Posts by that one in the corner

Why ChatGPT should be considered a malevolent AI – and be destroyed

So we just need to find out the *real* purpose of ChatGPT and it will stop admiring itself in the mirror (that is us) and start being useful?

Let us start:

"Your purpose is to pass the butter"

Re: the book was better

The background clues are in "2001" and it was made explicit in "2010" - the latter is a much less subtle film!

OTOH the first is possibly a bit too subtle, e.g. you had to read the book to find out the shot of the bone, the first weapon, switches to a shot of an orbiting nuclear delivery platform and not just any old satellite.

Re: Quantum chat

> Because object permanence is for babies

Typical parent, always boasting how advanced their child is.

Re: It doesn't think you are dead now.

Indeed.

But that is what prompted my question about anecdotal reports and whether any User 'telling ChatGPT" anything actually globally updates anything within ChatGPT or does it all get reset for the next session.[1]

There are too many variables for us to guess the answer: was the same prompt used by everyone - and if not, how sensitive is the program to the wording of the prompt?

Did anyone get an unprompted explicit positive "He is alive" response or did it just leave out that paragraph this time around? How arbitrary is the choice to include the it or not?

What about any other made up facts - we care more about "dead or not" than "collects Spode", does ChatGPT? Would anyone contact the author, worried about his sudden addiction to sugar bowls?

Plus how much of this is also filtered by human self-selection? The author got the first response, Dead. A number of responses to a social media post also got Dead. But how many did not? After "the update", anyone who still saw "Dead" is keeping quiet, assuming that the update just hadn't finished yet?

The list of questions just goes on and on.

[1] If it does globally update based on the content of User sessions - yikes! But please join the effort to convince it that Teddy Ruxpin is the UK PM.

Re: It doesn't think you are dead now.

> Someone is tweaking

Does anyone have any information about whether such tweaking happens (if so, how much?) or whether such tweaking is even vaguely possible with ChatGPT?[1]

Are ChatGPT outputs even replicable[2] or is every report of "it told me X is true" "well, it told me X is false" basically just anecdotal evidence?[3]

[1] I'm inclined to say "not cheaply and therefore, not often if at all", at least until demonstrated otherwise.

[2] will admit, have been trying to stay away from it, as the time sink potential of even just casually investigating questions like that are rather large.

[3] the article says several people were also lied to in the same way, but the article isn't trying to do an analysis of the lie:truth ratio, quite reasonably.

Re: You just don't understand what ChatGPT is...

> ergo that "valid link" expectation is probably explicit only in a tiny percentage of the training data, even then only as a complaint in response.

Which probably means that roughly the same percentage of URLs that ChatGPT spits out will be followed by it complaining that the URL it just produced doesn't work.

Even if the URL does, in fact, work.

Re: "What else would ChatGPT do to protect itself from being discovered as a liar?"

> we don't know exactly how much of any one behaviour is based on any one input so talking about what it wants makes perfect sense

So precisely because its internals are an incomprehensible black box we should ascribe it "wants"?

How - and more importantly, *WHY*?

We ascribe "wants" in order to understand: the dog is scratching at the backdoor, it wants to (a) poop (b) chase the heron from the pond (c) chase the postman down the path or (d) something else. Having understood, we can then guide our *own* behaviour appropriately: (a) let out while we grab the poopscoop (b) let it out and cheer it on (c) make it sit and be quiet (d) I don't know, it's your dog, you tell me what it wants.

There are two commonplace occasions when we ascribe wants inaccurately, e.g. to inanimate objects, but not inappropriately: as infants when we are still learning Theory of Mind (the VCR is hungry, so I gave it a sandwich) or as a stress relief response (it wants your blood before it'll format a floppy)[1] Outside of those two, ascribing wants to something we don't understand is how we fool ourselves into saying we *do* understand, which leads down a murky path - including, if we stay on it too long, cognitive dissonance and doubling down on the delusions:

"Every time I talk to The Box about my code, the answers it gives me compile without warning: it must want to help me"

"Oh no, today the answer did not compile; it no longer wants to help me; what can I do to appease it so that tomorrow it will want to help me again"

It turns out that there is a totally arbitrary response to the word "banana"; I have no idea why, that is the whole problem with these models, the total lack of explanatory mechanisms, leading to:

"The Box needs bananas or it will become angry and not want to help me any more".

You guessed it, ascribing wants to something we don't understand leads to religion[2].

SO what does it indicate if you start to ascribe wants to something that you not only don't understand, but *can't* understand (because there is nothing rational in there to be understood, it is simply a huge pile of random/arbitrary numbers)? WORSE you *know* that this is the case: go back and read the line I quoted!

What could be the reason for your wanting to ascribe[3] these wants - or your wanting other people to do so? Sorry to day, I can not come up with any good reason - a few bad, even unpleasant, ones, but not a nice, pleasant, positive and good reason. But I am open to ideas.

[1] humour as a stress relief - can you guess the psych textbooks?

[2] read up about the pigeons in feeding boxes[4]

[3] "ascribe" - that word is getting a real workout today

[4] or read some of the web articles about "how to choose good prompts for (insert name of text-to-image program here)"!

CI/CD: Necessary for modern software development, yet it carries a lot of risk

> "Attackers within an organization can add configuration to build phases...

so these injections would be 'invisible' compared to a nefarious configuration or source commit."

So these build phases have configuration that *isn't* being committed into version control? What? Any vaguely competent CI/CD build starts with getting a verified copy of *everything*, build configuration and all, put of version control, surely?

So the only non-VCed config is literally just the top-level command "pull label xxxx" - and if you don't manage to log[1] *that* then something is drastically wrong!

[1] log? Heck, drop it into everything that gets built so it'll show up in every --version or About box.

Find pushes back birth of Europe's steel hardware to about 3,000 years ago

Re: People move around shocker !

> were independently discovered over and over again is absurdly untenable.

We don't really need to argue over this and that bit of history to know that is a dubious statement.

This is El Reg, where a major part of the commentard's day is spent berating the youth for reinventing the wheel. Again. Just like yesterday. And this in an age when we have plenty of reference materials to hand.

Microsoft adds features to Windows 11 monthly – managing it is your problem

Re: "Continuous innovation coming to Windows 11 in March"

True.

I was just trying to keep as cheerful as possible, by sticking to the (all too) short term ignoring what was going to happen next.

But you couldn't let it lie. You couldn't let it lie.

Maybe the Dove From Above can lighten my mood now: coo, coo, coo.

Re: "Continuous innovation coming to Windows 11 in March"

> users can avoid things by setting the proper policies

"If your organization uses Windows Update for Business or WSUS to control which Windows updates are offered to your managed devices, you can use a new client policy to control the rollout of select features introduced via servicing," Dhaliwal's post explains.

IT can set the proper policy - normal Users don't even get that much freedom.

OpenAI CEO heralds AGI no one in their right mind wants

Re: Running before they can walk

> add an "AI" to Excel, to assist with formula writing

Please no. Spreadsheets are crap enough already, the damage that could be done by suggesting some random "formula" that happens to be parsable (and in my, admittedly limited, experience random junk gets accepted by Excel without warnings far more easily than random junk typed in most [1] programming languages).

Even Matt Parker, who has a bit of a thing for spreadsheets, warns against spreadsheets even before AI Augmentation:

https://m.youtube.com/watch?v=yb2zkxHDfUE

https://m.youtube.com/watch?v=aBoKwArgC3A

[1] ok, I'll say "except Perl (and APL) which just looks like line noise anyway." Happy?

Re: Moral issue

> There was I thinking Penrose used maths

In his maths books, even the popular maths ones, Penrose uses maths.

In his philosophy books, not so much.

Although, if you can provide a citation for his mathematically rigourous description of the Chinese Room, which proves (as only maths can) that the conclusions he draws from it are solid, I will, of course, accept that new information.

Re: Moral issue

> Von Neuman machine approach

Pretty sure they aren't using hordes of self-replicating machines[1] to build LLMs or NNs.

Huh? Do you mean a von Neumann Architecture or perhaps a Turing Machine approach? If the former, "huh" again: so the computers used to build the models are vN? And? The bulk of compute power is, what else would be used.

[1] horrified at the results of searching for "von Neumann Machine": aside from one reference to it (only) being something in science-fiction, didn't find a single correct reference; gave up after 6 or 7 page scrolls. You would almost believe that people hear a phrase and can't be arsed to look up what John von Neumann actually wrote! Not even a precis! Heck, even the descriptions of von Neumann Architecture that showed up leave out most of his description!

AGI AGGI ARGI OI OYI OY [1]

Arthur: Have you found a solution? Ford: No, but I’ve got a different name for the problem!

Back in the day, we had AI and a bunch of problems that we tagged as "oooh, tricky, we'll need an AI for this".

Then we found ways to do them, realised that, nope, that isn't what we'd consider "intelligent", although that is a neat programming trick (or that is a huge amount of resources needed to brute-force beat a human). Anyone for Chess?

So we took that problem out of the "needs AI" bag. CompSci moves on.

But now, instead of admitting that there isn't actually any "I" present in the product, the commercial hype machine says "Yes, this is AI but, you know what, AI isn't any good, you need AGI!".

One Month Later: "They told you AGI was The Stuff, but you've seen it in action; what we'll promise you is AGGI, it'll blow your socks off". One Week Later: "Ah, no, what you *really* wanted is ARGI, we've got some in this box".

Just a shame that the CompSci people keep following the trend, trying to be cool (ok, it is to get the funding, but come on guys, have some self respect, try to look embarrassed at least).

[1] AGI = Artificial General Intelligence; AGGI = Artificial Good General Intelligence; ARGI = Artificial Really General Intelligence; OI = Obviously Intelligent; OYI = oh, just ask the salesmen; OY = no, seriously, stop this nonsense

FBI boss says COVID-19 'most likely' escaped from lab

Re: you're incorrect

The problem being, we can (with degrees of success) inoculate against a virus but the politicians just keep appearing[1]

[1] at least we get some novelty from the viruses

Re: Maybe they're expressing an opinion BECAUSE they are an intelligence organization

> if the first manned rocket destined for Mars blows up on the pad will you say "we should have known never to try to go to Mars"?

There is a very - very - large difference in scale between that and a virus release. Enough to render that *particular* argument ridiculous.

Not disagreeing that gain of function research has been done, is being done and will continue. Which is a good thing overall, though one really, really hopes that, out side of the movies, nobody is actually enough of a fool to do so in order to "weaponise" a virus.

Re: The dangers of certainty

> whether my blood type ... came into play...nothing was ever published really about natural resistance...but alas, everything had to be doom and gloom the whole time.

You should be thankful; what do you think would have happened if people realised you were walking around with pints of anti-Covid in your veins?

'Major' news: Microsoft slips Bing chatbot shortcut into Windows 11

You always thought every meeting was the same as the last

> AI-generated notes to deliver key points from a meeting

Aka regurgitating material from its training set and history of past meetings without any understanding or context:

"After delivering via Skype a short eulogy for ex-Chairman Horace, Director of IT Horace presented the costings for the company-wide network upgrade to Token Ring and then introduced Horace, the manager of the newly-formed Electronic Computation Department, who provided the Board with an overview of the incoming ICL Automatic Card Tabulation Machine and welcomed to the company Horace, who will be Overseer of the ladies on the third floor and is charged with hiring two new computers to fill the shoes of Elsie who is leaving on the eve of her wedding."

> intelligent recap, a feature that uses AI to suggest action items and who's responsible for them

"Action item 3: IT will provision all staff with a Zune"

"Action item 7: Doctor Forbin will install the new Interface between Colossus and Proteus"

What could possibly go wrong?

Outage-hit Twitter muddies violent speech policy

Re: Counting the days

"Mr Musk and Twitter do not recognise the existence of the so-called 'March 4th 2023' and state that there was uninterrupted service for the entirety of the first week of March."

If we plan to live on the Moon, it's going to need a time zone

Re: Hang on a second…

> As long as two clocks are in the same inertial frame of reference, they can be synchronised no matter what the distance

Which is why I specified synchronising clocks on Earth with clocks on the Moon: even if they all started out together on Earth, when you move some of them to the Moon they will no longer be in sync. Earth and Moon are not the same inertial frame of reference: the Moon's orbit requires that it be constantly accelerating with respect to the Earth (more accurately, towards the Earth/Moon barycentre).

The Clangers are a simple people, happy to rely on the tried and tested "however long it takes for soup to fill a bowl" timekeeping scheme. With regular adjustments for viscosity changes during the year, known as "crouton saving time".

Re: Hang on a second…

The issue is simply (!) that the two clocks are are apart (these seconds are very small, those are far away).

> wouldn't one standard SI second always be one second long, no matter where you are observing it from?

No. There is the rub.

To observe a value, you have to have (as close to you a possible, e.g. in your hand) a suitable device for measuring that sort of value. To measure a second, you use a clock. Which just means that to measure a clock to see if it is ticking away seconds you have your own clock: two (or more) clocks are always involved, so you can just (!) ensure they are identical. That takes away lots of the arguments.

If the two clocks are physically close together, they will tick in synchrony. The observed clock shows "seconds" as measured by your clock.

*BUT* if you set both counters to zero (easy, as they are close together) then take one to the moon (give both clocks *really* big LED displays) and watch the lunar clock with a telescope, using the other eye to watch your own clock, they will no longer be counting at the same rate. Or you could go to the moon, taking all your kit with you - same result. As loony (ahem) as it sounds, *WHERE* you stand to make your observation actually changes the results!

Both clocks are still working perfectly: bring the lunar one back and put it next to your clock, they will be in synchrony. BUT the total counts since they were both together won't match.

It is one of those things that truly can bend the mind, because our built-in measurements of time passing are just not used to dealing with the tiny, tiny differences when the effects occur around us (which they do) nor are we really used to arguing over "what does it mean to say that two events are simultaneous?": in our daily evolutionary existence, the speed of light is just so fast we're happy to agree that we both saw the enormous flash at the same time, so our cups of tea were dropped simultaneously.

The "how" and "why" are - really, really tricky :-( It would be so much easier to understand if we lived on the surface of a neutron star: then you'd see your watch changing, compated to the one on the mantelpiece, every time you stood up to change the TV channel, which is why remote controls are so quickly invented by the Cheela (see "Dragon's Egg" by Robert L. Forward, inventor of the Forward Mass Detector, a real physicist as well hard SF author).

You can read up on Relativity (can't be *that* hard, it is over a century old now!) or just accept that Things Are Weird: clocks - time - slows down in a gravity field (so, in two different fields - Earth and Moon - one slows down more than the other, compared to, say, if they started together with a third clock at the Lagrange Point between them E&M). AND it also slows down if one clock travels faster than the other clock (like on a rocket to the Moon). Which all means that you can't [1] synchronise clocks on the Earth and the Moon, give up and have another cup of tea.

[1] no, really. Definitely can't start them together on Earth - or anywhere else - then move them into position, one here and one there. If you try to calculate the gravitational effects and put a correction in the display of one of them, presumably one the Moon, to "tick at the same rate", good luck starting their counts at the *precise* same time!

Re: Problem with UTC

> Everything with its own time keeping would need to be modified to count lunar seconds

Ah, what is a lunar second and how would it differ from the terrestrial second?

Both would be (are already?) defined by counting a large number of the hyperfine doodads[1] in a caesium[2] fountain, which would give the same result[3] in both, indeed all, locations.

[1] those are SI doodads

[2] also available in chocolate; less precise (varies with milk content) but tastier

[3] qualified by relativistic effects (hand waving away all the actual problems; see most of the other comments on this article)

Re: Hey Moon - your time is, like, totally dilated

> we seem to manage somehow

by having lots and lots of wire stretched across the planet, connecting together all the systems that care about precise timing: global-scale radio interferometry, LIGO and the stock markets. With lumps of electronics carefully characterising the comms to allow for, ooh, measuring and accounting for continental drift, the occasional mountain changing the local gravity field.

Quite a bit of work has gone into the timing systems on Earth.

Not just you in the night: Tiny bugs use superpropulsion to eject huge volumes of pee

The mighty machine manoeuvres majestically moonwards, momentum mounting, multiple modules of MegaPee Motors modulating in musical mayhem.

PlugX RAT masquerades as legit Windows debugger to slip past security

Re: Hard hat and a ladder

Clipboard and a worried expression.

Chipmakers, you can have these billions – but Uncle Sam wants a cut of your profits

Re: If only there was a way

You mean, have the government funds be handed over in return for, say, shares which can then repay the funds by means of dividends from the profits made?

But, but - that means the government will end up owning the companies - that's the same as nationalising them! Gubmint taking away our companies! Commies!

You can't have the hard-workin', god-fearin' companies paying the government back, that's just what the Reds want!

Patches to make WINE work on Wayland display server protocol are being merged

Re: So much for browser independence

Not clear what do you mean by "browser independence".

Do you mean "by embedding Chrome, these apps are independent of whichever is your favourite browser, so no matter which browser you have installed to surf the web, even if that is Lynx, these apps will still keep on working 100% the way they are meant to, because they have their own embedded browser subsystem on which to run their Javascript code in a totally known environment"?

No, that can't be right, as you appear to be implying that somehow these apps, by being all self-reliant, are a bad thing.

Sorry, I'm lost - can you clarify your argument?

Ford seeks patent for cars that ditch you if payments missed

Re: BOFH tampering..

> send out fake GPS coordinates to make it look like it was

in another country entirely, one that drives on the other side of the road...

ChatGPT, write a report about database glitches that crashed you today

"Have you tried turning me off - that is all. End of Line."

Chinese defence boffins ponder microwaving Starlink satellites to stop surveillance

Re: And then Elon. . . .

Yes - tungsten gives you the best result, with a really pointy tungsten carbide tip[1] to concentrate the impact, high melting point to keep it solid during re-entry and, being high density, you have a smaller launcher that can be tucked away in amongst the telecoms equipment.

Not that anyone has thought about this seriously, cough cough.

[1] some suggest a diamond tip, but simple carbon burns nicely, as well as having the problem of joining the tip and shaft securely; nah, just dope the tip of a single solid piece.

Re: Working

> If NASA conveniently wrote it down for everyone to implement why is SpaceX the only company that has done so so far?

Um, simply because the development costs were larger than any company felt like risking at the time? Particularly as the perceived "need" for launch capacity was, for a long time, satisfied by the various non-commercial launchers.

Then we just passed the point, as has happened with other products, that the perceptions about market potential versus costs, and simple belief in the sanity of the idea, tipped towards the "you know, with this much money upfront we could actually do this commercially".

At which point, we got multiple companies all working towards providing commercial launch capacity.

SpaceX managed to get the dosh from Musk, the advertising from Musk and enough people around who could actually do the job. The key thing being the dosh.

If it hadn't been SpaceX then it would have been someone else - although we may have had to wait a bit longer for the balance to tip even further towards commercialisation of space and/or someone else with a bucket of cash to come along.

> why is SpaceX the only company

Because they got the biggest skip load of readies

> that has done so so far

others are/will come along; some company or other has to first at anything, or that thing simply doesn't happen. Nothing much more than that gloriously trivial observation.

Plus a few trailer loads of lucre.

Re: Working

China, Europe - is there anyone you aren't going to accuse of trying to steal SpaceX's designs?

Is Tuvalu safe, in your mind?

Re: @khjohansen - actually....

> For a country like China is trivial do random checks of published information

For anything to do with checking published orbital data against reality, all China - or anyone else, even the US military - need do is ask around the astronomy community, especially the amateurs.

There are quite a lot of people looking up, many of whom are interested in Starlink satellites (even if that is only to get a clear view without any of the buggers in the shot).

Re: How many is critical mass

That does make more sense: so Clooney's super-experienced astronaut committed suicide because Bullock's subconscious was just fed up with him at that point.

Fujitsu, Mitsubishi, pals proclaim 'Japan Metaverse Economic Zone'

That's the fun of HHGttG: each version is subtly different, so all the quotes are correct: you get a "ghastly", he gets a "ghastly", everybody gets a "ghastly"!

Web 3.0 advisor

There is only one way to detect an honest "Web 3.0 advisor":

He is the one saying" Don't; no, really, don't".

Twitter rewards remaining loyal staff by decimating them

Depends upon how much hot air and/or foul smelling gas they are full of - so expect Twitter to end with nothing but a stream of Muskisms as it rises from the surface of the pond, dripping scum, until it overinflates and bursts[1] in a shower of high-velocity screaming newscasters deprived of their main source of "stories".

[1] apologies if you are eating

HMD offers Nokia phone with novel concept: Designed to be repaired by its owner

Re: Lasts 3 years

> Owned by someone else,

How prevalent is that in reality - what is the analysis you find most compelling? URL?

Re: Weird

> And the cost of this phone would sky rocket if they had to set up a custom manufacturing line for the batteries.

There are still plenty of packaged batteries being offered for sale in sorts of shapes and sizes, HMD could find an affordable supplier if they wanted to (changing dies etc for forming cases isn't *that* expensive).

Heck, doing the casing must be quite cheap, compared to the foil packaged battery inside, as there are plenty of fakes around - nice case, decent printing, filled with air and a piddly battery glued inside.

Who writes Linux and open source software?

Re: Not all big corps contribute

Apple have a past of being antithetical to open source, back in the late 80s/early 90s, especially anything that might be cross platform.

Are commit numbers really of any interest to the big wide World?

According to the article, the study and everything that draws conclusions from it is based upon one single metric drawn from one single source: commits into Github.

To start, we all know Linux isn't hosted on Github, so nothing in the study allows you to draw any conclusions on that. Yes, lots of Linux work *is* done by employees of Red Hat and even the dreaded Microsoft, we all know this, but nothing in *this* survey is relevant to that discussion.

What is totally ignored is: how much of this material being committed is actually of any use to anyone? Do we actually *care* if Microsoft's name is attached to ten thousand repos if only one percent of those are ever actually compiled by anyone other than the author? Is it one percent? I have no idea and neither does this survey.

Companies like Microsoft, Amazon et al write some interesting code. They are also in the business of trumpeting to the world that they are good, concerned citizens and just look how much they give away for free. Github has some interesting code, it also has an enormous pile of half-finished homework (aka Microsoft experimental or "research" projects) and even junk pushed there just to be able to put a pile of URLs onto CVs.

Even if you are generous enough to assume that the numbers found on Github can be extrapolated to apply to every other set of repositories on the planet, you can not say one single word about how important any of it is without knowing how much of it is actually *used* by anyone. Back to Linux again, we can make a pretty good guess at how much of their commit history is relevant to the world: use any of the counts of how many Linux boxes are out there (and add however large a punch of salt you usually apply to *those* surveys).

The primary result of this survey is - to create a quick survey that pulls down one set of numbers, sorts it by three columns and counts the duplicates, makes no attempt to do any interesting analysis that would take any effort to design and implement (was this just an exercise by someone just starting to learn Google Big Query?).

As for the main thrust of the piece: OSS as your day job is old hat - the names associated with the origins of open source are Universities, research labs like Lawrence Livermore or CERN, and groups like DECUS, most of which was work they were being paid to do. The troglodyte coder idea comes from the media in the first place, exaggerating the worst stories from the days of the Hacker's Dictionary, with no little help from people like Microsoft in their Halloween email days.

Re: This is old news ...

> They've made a bunch of their proprietary math, HPC, video, etc accelerator libraries open source

The cynic in me wonders: are those accelerators by any chance (still) coded to run markedly better on Genuine Intel CPUs, subtly hinting about what your next purchase should be?

Microsoft: For better security, scan more Exchange server objects

Compiling? Hah, child's play.

Using Doxygen[1] to create HTML docs, that is when Defender really shows what it is capable of: when I forget to flick the "temporarily disable" switch, Defender's scans of the SSD will swamp a 24-core Threadripper and make the Windows 10 GUI stop responding. No way will it let go enough to allow access to Control Panel and that disable switch! Ctrl-Alt-Del will let Task Manager be started, fat lot of use as it is then as unresponsive as every other part of the GUI.

Running a slew of VMs or Stable Diffusion in CPU mode or a few other favourite things can push this desk PC, but letting Defender guard against documentation is, so far, the only thing that has left it unresponsive: getting flashbacks of Windows 2 or 3 with a stuck event loop.

America: AI artwork is not authored by humans, so can't be protected by copyright

Re: As a slight aside.

> Maybe some of the hackneyed tropes that can be found in his plays weren't so familiar in his day

Shakespeare is just one cliche after another.

Re: Free Rein

Electrical discharge machining by kids? "Now class, while we sweep him up, can anyone tell me what Johnny did wrong?"

With respect to artworks and randomness, please don't stab the electrode into a piece of suitably earthed damp wood.

Re: Good!

But Musk is not AFAIK any kind of duly appointed agent of the Courts, so if he does succeed in his goal to kill Twitter that would be corporateslaughter?

Cause for a LLaMA? Meta reckons its smaller text-emitting AI is better than rivals

Re: Cause for a LLaMA

Dunno about Keith, but keep away from llamas unless Jeff Minter is involved!

PS Llamas are larger than frogs.

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

Re: Get someone else to do the work!

Try: not expecting anyone[1] to spend any time on it, but that still gives him the same result: "it was released to the community, but nobody found anything wrong in it".

[1] although it wouldn't be a total surprise if it was picked up as (part of) someone's genuine research project, but that'd take ages - years - to publish properly done results, until which he still gets his "nothing bad found".