

> Python is the only language I'm aware of that needs a "pass" statement

Absolutely.

I am still such a Python neophyte that, seriously, I only learned of "pass" last week - and then it took hours of scouring to find it - it is buried hundreds of pages deep in "Learning Python", skipped over entirely in lesser beginner's guides (as they sneakily avoid situations where it is needed) and just what the heck do you google for if you don't already know the name but are hoping something like it exists?

Re: The argument for keeping different horses for different courses runs thus...

> we'll have another look to see whether all that work to recreate exactly what we already have was really worth it.

Um, why would you, or anyone, want to recreate *exactly* what we have now? We have Call of Duty 6, you can play it now, why recreate it - and do so by changing over to a new language at the same time?

Maybe you are thinking that CoD6 is deserving, in the future, of the cult status that many retro games already have, so we must be sure to be able to run it, exactly as it is now, on the Machines Of The Future? Well, maybe, but then we'd use the same methods as we do for the current retro games - fire up a VM and run it in that.

In the course of time, unless you know of some fundamental reason why not (in which case, please share it), there is no reason to think that one couldn't build a video game of the scale of CoD totally natively to a Lisp-based system[1].

But it will get to that point piecemeal, just as CoD did on the PC (what, CoD came into being as a single project, not in any way pieced together from and building upon existing materials?). But did anyone promise the new platform as a Gamer's Paradise from Day One? Maybe, if it happens, it'll take over from a different starting point and, for the first couple of years after it trickles down to all the home users, there will be a resurgence in console gaming whilst the new stuuf ramps up behind the scenes.

As for the database - why recreate it, especially as it is running on a server and we can just talk it over the network? Heck, if it is a big industrial database it is probably already running on a different OS than its clients are!

But we can write some new database clients using Smalltalk - after all, how many people actually write the database server compared the number that write new databases to live in that server and new clients to access that data in User-appropriate fashion?

Or did you actually mean that you don't see any purpose to a new programming idea until it has been used totally replace the current crop, in one fell swoop?

[1] Lisp-based doesn't restrict you to only writing raw Lisp, of course; just like the use of C in CoD doesn't limit it to being written in C (in fact, Googling it, C isn't mentioned - only C#, C++ oh, and Lua, which is in C but aren't writing C, they are just using the language that is implemented on top of it. Many a domain specific language has been implemented in Lisp.

Re: Hit-and-Miss

> That's application level detail

Hmm, nope. It is the difference between tacking version numbers onto files (which you can do manually, VMS just did it automatically) and "version control" as per a "version control system" or VCS. There are "application level" features built on top, some of which are so common that we associate them with the concept of a VCS, even though they are totally generic facilities: most importantly, diff and merge (which you do to anything, with varying usefulness, not just the contents of a VCS).

Ignoring the use of compacted storage (e.g. just saving the results of the diff) - which is a pretty big thing to ignore, but never mind - version numbers do not a VCS make.

Even the crudest VCSs, which operated on the single file level, such as SCCS and RCS, provided more than VMS versions when you saved the next revision - basically, more meta data was added. Some of that VMS retained - eg, the timestamps on each version.

But nowadays, we tend to shudder at the thought of going back to RCS and file-by-file versioning. The use of CVS (and later) features, such tracking renaming and grouping changes to multiple items as a single atomic change are beyond what VMS would do for you.

Perhaps cheekily, let us not forget that automatically saving FRED:001, FRED:002, FRED:003 etc does not even provide the level of meaning we attach to "saving the copy with a new version number", such as FRED:1.2.0, FRED:1.2.1 and FRED:2.0.0 (yup, the third copy was a major change in the file).

Re: Hit-and-Miss

> The cost of PMEM is not such a critical aspect as, one might presume, it becomes just a tier in a seamless pool of storage

We can already do this without the PMEM: you just[1] have to memory map the entire raw disk, without worrying about creating a file system first. You can even leave all that to hardware and not bother the (programmer' model of) the CPU at all, giving you a transparent memory model from L1 associative cache down to spinning rust. Which I'm pretty sure a few applications have done, albeit ending up laying down a data-structure that you could easily just call a task-specific file system (e.g. a database taking over a raw partition instead of using an OS provided file system).

Which is fine and dandy, until you wanted to use more storage than the CPU could address (ok, we hit that immediately for 16-bit CPU addressing, very quickly for 32-bit but, whilst 64-bit made a mockery of the 48-bit LBA hard drives and their feeble 100-odd terabyte limit, by then we were rather stuck in our ways).

But once you do have your memory-mapped drive *or* your PMEM, to protect against software (and some hardware) crashes you apply the same methods to data structures in all of memory as we do now to those on the hard drive: i.e. journalling, atomic writes to switch representations of a data structure from one consistent representation to a newly copied-and-modified-then-verified-consistent representation and so forth. Which means a *lot* of discipline by programmers (not everyone actually writes database programmes all day, so aren't very used to thinking about doing everything by read-modify-commit transactions) but preferably compiler/language runtime support (which sadly isn't, as far as I know, fundamentally built into Smalltalk or its ilk, as per the suggested languages in the article; anyway, how do you automatically decide when a totally arbitrary data structure is in a new and consistent state?).

[1] he says, glossing over the issue of unexpected power loss, but keep on reading.

Re: In the absence of files...

> If another application wants that data it has to go through the owning application.

So do I *have* to have Word installed, as it is described as the "owner" of the file that you sent me? Or can Libre Office be used? What if I have both installed, because I'm supporting Users who each choose a different program?

Is that SVG file "owned" by Inkscape or by the web browser from which I used "save link"? What about the SVG file I created using Inkscape but want to view in the browser? What about when I uninstall Inkscape? What about when I just open the SVG in a plain text editor to do a quick search and replace of an element?

Is that JPEG owned by Irfanview, my favoured quick'n'dirty viewer, or by the Sony program that copied it from the camera? What about the JPEG copied across from the Olympus DSLR? Or the images from my mate's Canon? Maybe they are owned by Lightroom whilst I'm in an editing frame of mind, trying to correct the colour balance? Or back to Irfanview as all the editing required is a simple rotate and crop?

Re: In the absence of files...

> This is probably because there is no physical analog.

There is - a directory! Or a set of directories.

You can use the directory of local trades, or the directory at Companies House, or the local phone directory of every subscriber, or the directory of local council contractors, or the directory of members of the plumbers & gasfitters union, or the directory of Corgi[1] registrations - all of which can lead you to Joe Bloggs, of Bloggs and Sons (gas) Ltd, who fitted the new mayoral kitchenette.

Which is why computers actually have directories full of files, not "folders"; which is just a weird UI choice made by someone who created a GUI and it unfortunately stuck (a great shame, as we definitely still had lots of examples of real, paper, directories in use when that silly choice was made).

[1] old name, but far more memorable than whatever it is now

Well, at least this reminded me to have a play

Despite not having had success applying Lisp or Smalltalk in the Real World[1] this article has reminded me that how exciting all this was.

Now that I have some more time on my hands, I have at least made a few notes, reminders to dig back into the piles of books (Blue, Green, Red and Purple - I still have them all in dead tree form), look up the latest installer for Squeak (and Pharo, why not) and give the middle mouse button some attention.[2]

I might even dig out my hacked copy of Budd's "A Little Smalltalk" (although that reaks of masochism these days, for a while it was the best - yikes - that was actually available)!

Thank you, Liam, for a definitely thought-provoking article.

[1] we tried, back in the '80s; I was even hired 'cos I could namedrop "object browser" and draw a picture to explain CADR, although they did demand I came back looking a little less like a programmer and a bit more City of London!

[2] But first, I am still gritting my teeth and trying to convince myself that semantic white space is bearable and Python *can* be learnt! Oh, why didn't they release the "Raspberry Lua"?

Re: Ah, LISP

Try The Implementation Of Postgres - I wasn't presented with a paywall.

>> Our feeling is that the use of LISP has been a terrible mistake for several reasons. First, current LISP environments are very large. ... As noted in the performance figures in the next section, our LISP code is more than twice as slow as the comparable C code. Of course, it is possible that we are not skilled LISP programmers or do not know how to optimize the language; hence, our experience should be suitably discounted.

>> ...

>> However, none of these irritants was the real disaster. We have found that debugging a two-language system is extremely difficult. The C debugger, of course, knows nothing about LISP while the LISP debugger knows nothing about C. As a result, we have found debugging POSTGRES to be a painful and frustrating task.

Re: NV RAM never entirely went away & predates Optane

> NV RAM / PRAM. It was static memory with a lithium coin cell. Microcontrollers could be powered off and resume.

As used in the Palm Pilot PDAs (although the power was triple-As with a small backup whilst you switched in new batteries - which was *not* a daily occurence!)

> Some could "sleep" due to having a static design. You could literally stop the CPU external clock to pause execution.

The CDP1802 COSMAC ELF was a brilliant first computer to learn assembler on, for exactly that reason. Make a simple two-transistor logic probe[1] and watch it all happening, one clock at a time.

[1] rich sods built a couple of dozen probes and soldered them onto the system busses; flash gits. Mutter.

Re: Windows NT

Under all the Usual Suspects operating systems, you can happily memory map a file across the network. With caveats.[1]

You should be able to use memory mapped files in any programming generic language, Python, Smalltalk, Lisp etc, unless they are being a pain in the backside and not allowing you to take advantage of the OS features.

[1] the most obvious being consistency[2] if multiple processes are mapping the same file and at least one is writing to it: across the network, those processes can now be on distinct machines, so unless you (use a library to) do the hard work, you are not guaranteed that both processes will see the same contents at the same time.

[2] actually, I suppose *the* most obvious is that the server can be unplugged without the client being told and your next attempt to page the file will cause an exception! But that is hard to prevent just b linking in a helpful library.

Re: Costs and Benefits......

The Lisa required far more hardware than the Mac, and definitely way more than the early IBM PC, which helped push its price way up there. Who on Earth would ever want a computer that could take 2 MB of RAM and a hard drive? One floppy and upto 512kB was enough to get the work done all around the office.

And a soft power-off function? Pah, the BRS is cheaper and all you need!

(The Lisa GUI was a lot nicer than the IBM PC, but I'd been spoiled - the PERQ had a bigger monitor)

Chickens and swans: they are harder to tell apart than one might think

Is there *really* some magical difference between using the fondleslab and the PC? And I'm admitting here to using Windows most days[5]

> You have a grid of icons and you can't directly see your "documents". Open an app and it picks up where it was. Turn off the fondleslab and when you turn it on it's right where it was, although some things will be slower to open at first.

You know, that sounds like my day to day experience with the PC for, ooh, decades now. *The* "productivity tool" I use is a plain old programmer's text editor - and those have been reloading the last state for donkeys years, including cursor position[3]! And way better than that, tracking a whole lot of different sets of states that I can switch to easily.

Just quickly comparing the Android[1] devices sitting next to the PC, on both in their normal configuration:

* I have to start up a special program in order to 'directly see my "documents"' (although I can create an icon that will directly open a particular one): xplorer on the PC, Ghost Commander on 'phone & tablet.

* Some programs will automagically reload the previous state, some won't; the web browser is the obvious example (which started doing that on the PC back in - not sure, was it Opera around about 1997?) but the PDF reader(s) will as well[2]

* Other programs steadfastly refuse to reload the previous state, but do something more useful (e.g. load the night sky as it looks today, right now, by default)

* Other programs don't (and I sometimes think I'll hunt down something that will, but it is more faff to do that than not bother; hey, I probably just need to check the options): on both, opening the photo browser program directly means I then have to navigate to whatever I want to see today, even if it is the same as that thing yesterday.

* When I turn on the PC[4], it starts all the bits I *want* it to start, in their previous state (if that is appropriate)

Oh well, maybe my PC is an outlier, running so much weird stuff that it is unrecognisable to Joe Bloggs.

Okay, on the PC it is a *lot* easier to drop into the Really Oldie Days way of doing things, and I do do that (frequently) as, strangely enough, using a command line can be a really efficient way of doing *some* parts of the task. So maybe that is the big difference between the two? But if I decided to load

[1] full disclosure, haven't used an iOS device for more than a few tens of minutes, but I'm told the experiences are similar

[2] actually, both the PC versions win here, as they reload *all* the docs I had open last time, right at the relevant page; great for hardware refs.

[3] seriously, donkey's years; the *current*, latest, exe for my favourite editor, the one I open first in the morning, CodeWright, is dated 01/08/2003

[4] if I used my laptop more, it would hibernate and come back up in its last state even faster!

[5] I have a Real OS running as well, honest, Windows is just in a VM - with sole access to this GPU, so it *is* all very hackery and clever; stop looking at me like that!

Re: Hit-and-Miss

> Let's talk about "version control" and LISP and Smalltalk. Oh, they don't have that, so we're back to BASIC-style "SAVE PRG001.LSP", "SAVE PRG002.LSP", etc.

Absolutely. And forget about diff'ing the saved images to get a meaningful view of what has changed from one version to the next.

And, for that matter, wanting to run different images alongside each other, for the sake of comparison: at that point, you are into loading VMs and selecting the images to run, which is a sudden jump in complexity (especially in Bygone Days).

As much as I had wanted to like Smalltalk, since reading the dedicated Byte issue, that was one of the ideas that I really was never comfortable with, so never really got to grips with "the proper Smalltalk experience" enough to complete a product in it - not helped by being asked how to remove all the debugging and object examination features to turn it into something that was deemed shippable![1].

(FWIW All my LISP work was done the normal way - editing the source in the same editor used for Pascal, roff etc; and I got a couple of the Squeak books - and so many versions downloaded - hoping to "really get into it this time" but, well, not so far, despite being one of those C++ programmers who still really likes using OOP)

[1] There probably was a way to do that but it didn't seem to be documented at the time - maybe we hadn't bought the full "commercial-grade package"? - and, unlike nowadays, you couldn't Google and find a Youttube video entitled 21. Tutorial. How to distribute a Smalltalk application.

Judge slaps down law firm using ChatGPT to justify six-figure trial fee

Re: Second time in two days ...

> But it does seem as those most of the hate here is based on fear

Hmm, but you said

> I have already openly admitted that I have experienced ChatGPT be very wrong about very cut and dry factual matters.

So it must be fear, not a desire to have reliability, especially when there are apologists around (hint: mirrors exist) to wave away the unreliability as unimportant.

So, unreliability is unimportant in something that is being literally thrust upon the masses, who have no way of even judging whether what they see came from that source? Let alone, if it did come ChatGPT, in which direction it has decided to be unreliable today.

Relying on the unreliable is a common trait, but at least you *know* that you've just read an Info Wars article.

Re: Second time in two days ...

> That is, generally, a good thing.

But not a thing that is related to the point doublelayer was making.

Reacting to the context of the conversation would indeed be a useful feature, if it were based on a solid foundation.

Although, what is the conceptual difference between "reacting to the context of the conversation" and simply re-starting from scratch and feeding in the whole conversation again as the prompt?

In other words, "reacting to the context" is really an amazing feature only insofar as they are spending the resources just to keep your session alive whilst waiting patiently for you to type something, rather than dropping it and letting another process use the expensive hardware.

Re: Bullshit

> a fairly new technology

New? In comparison to other computer tech?

Neural Nets predate WYSIWYG wordprocessors. By decades.

The only "new" things here is the scope of the hype, the way that has reached pretty much everyone who uses computing devices on a daily, nay, minute-by-minute basis.

Actually, even that is barely true. As noted, NNs are old and people have been building them and offering them to compulsive computer users pretty much the whole time. Without much success, beyond making money before the bubble bursts (to the point of just-inside-the-letter-of-the-law fraud[1].

There have been - still are - places where NNs have done good - have saved lives[2] - so there is certainly no reason for thinking that there is some irrational anti-Net bias: but those genuine successes are not the subject of the massive hype machine.

The change now is that so many, many more of us are now in the minute-by-minute computer users category and the economy of scale means that spending one dollar per user has allowed a (very small number) of players to build some very big 'nets.

But being big doesn't mean the maths and logic behind these things has in any way suddenly changed[3]. All the flaws are still there[4].

[1] e.g. 1980s/90s, as automated trading took hold: take a stock exchange feed, train up as many 'nets as you can get PCs to run them - don't worry, the data rates are low enough and you don't need to waste 'net nodes on pseudo-parsing natural parsing language - then switch them to output mode before trading opens the next day: ta-da, predictions for the market that day. Repeat for, say, a month. Most models' predictions made losses - but this handful won Big Time! With a totally straight face, sell copies of that handful of 'nets (with witnessed guarantees that they did make those predictions!) to anyone with a big old pot of cash.

[2] 'nets used in medical image screening - *but* we just happen(!) to give very high value to the true positives, enough that we are willing to forgive the false positives - and quietly shrug our shoulders over the false negatives, because we are still catching more than we did previously.

[3] there are good reasons why AI research spent time on more than just NNs - and it wasn't just because the hardware wasn't available: Moore coined his law a while ago (and the current LLMs are appearing in a time where we are concerned that said law is losing steam)

[4] big bug bear: inability to explain their reasoning *and* to have that path tweaked to improve the results; instead, next run (irrespective of whether you change the prompt) you get a totally new output and need to go over inch of it again to look for flaws.

> it makes up nonsense, but it suggested a line of attack he hadn't thought of, which worked.

These anecdotes, where someone - anyone - has used ChatGPT and "it didn't solve the problem, but I came way with a new idea" make me wonder if they could get the same result by using "Rubber Duck Debugging" and a random choice of new approach to the problem by, say, drawing a card from Brian Eno's Oblique Strategies deck[1]. Cheaper, far fewer resources needed.

Now, if you have a citation, including the transcript, so we can see that it was, indeed, ChatGPT that came up with the idea out of the blue, without either being led by the nose[2] or it being one out of dozens of otherwise useless ideas[3] then we can reevaluate the level of incredulity.

[1] hmm, to make it more Modern and Computery, we could drop a copy of those cards into ELIZA and save on buying rubber ducks; I mean, have you seen how much they charge nowadays for a decent model, like a Shakesduck or an Isambard Kingduck Brunel?

[2] I've had some useful info out of ChatGPT, because, unlike Google, after it has come up with the latest, most repeated/hyped, suggestion, you can tell GPT "no" and it'll dig out something a bit more obscure; did that a few times and it finally spat out a reference to some 1970s tech that turned out to be what I wanted to find.

[3] carry on saying "no, that isn't what I want" and it'll happily apologise and then spit out yet another semi-random selection, throwing stuff on the wall and seeing what will stick. Then a simple bit of confirmation bias[4] will get you to remember - or just retell - the sticky bit and forget to mention the mess piled up on the floor or the bits that were so off-target they flew out the window.

[4] even the smartest people are subject to this; I'd even wager the smart ones are more used to having lots and lots of thoughts about a problem that they are really used to just forgetting about all the ones that never lead anywhere that they may not even notice how many such fly past.

"was misbegotten at the jump"

At the "jump"? Misbegotten from the start, surely, at the very point someone on the legal team even had an inkling to use[1] ChatGPT for this purpose, let alone the point where, having prepared everything, they finally made the jump and submitted the prompt.

[1] would say "misuse" here, but that would imply that there is a sensible way for these people to "use" ChatGPT in the normal day to day[2], and it is difficult to come up with a scenario where anyone in the legal profession should ever hand over responsibility for writing anything to ChatGPT, let alone something that is intended to be examined by the one person who has such power over the case!

[2] but what about simple, repetitive, things, like writing flyers to leave under windscreen wipers, I hear you cry? Well, if the interns aren't doing the simple assignments properly first, they'll not be getting the practice needed to get the meaty stuff right. Though if they are in a firm that advertises by interfering with parked cars...

China breakthrough promises optical discs that store hundreds of terabytes

Re: Read/write speed may be the barrier to adoption here

With 100 layers, add as many heads as you are willing to pay for. The rotational latency will remain the same, but you wanted throughput.

AI comes for jobs at studio of American filmmaker Tyler Perry

> Hasn't he heard of matte painting...

You can't expect a man of Perry's stature to know anything about the history[1] of the craft, he is a Going Forwards only kind of guy.

[1] "history" of moving film? Good grief, it is so young that it barely has history! Compared to theatre, say... Next, you'll be telling me people involved in something newer, like computers, don't bother with their "history" either! Preposterous!

Europe's data protection laws cut data storage by making information-wrangling pricier

> The cost savings appear to assume that there was no benefit to holding the information. Doesn't sound likely.

If you're not concerned with any privacy implications, it is (initially, at least[1]) dirt cheap just to grab every bit of PII you can and hang onto it, rather than spending time & effort trying to decide if there really is any benefit to you from hanging onto that data: it looks juicy and interesting, we'll think of a use for it.

A lot of stuff, of all sorts, gets hung onto without any benefit, or the apparent benefit evaporates. Haven't you ever been a part of the occasional "emptying of all the cupboards and dusty storage bins" in the office/factory?

[1] as time passes and the company/clients/customers increase the unnecessary information can take up enough space to become notable - by which time it continues out of habit, assuming *someone* knows why it is being kept around.