Comments

Clive Robinson • February 14, 2025 10:10 PM

@ Bruce,

Italian’s caught at it again.

As you are probably aware I point the finger at both Israeli and Italian “companies” that closely align with their respective Government Agencies to produce “Spyware” and worse. That they also make available –often through Bulgaria– to just about anyone with the money to pay… And so people get hurt and killed by these companies profit making.

Well here is another article confirming an Italy Company and it’s Government are still at it,

“Italian spyware maker SIO, known to sell its products to government customers, is behind a series of malicious Android apps that masquerade as WhatsApp and other popular apps but steal private data from a target’s device”

https://techcrunch.com/2025/02/13/spyware-maker-caught-distributing-malicious-android-apps-for-years/

And apparently not just Android,

“Kaspersky said in a 2024 report that the people behind Spyrtacus began distributing the spyware through apps in Google Play in 2018, but by 2019 switched to hosting the apps on malicious web pages made to look like some of Italy’s top internet providers. Kaspersky said its researchers also found a Windows version of the Spyrtacus malware, and found signs that point to the existence of malware versions for iOS and macOS as well.”

Dancing on thin ice • February 15, 2025 11:16 AM

@ Tom Kenney

No, much of what is being accessed includes classified or personal information like social security numbers by those that would not pass a background check..

This site is about security issues.
Young inexperienced engineers not following basic security practices should not be glossed over just because one blindly agrees with a certain political viewpoint.

Dancing on thin ice • February 15, 2025 1:18 PM

@ Tom Kenney

Examples of things that not everyone needs to know are our social security numbers, military capacities of allies and advesaries and criminal investigations.

Things that may raise red flags in a background check are people I knew almost 50 years ago but have worked with local political leaders of both parties for the past 15 years or so.

Clive Robinson • February 15, 2025 3:24 PM

@ Bruce, ALL,

The wall Sam denied he has run into.

Prof Gary Marcus who has madeca few comments and books about current and earlier AI systems over the years has written a piece that people should read,

https://garymarcus.substack.com/p/breaking-openais-efforts-at-pure

His opening paragraph resonates with my thinking and what I’ve been saying for a while now, so I’m neither shocked or surprised, but others almost certainly will be.

As he indicates for Altman the dog has not barked… and from my perspective it almost certainly won’t with current LLM and ML systems.

As Prof Marcus dryly observes,

“All this time, it turns out, Altman was bluffing.”

I’d disagree, bluffing is not the word I’d use… Of the systems Altman has pushed there are two terms of art “Hallucinations” and “Hard Bullshit” you can pick your choice which applies most to Altman’s words.

However consider what he and similar have done in terms of “knowingly misleading” those who would be and have been invesstors.

It’s hard to say just how much money has disappeared into the current AI money pit… But we do know that it more than tripled the value of Nvidia well into the trillions.

The same nonsense is what “Hell-on Rusk” is selling to POUTUS to “run the country under his control”.

It would be considered “unfair” if I was to scream “Scam”, but I have been warning that the current AI LLM and ML “GPT” systems bubble really is “puffed up” and full of “hot air” and lacks any recognisable substance beneath all the “windmill in a storm” arm waving, which is rather more than a broiling “tempest in a teapot”.

Because despite Altman’s confident words Orion is not and nowhere near Altman’s GPT-5 and AGI… So Orion is re badged into potential ignominy of at best 4.5…

Put simply “Scaling has failed” it won’t give you AGI, but lets be honest neither will,

1, The worlds largest library
2, The worlds fastest database

Which at the end of the day is all LLM systems are, in the same way all current computers are in fact Turing Engines. They are useful but some things they just can not do.

What will cone after LLM and ML systems, all anyone can really say is,

1, I don’t know.
2, It’s unlikely to give AGI.

Which begs,

“Does this make LLM and ML systems useless?”

No far from it, they will however eventually become niche rather than general as all the systems before have. But it leaves a rather awkward question just hanging on a single hair at the feast like the sword of fame,

“How much damage will they do untill then?”

Untill a few days ago, I’d assumed that the “sign posts were sufficient” for all to see by now. But apparently not… DOGE under Musk appear determined to keep such systems bringing in the billions “for the chosen few” from the nations Tax Dollars… By in effect playing a “shell game” with them.

Thus it should be asked,

“Who apart from the corrupt, want hallucinating, hard bullshiting, glorified random text predictors running the country?”

Think carefully before you speak, lest you get the equivalent of being strapped in a chair and forced to watch endless reruns of “Fawlty Towers”.

[1] You can read more on the Prof at,

https://en.wikipedia.org/wiki/Gary_Marcus

jelo 117 • February 15, 2025 5:20 PM

There have been some recent articles examining record game runs for evidence of fraud. The analysis method goes through effectively the entire digital game data space, including the large but in the end finite “randomly” generated level and map data, to see if the record run falls within the data space
legitimately reachable by the program code.

That is, the complete set of trajectories of the discrete digital dynamical system that is the game is examined.

Perhaps this general method could be used to improve performance and security of code in general. It might even assist in analyzing issues arising from hardware anomalies induced by strange inputs.

Clive Robinson • February 15, 2025 6:53 PM

@ jelo 117,

With regards,

“The analysis method goes through effectively the entire digital game data space, including the large but in the end finite “randomly” generated level and map data, to see if the record run falls within the data space
legitimately reachable by the program code.”

Ever hear of a “Drunkards Walk”?

There are others better qualified to answer this in depth but you are running into the same issues we do with the movement of particles in a fluid when subject to uneven thermal excitation.

On mass the particles behave in a statistical way and you get what you were taught in school as “Brownian Motion”. However individual particles do behave entirely randomly unless they interact with another particle or boundary.

Without resorting to a white board and lots of squiggly notation it may not be immediately obvious that each and every particle is in effect acting as a Turing Engine and thus the “Halting Problem” and “undecidability” apply…

This actually means that for the game, you can due to statistics show cheating is going on, however it also indicates that on the individual game play you can not decidably show cheating is responsible for the moves an individual makes in that one game.

In part it’s the issue that gives rise to statements like,

“Correlation not causation”.

In short “Random is noise” and “you cannot show a signal exists in noise when the bandwidths are equitable (it’s why using artificial noise you can put a narrow band signal well below the noise in a wideband channel, and is the basis for not just all “Spread Spectrum”(SS) “Low Probability of Intercept”(LPI) systems but also some of the more modern MIMO systems).

In fact there is a body of work on this from the mid to late 1990’s and it falls under “Digital Rights Management”(DRM). The reason DRM of this form failed in the end was that an opponent could control the characteristics of the channel the DRM SS equivalent “signal” was sent in.

Put simply humans can deal with various types of distortion without much perceptual interference. However the DRM signal being very reliant on auto-correlation can not deal at all well with those types of distortion. In effect you end up in an “arms race” where the DRM signal either fails, or is of such a level it noticeably interferes and is apparent to the user in an unpleasant way and also can be “tracked” thus “removed” by what is in effect “an attack in depth”.

Clive Robinson • February 17, 2025 6:38 PM

@ Bruce, ALL,

I don’t know if you’ve seen this,

https://www.cs.auckland.ac.nz/~pgut001/pubs/bollocks.pdf

Yup it is from who you think it is and to say he’s,

“Not best impressed with Quantum Cryptanalysis”

Might be considered an understatement 😉

P.S. As some know my “First love” work wise is hardware design for communications. I’ve designed more “embedded systems” with communications than I can actually recall without looking them up. My fun for thinking about was “security” way back when even 8bit CPU’s were “new and risky”. That said whilst hardware design brings in the bread[1], the security stuff puts a nice thick civilising dollop of cream and another of jam on top spread thick enough to get in the beard 🙂

[1] On one side of the pond, you call them biscuits and eat them with “redeye gravy” –strong black coffee and bacon drippings– for breakfast. On the other side we call them scones and consume them with lashings of high quality clotted cream and fruit jam along with a good pot of tea mid afternoon. I know which I find more civilized 😉

I’m told that in NZ afternoon tea is also considered “most civilized”.

Clive Robinson • February 18, 2025 3:52 AM

@ CDN Man, ALL,

Your,

“With the power to… doubt?”

Reminded me of the old comment as to why orthodoxy was dangerous by way of alluding to lemmings following each other over a cliff.

Thus,

“To doubt not follow, to question not accept, to prove not presume, are the steps to insight and understanding.”

Which was fresh in my mind after thinking about the issues we currently see in Security.

And why in some parts of the world to question the presumptions of,

“Short term thinking, shareholder value, neo-con mantras and the capitalist way of forever debt.”

Is a way to get your thumbs “nailed to the wall” for the crimes of “impertinence and heresy” or just “cutting your own path or furrow”.

But the supposed advantages of the neo-con capitalist presumptions are when you cut down far enough all based on false axioms about growth and “growth for growths sake”.

What brought that to mind was reading,

https://www.scattered-thoughts.net/writing/small-tech/

Again a near half decade after it was written, and thinking about if it’s points are still valid…

Dalin Owen • February 18, 2025 6:42 AM

@Clive

AI is powerful, no doubt, but it also comes with some serious unknown risks.

Take DeepSeek, for example; it’s a tool that can analyze huge amounts of data, and while that sounds amazing, it can get dangerous fast.

If it’s not used responsibly, it can perpetuate biases, invade privacy, or make decisions without anyone really understanding how.

Plus, there’s always the risk of bad actors using it for things like manipulation or surveillance.

We’ve got to be careful and make sure these technologies are developed and used responsibly before they spiral out of control and I’m doubtful that the recent AI summit will help at all…

Clive Robinson • February 18, 2025 12:45 PM

@ , ALL,

A lot of the risks have been very well known for well over a half decade. And can be seen given in,

https://www.brennancenter.org/our-work/research-reports/predictive-policing-explained

One of the “new risks” is something I’m still gathering information about.

As I’ve indicated for a while now current AI LLM systems are basically DSP filter systems where the “output spectrum” is a semi-random “next likely word”. With the additional ML component added making them similar to “adaptive filters”. Where adaptive filters were first used noticeably for “echo cancellation” on long distance phone communications.

Hence some call the current AI LLM systems glorified predictive text engines because of the way they pick words.

However it’s been realised by some that the words chosen and presented to the LLM system can be used to create cognitive bias in the user, that is usually unnoticed by them, but is sometimes picked up by those that are around them and know them reasonably well.

Which tends to make me think we’ve hardly started on the “biasing the user” issues (as opposed to selecting a user by bias).

The alleged “Ancient Chinese Curse” of,

“May you live in interesting times”

Appears to apply more and more on an almost daily basis with current AI LLM and ML systems. The difficult part is telling the “actual reality” from the “talked up” hopes / fears / day dreams of those “with skin in the game”.

As for the AI summit, it revealed quite a lot about the participants natural predispositions to ICT and Corporate interests.

The UK and US essentially “walked away” with a “nothing to see here” attitude, whilst the continental Europeans were a good deal more cautious and in effect saying “there be dragons here on the AI future map”.

If anything I’m with the Contenental European view point, and make a tentative prediction most of the issues will arise from the UK and US and other “nothing to see” or “hands off it’s the future” view points based on “short term thinking” driven by “investment bubbles”.

As I’ve said before these current AI LLM / ML systems are a “follow on” from the Web 3 / NFT nonsense and earlier Crypto Coin speculation.

Way to many people think “churn” created by these nonsense technologies is the future “economic activity of nations” to which the only logical response is,

“God help us all from the plague and blight it will bring.”

As for the US players as I’ve already indicated Alphabet, Meta, Microsoft and others have a business plan for these systems and it’s “heavy weight surveillance” based on,

“Bedazzle, Beguile, Bewitch, Befriend, and Betray.”

How far you fall along this path depends on your susceptibility. But it’s fair to say that “Befriend and Betray” is what the “AI Agents” these Corps have planed are very much designed to do.

The real issue not being,

“When they Betray an individual”

That’s expected from the “get go”, but more importantly,

“When the individual finds out and by how much harm it has caused them.”

It is the nature of these things that they will destroy people before there is sufficient push back created for hard legislation to be enacted. And we can tell from the behaviour of Sam Altman etc that they are very Cognizant of this hence their “bull to the gate” attitudes.

As the old saying has it,

“You have been warned, take note if you value your life.”

Clive Robinson • February 19, 2025 9:37 AM

@ Bruce, ALL,

The next snouting venue…

There are at least two Camels in the US Executive shoving their less than desirable appendages where they really are not wanted, not just into the flaps of tents.

As is suspected by many the current POTUS for some “political reason” –see Orwell’s 1984– is creating issues with China and claiming it’s all “National Security”…

The reality is a little more complicated and the recent “OMG Shock Horror” of the US AI cleaque getting trounced by a bunch of Chinese “alleged” startup students when “DeepSeek” released it’s R1 system. Giving rise to MSM titles such as,

“China’s cheap AI melts NVIDIA supremacy”

And the US stockmarket wiped $600 billion of NVIDIA, and well over $1 trillion from US tech and energy companies with upto double digit percentage losses in value.

Thus POTUS had to bite the bullet and say the usual nonsense of US industry must nerd harder, and take the Chinese technology…

But in reality that is not POTUS’s game plan… Some might remember last time he was making wild claims about China stealing US Telecoms technology and when that failed started his anti 5G nonsense…

The reason is that much of 5G is based on Chinese developed technology that is Open Licenced through the GSM standards. However US telecommunications corps could not compete with the two Chinese Corps and with the NSA being caught red handed “backdooring US products” the international market was decidedly fridgid with respect to US Corps products.

So POTUS claimed all sorts of nonsense, not because he wanted to stop his dreamt up spying idiocy but because he was advised to “Kill 5G” so “6G” would give US Telecommunications Corps a new chance to be preeminent.

Because half a decade and more ago 6G was effectively just a marketing projection label. Because there were no standards, specifications or radio allocation. Even algorithms and methods were just buzz words.

Since then some things have changed though 6G standards etc still really do not exist. Politically there are those in the US who thing they can “Grab 6G” to “Make America non-G Again” and make the world follow and pay pay pay, but it’s actually quite unlikely.

Worse 6G may not happen… Most are finding 4G-LTE “meets the needs” not just of Service Providers but their customers as well.

Certainly in the US the bad behaviour of the big Telco’s is such they are actually killing the mobile phone market… But also customers world wide are not using bandwidth as expected so 5G is “over the top and well over priced” which means 6G with all the problems it has so far is not a realistic prospect even for 2030,

https://www.theregister.com/2025/02/19/ngmn_6g_report/

Oh and as the comments note even in the UK 4G is something of a scarcity out of built up areas and motorways.

Fun fact the UK “Trunked Radio for Emergency Services” based around TETRA has been one disaster after another and is finally getting replaced. The proposal is to take 4G UK wide as a backbone untill at least 2050 to do this as well as make it public accessable even a few miles out to sea. Railways however due to EU Regulations are likely to remain 2G at best for another few decades.

But as I’ve noted before, there is so much infrastructure investment with 2G it’s probably still going to be around after most of us have had the lid nailed down…

So 6G… Probably not, I can see global coverage of 4G from Sats being more likely, which will no doubt please those who go to sea.

Patrick • February 19, 2025 8:35 PM

Thought you’d be interested in this from the POV of: what happens when you can’t trust your email client anymore.

Yahoo Mail replaced the email subject line with AI summaries, without any indication in the UI that they were doing it.

Caused total chaos at my startup which runs online sneaker launches. I got paged over the weekend when thousands of people were told they’d “won” a sneaker raffle connected to NBA All Start Weekend.

Awful to think how it could affect people getting emails about medical tests, job application results, etc.

I wrote up the story of how we figured out what was happening here: https://www.eql.com/media/sneaker-bot-ai-error

Clive Robinson • February 19, 2025 9:24 PM

@ Patrick,

Hmm current AI at it’s finest…

Atleast you have sufficient “messages in depth” to diagnose the problem.

I can see such an AI mechanism changing things such that some one could end up looking racist or worse “hurtful” of others feelings.

In the UK there is now legislation that can criminalise a person for allegedly hurting another persons feelings. Without there actually being another person or a person saying their feelings are hurt…

All it requires is an “in the opinion of” decision by a person in nominal authority…

So you can imagine how much the UK Police and other Gov agency personnel are already abusing that legislation…

Such changes from pro to anti could cause real issues…

As you know poetry and similar can use dramatic effect. So as the 14th has just gone by I wonder how something like,

On this day,
I find I must say,
You are not my friend,
You are not my love,
You are in every way,
My soulmate true.

Could get summarised by an “on the cheap” / “low budget” AI…

Clive Robinson • February 19, 2025 10:08 PM

@ Bruce, ALL,

I’ve mentioned in the past that Silicon Valley Big Corp’s business plan for current AI systems is,

“Bedazzle, Beguile, Bewitch, Befriend, and Betray”

And that as “End to End Encryption”(E2EE) NOBUS was sufficiently bad news in even the MSM it was being made in effect “impossible” to comply with LEO “KeyMat” requests etc by developers of messaging Apps… That “Device Side Scanning”(DSS) of User Interface”(UI) “plaintext” to get around encryption entirely would be used instead.

It now appears to be the case that all Mobile OS’s etc have DSS “built in” and designed not to be disabled or removed as Google inflicts DSS on Android.

It appears I’m not the only one getting annoyed at the “misinformation” being put out about DSS,

https://m.youtube.com/watch?v=6bPJzLXdEcA

For those that do not like being “Spied On” in this way for either Professional or Personal “Privacy” reasons, your options are at best few. Because it will quickly become a “Red Queen’s Race” where every OS upgrade will turn DSS on in new hidden ways to stop it being turned off by the user who payed for the device.

The only option is not to have sensitive “plaintext” or “KeyMat” “pass words/phrases” on the mobile device.

I’ve warned for some time about the issues of this to do with “secure messaging apps” actually being insecure due to the “security end point” and the “communications end point” being on the same device. And indicated you need to get the “security end point” and any sensitive plain text “Off Device” and so out of reach of the OS and “communications end point”.

This is even more true for “Device Side Scanning” no matter what it is called.

lurker • February 19, 2025 11:14 PM

Quantum computing in our time? MS has invented a chip using Majorana particles (BBC description, not mine). If their hardware is as good as their software, what could go wrong?

‘https://www.bbc.com/news/articles/cj3e3252gj8o

Clive Robinson • February 20, 2025 7:03 AM

@ Lurker,

With regard “Majorana particles”

A readable MSM article from a little over a decade ago,

https://www.theguardian.com/science/life-and-physics/2014/oct/05/majorana-princeton-particles-fundamentally-confusing

The author is Prof Jon Butterworth at UCL,

https://en.m.wikipedia.org/wiki/Jon_Butterworth

I’ve read some of his works and sat in on one or two talks/lectures before when I had reason to “stomp around” UCL. They appear to my eyes and ears as what you would expect from a particle physics researcher, that can actually “communicate” with meer mortals 😉

So make of the article what you can/will.

Bob Paddock • February 20, 2025 8:36 AM

@Clive Robinson, I know you’ve discussed the problems with these in the past.
Because they can be overloaded with things like RF, is that why they are now becoming the Industry Standard?

“CEA-Leti Launches OpenTRNG, an Open-Source Project For True Random Number Generators Using Ring-Oscillator-Based Architectures”

“…open-source project to produce physical True Random Number Generators (TRNG) using ring-oscillator-based architectures. Targeting industry and academic researchers, the project’s comprehensive toolkit includes reference designs, emulation tools and analytical tools to facilitate development and characterization of hardware TRNG implementations. …”

‘https://www.cea.fr/cea-tech/leti/Pages/actualites/Communique%20de%20presse/CEA-Leti-Launches-OpenTRNG-an-Open-Source-Project-For-True-Random-Number-Generators-Using-Ring-Oscillator-Based-Architectures.aspx

Clive Robinson • February 20, 2025 12:05 PM

@ Bob Paddock, ALL,

With respect to “ring oscilators” as “True Random Number Generators”.

There used to be a truism said about Microsoft of,

“What ever the question, the answer is not Microsoft!”

The same can be said of “Ring Oscillators”.

I’ve mentioned in the past why Intel had to hide their “Ring Oscilator” TRNG source behind a Cryptographically Secure algorithm, due to Ring Oscillator deficiencies and the fact whilst their outputs appear random, they are not even chaotic but reasonably predictable as the fraction that is “true entropy” is incredibly small to vanishing.

Further consider the basic design, of two oscillators feeding the input of a D-type latch or XOR gate with the output being the raw entropy source.

As I’ve explained in the past the D-Type latch behaves like a “mixer and low pass filter 1bit digitizer”

What it outputs is a one bit wide signal that if integrated will produce a very pure sine wave at the difference frequency of the two input oscillators.

The thing is, if you look “close in” at the one bit output with an oscilloscope “it looks random” to the human eye. However open the time base out and you start to see the pulse widths bunch up and apart cyclically. Open it out further and you can see this bunching forms a pattern that forms a sine wave.

Thus the output of the D-Type is f1-f2 + phase noise. The phase noise depends on,

1, The ‘non’ correlation between the oscillators.
2, The independence of the inputs to the oscillators via power supply, circuit and mechanical isolation.

Which on a single silicon chip is such that they are all very very low, thus the phase noise is likewise very low.

All to often to try and improve “the look” one oscillator is tied to the CPU XTAL that provides the entire chip “clock” and the other oscillator is sort of “free running”, but it’s frequency usually can be seen to be a significantly correlated to the chip temperature and other predictable factors…

Which means an attacker can get one oscillator input frequency just by observing the CPU outputs to get an accurate fix on the XTAL frequency. And predict the other from being able to deduce the CPU temperature (from Delta-f of the XTAL)…

Knowing this I suspect many can work out an attack plan on this supposed TRNG in effect “passively” say by looking at “network traffic” timing (I’ve discussed this on this blog in the past). To save the grey cells others have written about it,

https://lcamtuf.coredump.cx/oldtcp/tcpseq.html

https://lcamtuf.coredump.cx/newtcp/

But also consider who is behind this “recommending” organisation?

And say their connection to ETSI through the French Government…

As we know the French Government detest cryptography being in “Non Government or Foreign Hands”. We also know that ETSI have backdoored or in other ways weakened all communications security in several communications protocols over the years like those for GSM mobile phones and digital hand held radios.

We also know that all “Roots of Trust” used for security, must be at the very minimum,

1, Entirely random with high entropy.
2, Unknown / unguessable to all third parties.

I’ve pointed out that the NSA most likely are overjoyed with the very very weak entropy in embedded devices used at the “network edge”. Where roots of trust are made “on device” very very soon after initial turn on. Thus the actual entropy is at most down to little more than a few bits (say 2^10 or slightly above). Certainly within a Brut Force search range.

It’s known that ring oscillator TRNG’s are especially bad at turn on and really should not be used. Something that “Mr Linux” got all hot under the collar about with the RNG in the Linux kernel and then had to apologise about shortly there after.

https://www.reddit.com/r/linux/comments/1lucdy/did_linus_torvalds_backdoor_linux_random_number/

So yes my nose is twitching at this recommendation and it smells quite rotten.

Clive Robinson • February 22, 2025 11:38 AM

@ Bruce, ALL,

Update on Apple v UK Gov.

https://www.bbc.co.uk/news/articles/cgj54eq4vejo

Apple pulls data protection tool after UK government security row

“Apple is taking the unprecedented step of removing its highest level data security tool from customers in the UK, after the government demanded access to user data.”

This raises several technical questions about not having your data in Apples Cloud.

The first being

1, Can you really stop Apple uploading “your data”?
2, Can you really stop Apple doing device side scanning?
3, Can you upload data to the Apple cloud from an App etc that uses only a key that is always in the users control only?

Oh and of course,

4, When are the idiots in the UK Government ever going to “grow up”?
5, What are the UK Gov going to do when “Off-Device Encryption” replaces E2EE as the next user security step?

But importantly when are politicians and those unelected in Government going to stop lying to the citizens and voters about “back doors” etc being for the protection of the public, rather than the actual truth that they are being used to spy on innocent people without legal oversight.

Clive Robinson • February 22, 2025 4:51 PM

Sometimes a little humour helps…

The news has been fairly dire this past month and honestly I can not see it getting any better for quite a while.

So anything to raise a smile is needed, which is why this caught my eye,

“California goes ape with bill to crown Bigfoot official state cryptid”

I thought “have I been asleep for a month and now it’s the 1st of April…

https://www.theregister.com/2025/02/22/bigfoot_california_state_cryptid/

Any way it raised my curiosity, and I know some strange things go on down there but “Sasquatch” really?

I guess it has the advantage of not ever going to “go extinct”.

But then I saw the sub heading, and “Smile much I did”

“Beast remains as mythical as the return on AI investment”

You know something is “doomed in the public eye” when people start using it as the butt of a joke.

In case “we” are all wrong… let me know how your investments in current AI LLM and ML systems are going, any better than NFT’s and Web3.0?

Bob Paddock • February 23, 2025 7:34 PM

@Clive Robinson the Chaotic/Strange Attractor stuff is Interesting. I’ll have to figure out how to plot some of my data that way. There is some speculation that the Brain uses such ‘Neural Noise’ in its processing. I can provide references if anyone is interested. Does the Brain ‘Attract’ the next frame of Reality? Different blog…

Clive, what kind of Random Event Generator/Random Number Generator would you design with today’s technology?

Clive Robinson • February 24, 2025 12:39 AM

Bob Paddock, ALL,

With regards,

“Clive, what kind of Random Event Generator/Random Number Generator would you design with today’s technology?”

The “trite answer” is a TRNG that is fully observable and based on fundamental physical properties from which bits can be extracted that are fully independently, without bias, and can not be interfered with.

At our current level of understanding that implies a quantum device of some kind. But as most such physical devices are subject to “interference” along the “chain” from quantum effect to logical bits. There are still many issues to be addressed for them to be considered secure.

Oh and another issue is “safety”. Consider you need a very high number of random bits in any given period of time. Lets assume you use a radioactive source… The implication is you need a very large quantity of detectable radioactive elements emitted from the source and for various reasons you want to keep it’s volume low. This means that the amount of EM energy produced by the source is very high from a quite small surface… Which implies a lot of radiation be it ionising or not things are “going to be hot”.

Clive Robinson • February 27, 2025 1:55 AM

@ Bruce, ALL,

As most readers here will know their is an ongoing argument about “intelligence” with people making wild claims about boxes of logic circuits having what the consider General Intelligence” then as famously playing the two foot shuffle and apparently retracting their claims…

This is not new by any means many years ago there were claims about Conway’s game of life and finding the right sequence to promote what appeared as “intelligent behaviour”

Few realise that the argument goes back way before that with Charles Darwin making claims about plants having intelligence.

The reality is that we have no clue about what constitutes intelligence for a couple of reasons. Firstly we have no definition for intelligence other than the old,

“You’ll know it when you see it.”

And secondly that nonsense about,

“There is something special about humans.”

It’s like racism but we don’t really want to admit it. In part because it brings into question the fundamentals of law from “Individual Rights” for the notion of “property” and “ownership” over “Social Responsibility”. Which is what “authoritarianism” via “Might is Right” and the “King Game” insists must be “the one true way” even when provably false and destructive.

The fact is if you can keep an open and reasoned outlook, you discover even the simplest of life forms exhibit behaviours that are not,

“Random, chaotic, or fully deterministic.”

But exhibit features of all of them, along with the ability to show changes with repetition to stimuli over time, that equates to the ability to remember and learn.

Something Conway’s “game of life” did not.

In effect the two types of “evolution” by themselves do not account for “intelligence” there is something missing…

This is discussed in,

https://www.noemamag.com/a-radical-new-proposal-for-how-mind-emerges-from-matter/

And it boils down to “communication”.

What it does not cover when mentioning neurons, is the fact an individual neuron is actually in effect a “pipe” from a sensor to the brain. Not to dissimilar to a length of insulated wire from a sensor to a controller in a machine[1]. It misses out on the synapses where a lot of interesting things actually happen, not least is communications across synapses are to put it politely are “complex” with in part their behaviour modulated by the immediate environment around them.

Maybe some one should “revisit” Conway’s Game of Life and add “communications” between the automata and a little memory and agency to control the environment around them…

[1] Such controllers that within living memory were made of “Ladder Logic” made of rotary wafer switches, electromechanical relays, resistors and diodes implementing basic logic circuits and fixed memory… Not what most would consider to be fundamental components of intelligence. Yet performing what some considered the basics of intelligent behaviour of having outputs that respond and vary to inputs in ways that are both complex and variable.

Leave a comment

Blog moderation policy

Login

Allowed HTML <a href="URL"> • <em> <cite> <i> • <strong> <b> • <sub> <sup> • <ul> <ol> <li> • <blockquote> <pre> Markdown Extra syntax via https://michelf.ca/projects/php-markdown/extra/

Sidebar photo of Bruce Schneier by Joe MacInnis.