Schneier on Security

Clive Robinson • November 28, 2020 5:37 AM

@ MarkH,

… and simple grifters whose only focus is extracting cash.

That appears to be the real reason behind the court cases.

After all how much do you think “Dear simple Sidney” is billing for her work?

Especially as others have found it does not appear to be “Dear simple Sidney’s” work… But somebody else, a clear “Cut-n-Paste Queen”, with a missing “Z” and a “chardonnay name”[1]. I wonder if she looks both ways when she crosses the road, at her time of life? As others have noted sometimes a bus can be hard to see when it’s heading your way…

[1] A lovely expression from England before the turn of the century when “Friends” was still new and Starbucks was just a spin off of Melville’s Moby Dick. It implied some one who who might also now be called a WAG that drank copious quantaties of cheep white wine and changed their names to sound what they thought was more trendy. Usually this involved putting an “i” on an abbreviated form of their first name. As one caustic observer at the time noted “The only time “I” comes last in their lives”.

Clive Robinson • November 28, 2020 11:15 AM

@ Czerno,

Doubts over the Oxford vaccine-candidate

It’s a poorly carried out smear.

The Oxford vaccine passes the WHO efficacy standards at both the 50billion and 25billion doses.

As for the half dose plus full dose, this happened to only one small group and it showed efficacy around 90%. Whilst the other groups on two full doses showed efficacy around 63%.

Oxford And Astra are now recruiting more people to do further half+full and it’s said potentially half+half groups in a much larger size.

Because for obvious reasons, if they can find an as effective or better dosing with less vaccine, then more people can be vaccinated for the same amount of vaccine produced. That means getting it into more arms world wide faster, thus hopefully bringing the end of this pandemic faster.

The two US modified RNA (mRNA) vaccines are at best going to only be available to some first world nations. Not only are they nearly ten times the price for the vaccine, the required “chill chain” at -70C is dangerous in of it’s self and requires all sorts of expensive technology that has not yet been developed. The Oxford vaccine however can survive in an ordinary fridge for months, to get it to peoples arms requires only “cool bags snd ice blocks” of the sort many use for picnics, thus is not just safe and inrxpensive, they are readily available now and in most parts of the world such vaccine “chill chains” are already in place.

As for the author going on about the now fully discredited HCQ even with zinc, it shows the author is either significsntly biased for some reason or not upto doing actuall basic research.

I strongly feel that the authorvis biased, that is they are quite deliberatly pushing a political or other such as anti-vax line for some reason they have not declared. Thus rather than “go with the science” they are going for an ad hominem attack against an individual.

But is it just anti-vax, I suspect not as they have not mentioned the mRNA vaccines at all. Thus I suspect it must be specifically a US / Politics related bias or even the gruby business of profits for US big phama…

Oxford / Astra unlike the US companies are not making a profit and are encoraging others to make the vaccine localy thus the current price which is about 1/10th of the US companies will drop further. Also when you look at costs so far the two US companies have consumed billions of US tax payer dollars the Oxford / Astra has consumed only a small fraction of those sums.

The US companies have been trying for years to get mRNA therapy to work, and they hope to turn it into a highly patented cancer cure. As a treatment mRNA is not only new it’s untried and quite radical, and there is absolutly no data on it’s longterm safety nore will there be for a decade or more. The Oxford / Astra vaccine whilst new is based on well tried and tested principles and is a modification to a chimpanzee cold virus.

I’ve already said I will not have the mRNA vaccine, simply due to the fact that I realy do not trust the mechanism behind which it works for safety. However if my Dr sent me a letter tommorow asking me to come in for the Oxford vaccine I’d roll my shirt up before putting my coat on.

Oh one thing else to consider. During the mRNA trials some in the test group did get COVID and did end up in hospital. With the Oxford vaccine whilst some in the test group did get COVID NONE went into hospital. This tends to suggest that whilst the vaccine might be less effective than the mRNA vaccines at stopping infection, it is not an “all or nothing” that is those that got COVID probably got it a lot less severely than the mRNA failures…

But we are still a month or more away from getting a needle in our arms, whilst the efficacy stage of the phase three testing is over, there is still the two month additional safety trials to complete. Thus realistically most of us are not going to get the shot in the arm befor spring next year.

Which brings up the issue of getting through “flu season” information we have suggests that having both flu and COVID together is realy a very bad idea. So as this years flu shot is a quadriplex it might be worth getting a flu shot as soon as possible. Certainly in the US where flu is alrrady on the rise, earlier and faster than expected.

Clive Robinson • November 29, 2020 8:06 AM

@ MarkH, xcv, ALL,

I would expect the gradual weakening of the source to lower its output rate, but not to introduce bias.

As the source weakens the number of detections in a unit / period of time goes down and thus the average time between them goes up. That is a form of bias as the % increase in time period after time period remains the same irrespective of how long or short you decide to make the time periods.

So for your cycling counter the average count goes up by the same percentage time period after time period. When the counter “wraps around” you effectively get a sawtooth error function the frequency of which is directly related to the change caused by the % difference.

Such waveforms can “hetrodyne” with other sampling times and the lowest common frequency difference appear super imposed on the output as phase modulation.

If you have an oscilloscope and two medium or high frequency square wave oscillators, with one going to the D input and the other going to the CLK input of a D-type latch[1]. The Q outputs look chaotic or random close in, in fact they are anything but as you will see when you open out the time base far enough. Put the Q output through a lowpass filter or leaky integrator[2] and you get out a very pure sinewave at the lowest sub harmonic difference frequency. That is that chaotic looking signal is actually a high resolution Pulse Width Modulation (PWM) signal of a sinewave at the lowest difference frequency.

Thus even nano second differences in the timing counter chain fold back in the sampling domain to the lowest frequency quite sufficiently to significantly phase modulate the output. Thus your supposed “random signal” is phase modulated by that sawtooth caused by the % change per time period even if it is very very small.

The problem with phase modulation is most people can not see it for what it is when decimated by sampling. They tend to think of it as “random jitter” caused by noise thus potential entropy, which it is most certainly not.

It especially rears it’s ugly head in the sampling of “ring oscilators” that are standard in many CPU “TRNGs”. Because the phase modulation dominates the output, manufactures hide it from view by “sprinkling magic pixie dust thinking” on the problem by using crypto grade algorithms to obscure it. The result is if you do not know the “key” you will see what you think is high grade entropy. If you do have the key, then you see the PWM waveform with just a tiny bit of entropy. Which means you can phase synchronise with the wave form and turn what appears to be 256 bit of entropy into maybe 7 or 8 bits which turns an impossible brute force search into an easy brute force search…

I’ve mentioned this a few times in the past on this blog before, but I guess most people don’t understand the implications thus trnd to put it “out of sight” then the inevitable “out of sight out of mind” follows it’s usual path.

[1] If you don’t have the electronics and test kit to do this you can do it with a pencil and regular squared graph paper, or write a computer program to do the donkey work for you. If you are going to write a program I would suggest you use two phase accumulator Numericaly Controled Oscillators(NCO) as they are trivial to write and can give you a frequency resolution of very fine degree the more bits in length the phase accumulator has the finer the frequency step can be,

https://en.m.wikipedia.org/wiki/Numerically-controlled_oscillator

[2] An integrator is a mathmatical construct and can be seen as a “counter” counting the area under the curve. A leaky integrator removes certain types of bias and alows you to get an RMS zero for the waveform. You can use an up/down counter or similar to recover the waveform and then feed it to a D2A converter. Similar techniques work just as well in software.

Clive Robinson • November 29, 2020 5:53 PM

@ Cassandra,

I started reading the paper and up to section 4 it says exactly the same as I’ve said on this blog one way or another several times over the years.

However the first sentance of section 4 brought me to a screaching halt and various other statments in section 4 made me shake my head sadly. I don’t think they have considered various technology issues. Not least of which is a CMOS inverter is provably an analoge amplifier with sufficient gain to be wired up like an inverting op-amp. This has been well known for half a century and if you look in the first edition Motorola Seniconductors 1973 Mc MOS Handbook Chapter 3A.1 it gives an indepth descryption and theory of the operation of an inverter treating it as what it is two complementary fets in an open loop amplifier configuration. Chapter 8B through 8D explain how to exploit the analog behaviour not just as AC coupled amplifiers but multistage gain blocks, noise generators, and oscillators inclusing ring oscillators… Something tells me the authors have not read it or a more modern version.

I shall carry on reading the paper but I’m not hopefull. They do not appear to have addressed a number of issues that I know cause problems. Not just what I mentioned with the fold back when using a D-Type latch but the issue of “injection locked oscillators”[1] where the current spikes caused by the output stage of a CMOS gate “passing through the output linear zone” gets back to the input circuit of other gates and effects them as they are “passing through the input linear zone” thus causing gates to start to fall into “lock step”. Normally with “clocked logic” this does not cause a problem but can increase metastability issues in latches. But is obviously very much a problem when you are looking at jitter on ring circuit edges as they start to correlate with each other.

There are several other things that concern me like their assumptions with XOR gates and their transition characteristics…

I will go through it ib more depth and get back to you.

[1] The fact that oscilators fall into lockstep has been known for centuries where pendulums were observed to do it and investigated by invebtor of the pendulum clock Christian Huygens who wrote a letter to London’s Royal Society about it in 1665.

Clive Robinson • December 1, 2020 6:39 AM

@ SpaceLifeForm,

when you need to confirm you’re not a robot

Thank you for that I’m in need of a laugh or three…

Not feeling particularly “Compost ment-hay” at the moment, the broken foot is playing up again after going out to get food in, and it feels like I’ve a cold comming on in sympathy with the change in weather (add “cold and wet” to the usuall British weather of “damp and grey”, and you thought it was warm beer and bunions that made us miserable, nope it’s broken bones that act like “seawead”[2] and ache when it’s going to rain, so they ache most of the time 😉

Which would normally be managable by meeting friends up for an endless tea/coffee and meal with a chat and joke swap at our local “spoons”[1].

Of course what we should do as “security experts” is not laugh but nod sagely and point out that the two main offenders are Cloudflare and Google and neither should in any way be trusted. Then start shaking our heads sadly and muter into our beards about rampant AI and how it’s claiming we are supporters of political party X because we clicked on only Y boxes and went left to right not right to left etc etc.

Which of course saves us from COVID better than a lockdown because every one avoids us all the time as being “grumpy miserable old farts” 😉

But speaking of spoons and lockdown it’s not been possible due to a month long “lockdown”…

But to show the idiocy behind thinking in the powers that be… The lockdown ends this comming Wednesday so people can start Winter Solstice celebration shopping (oh the ecomomics of it, who can not but shead a tear for a lobbyist in fear of loosing their pay packet). Then… Many other sensible anti-pandemic-spread measures on travel and going to meet people etc are going to be lifted. Which will probably mean that by the time Rabbi Burn’s night / Australia Day[3] 25/26 Jan gets here we will be in another lockdown. But… based on their apparent current thinking that will get lifted for Valentine’s day and it’s weekend…

[1] Spoon’s is a sortened name for Weatherspoons’ which is a chain of pubs with reasonable beers at sensible prices and quite nice meals for the price you pay. Oh and just remember “spoons” can be said not just when you are sober or some what imbibing merry, unlike “weathers'” where those silent “aitches” and “s apostrophes” always trip your tongue.

[2] It was only fairly recently that medical science caught up with the reality that most knew by experience. When you break a bone and it heals it often does not do it properly and this makes it sensitive to slow changes in preasure such as you get with weather fronts. Much as some peoples filled teeth do, oh and nearly every one’s ears do to faster changes in preasure.

[3] Both celebrations arise from just 29years appart the first is Scottish Poet’s Birthday in 1759, the Second the supposed first landing of people on the Australian Continent in 1788. Thus Rabbi would have lived to see the first of many of his country men sent south for ever by the English. Thus for some a day of celebration followed by a day of reflection.

Clive Robinson • December 1, 2020 8:00 AM

@ MarkH,

Anyway, I hope your “pedal” barometer stops working and before long you’ll need to look at “the glass” to see how pressure is fluctuating.

I suspect that it will join all the others that are the “badges of honour” of a “misspent youth” of contact sports and wearing the green.

What they do not tell you in an army recruiting office is why if you are lucky enough to survive, as an NCO why you get “early retirement”. It’s because all those training and combat injuries you get not from actual fighting accumulate and slowely turn on you. So when you get towards 45 you are effectively just on the point of being on your last legs / feet / spine / other body part.

I see film clips of young soldiers “yomping” with packs well over a hundred pounds on their back, more on their webbing and in their sagging pockets. So unbalanced they are barely able to stand, yet some “red tab” staff officer in an air conditioned truck or building expects them to maintain a 5mile/hour speed to some target area and deploy potentially under fire…

I just think about how much they are “borrowing from their future” and how bad “the pay back will be”, a few faded medal ribbons are not much reward for what they will probably suffer.

Clive Robinson • December 1, 2020 3:19 PM

@ MarkH, Casandra, SpaceLifeForm, Winter, xcv, et al,

“If any reader believes that I’ve gotten the math or physics wrong here, it would be most generous of you to identify the error, and cite some reference we could all consult.”

This example should need no refrences as most get taught it in science at school.

“Brownian motion”

The path of the “individual” particles in the “working fluid” are considered currently as nondetermanistic. But the over all average behaviour easily predictable.

And the underlying cause is similar to that of isotope decay, in that energy is added to the particle in some way from the environment it is in.

In Brownian motion it is reversable and makes the particle vibrate more thus causes where possible an expansion in the working fluid (see Boyle’s Law and it’s implications to the statistical mechanics Ideal Gas Law). For obvious reasons isotope decay is not normaly reversable as both energy and mass are lost from the sample the particle is in.

Individually in both Brownian motion and isotope decay what individual particles do is currently beyond our ability to determine accurately.

However as I indicated and SpaceLifeForm pointed out more determanisticaly it is not the individial behaviours that are of interest but the “average” of such behaviours over time in a bounded volume or closed environment of the sample being tested or judged by measurment.

Whilst a Geiger-Müller(GM) tube can register some isotope decay it does not measure it all from the nominaly closed environment of an isotope sample and it can and does produce false readings from other sources in it’s proximity, including that super nova just down the road a few thousand light years back.

But have a think about how a Random Bit Generator(RBG) actually works. It is disinterested in singular isotope decay, because it is nothing more than a point in time that is effectively dimensionless. The RBG actually works with multiple decays giving two or more points in time that can be measured. That is some kind of time measurment is made either by counts per minute or time between counts from the GM tube (remember that a GM tube is bandwidth limited and has recovery time as well as other issues so the points it gives in time are actually not accurate but biased in some way).

As SpaceLifeForm pointed out succinctly the average effect of the counts changes very predictably with time. So predictably in fact scientists find it one of the most accurate ways to date closed environment objects.

Such predictability is almost the very definition of a “known bias” that some over optimistically think can be countered or negated in some manner. But there is a difference between knowing a bias is in a measurment system and removing the bias, and thereby lies the gap between theory and practice that hides a very large attack space, that may or may not matter depending on your application. For security it usually does matter to a very high degree (1 in 2^4096 for some Public Key crypto currently).

Which is why a bias of a very small amount on a 50:50 judgment normally goes unremarked in most measurment systems. But with each measurment judgment the bias accumulates thus after just 50 judgments a constant bias of just 1% would have probably effected the result. But which one if any of those 50 judgments do you change to correct the bias?

You can not actually work it out, the best you can do is take many readings and average them out and then apply a fractional correction factor. But that does not solve the problem in fact in some cases it makes the problem worse (law of small numbers applied to residues). But worse yet the number of judgments you have to average to gain an improvment in accuracy goes up alarmingly when you are looking for entropy at 2^32bits, and reaching impossible for 2^4096 bits for an 8K RSA Public Key.

But also all the time though you will still have bias you cannot correct on an RBG no matter how many judgments you make. Because the reality is you have just as many issues measuring the bias it’s self. Which makes the user of the RBG “time constrained”. Made worse by the fact few if any RBG’s people will use will have continuous test / evaluate / update loops running. In fact in the case of IC based RBG’s that is deliberately prevented for reasons the chip manufacturers like Intel refuse to explain.

But an attacker has an advantage over the RBG user on this score as they are not as time constrained. If the user takes more than one output from the RBG then an an attacker can average those outputs. Thus the attacker has the advantage of being able to average out an error function over a much longer period in time.

Whilst there are ways for an RBG designer to push the attack further out in time –look up “dithering” for one example– if this is done by either a deterministic or chaotic signal then an attacker can still average it out. Thus determin what it is, synchronize to it and strip it off. Most importantly the attacker can then wind it back in time, which if they have a ‘Collect it all’ policy in place means your past messages etc become vulnerable. Thus the signal you use to dither or other obscuring method also has to be ‘truely’ random.

Thus you get into an almost circular argument, and the only solution is to go with a “turtles all the way down” mwthod of using multiple independent true RBG’s and keeping your fingers crossed about certain types of measurment bias. Which brings us back to the issue of ‘independence’. That is how do you ensure your sources and the judgment systems attached to them are not in some way synchronized under any and all conditions?

That’s why it’s easy to design a “noddy TRNG” with electronics from scrap devices, harder to produce one for maths simulation etc, but very hard to design a TRNG of sufficient quality for todays high end crypto and communications needs, especially where high volume traffic thus high usage of good RBG distribution is to be expected.

Clive Robinson • December 2, 2020 1:18 AM

@ ifb,

There’s a pernicious problem of indebtedness…

Yes, and it’s one that some employers try to exploit occasionally viciously.

I might be unusual, but I was aware of the issue long before I had my first full time job. So as soon as I started working I built up savings that amounted to “six months drop dead and walk away money” or if you prefer “six months seed money” to start my own business. Even when I purchased my first property I maintained the “walk away money” to include the payments. Also I do not borrow money I save, the mortgage was the only loan I’ve ever had as an individual, and it was an investment.

One foolish managing director tried to exploit the “indebtedness” he assumed incorrectly I had (I was young and owned a house he assumed there would be leverage there I guess). I disabused him of that assumption rather rapidly and would have seen him in court. However the Chairman of the Company and I believe the majority shareholder at the time decided a messy trial was not what the company needed. I got a couple of months extra pay as a “bonus payment” (tax reasons). By the time he made that decision I had already found another job that was better paying and much less stressful, so it got added to the fund.

From thinking about what had happened I realised from that incident that a thoughtfull employee could reverse the employer employee relationship if needed you just needed your own leverage (the old “walk softly and carry a big stick” thinking).

Some time later I was working at a large very plesant organisation that unfortunately got taken over, and the new owners were rapidly developing bad employer employee relations. So I decided it was time to find another job which I fairly easily did. However the potential new employer wanted me to start as soon as possible which normally would not have been a problem. However they could not wait the three month notice period that had just been put in place by my soon to be old employer who was using the courts to try and enforce the period on people who were already jumping ship…

So I engineered a dispute and a leagl friend drew up a 14day notice of intention to start proceedings for “judicial review”. The employers legal team must have told them of what their chances were so I got “instant dismissal”. But they found that was not good idea either because they got another 14day letter for “breach of contract”. Their lawyers again must have advised them of their chances, and I got the three months wages plus an “undisclosed sum” and importantly a letter saying I’d been made redundant along with a signed legal agreement that redundancy was what was to be put in my work record and given to any future enquiry made and the real prize was keeping my Intellectual Property. It was not long thereafter that the organisation “imploded” so I was possibly the only one to walk away better off.

Before you get the wrong idea I was usually an ideal employee because I only worked where I was going to be happy to work that had interesting and often challenging work to do. It was the companies “going bad” for various reasons. Take the organisation that imploded where I got out with my IP, it was realy a very nice place to work before it got taken over and “New Managment” tried to make it hell as only “asset stripper Venture Crapatalists” have a habit of doing. As for the earlier job, well it was the managing director making promises not just to share holders but government agencies that supplied funding with claw back clauses that caused the issue. He had made promises neither he nor the other employees could deliver on. I delivered but the price was to high (moon shots usually are). So in return for getting his butt out of the hot seat, he had decided it had to be my fault he sat in it in the first place… Even though he was sitting sizzling when I started working there and it was only the fact he was in trouble that made him take my on. When people get that kind of mind fix the best place to be is somewhere else with a nice safety gap in between.

At the end of the day the leverage only arises from indebtedness, if the employee gets into that position. Whilst some people are so poorly paid they have to work hand to mouth that’s generaly not the case for the “aspirational classes” they spend till they are a month or more in debt on credit cards personal loans and the like when there is no real reason to. What they do not think about is that “month in debt” is costing them another “month in interest” every year. So in effect they are working 12months but only seeing the benifit of 10months. In the main they are kiding themselves they are being financially creative. The point is they are not investing in assets that usually appreciate in fiscal value like a house, but “life-style status”.

Way back in the early part of my career I got to know a consultant who had what are sometimes called “bankable skills” he was earning three times what I was for doing nearly the same job. We got chatting about what he did with the extra money. It actually wasen’t much. Yes he had a nicer house, but the bulk of the money went into a better cut of meat for sunday lunch, going out socialising in better places, better shoes, cloths and holidays and a nice car and a few other comforts in life like having “Harrods toilet paper/tissue”.

We kept in touch afterwards and both of us got to thinking about the conversation and as a result we both put money in a joint venture that was quickly paying dividends and building up. His wife did the admin and as the kids got older came to work full time and we soon had a couple of employees. Sadly he had a fatal road accident, I did not need the extra income but for his family it was all they had coming in, so I sold his children my shares for a pound rather than disolve the business. His wife made a go of it and the children are still running it today, and they now employ quite a few local people. They send me an Xmas card each year and I visit and give a “founders talk” to the apprentices they take on. So yes I’m more than happy that I did it and I hope the business goes on long after I’m gone. I guess from the satisfaction view point it’s one of the best investments I’ve made as it’s now given quite a few young people a future. I was starting to do similar with an old school friend, but sadly he died unexpectedly earlier this year.

Clive Robinson • December 2, 2020 8:14 AM

@ MarkH, SpaceLifeForm,

As I already wrote a couple of times, I’m trying to reason these questions through methodically step-by-step.

And you’ve started in the wrong place.

As @SpaceLifeForm and myself have told you looking at individual decays is not telling you anything about True Random Bit Generators (TRBGs) that use isotope decay or any other singular event.

Singular events give you a point in time which effectively you gain nothing by measuring no mater how accurately you try.

It’s not how TRBGs work, they measure “relative time” by using multiple events. There are two basic methods,

1, Count the events in a given time period.

2, Count the time between two or more events.

Thus talking about if a singular event can be determined either now or in the future and if it takes energy from the environment or not is not gaining you any insight as to the bias problems and where they come from. But remember that there is an issue you first have to understand which is the Quantum Zeno Effect or Turing Paradox, I’ll let you look it up but to whet your appetite simplisticly it says that a quantum state change can not happen at the point of measurment, thus if you rapidly measure a quantum effect you can in theory stop it’s transition… Whilst Turing was perhaps the first to talk about this big hole in quantum theory, it was not realy taken seriously untill this century when experimentation had reached the point it could start to be investigated in more interesting ways…

But to humour you a little about isotope decay,

1, One of the basic dictates of science is “For work to be done energy has to be expended”

2, Also neither energy/mass can be created from nothing after the big bang event had settled.

3, Energy/matter move from the organised to the disorganized state (entropy)

So far, as far as I’m aware, all practical tests including trigering an isotope to decay have confirmed the above.

Thus the idea of spontanious decay without energy input is going out on a limb rather more than a bit and taking a chain saw with you. Theory as they say is not the same as practice, in fact theory especially quantum theory is part of the “shut up and calculate” school. The results of which need to be tested for the mathmatical models to move from being elegant formulations to proven predictive tools. Repeatable real world experiments are in effect not just the “gold standard” but “the only standard”.

The argument for alpha decay to cross a potential barrier is quantum tunneling of the particles. Put simply quantum mechanics indicates, these small particles can, with a very small probability, tunnel from one side of a potential barrier to the other. The way these particles cross the potential barrier is interesting. The argument is the particle, in effect, borrows energy from its surroundings to cross the potential barrier. It then immediately gives the energy back by making the electrons reflected by the potential barrier more energetic than they would otherwise have been. Thus the three dictates mentioned above hold.

Is this actually what happens who knows, but “the accounting balances” the way described. Hence the point both @SpaceLifeForm and myself were making.

But that still leaves the question unanswered as to what triggered the tunneling event. Part of the argument is that alpha particles are at best only very loosely contained within the isotope nucleus and bounce around back and forth against the potential barrier around 10^21 times a second and even with the heaviest of istopes where the containment is very weak it could still take several billion years of this bouncing around to happen before an alpha particle tunnels it’s way out.

But it still does not explain what caused it to cross the tipping point to be able to tunnel. Is it an effect of the alpha particle or the barrier or possibly something we do not yet know about (Quantum mechanics is known to be incomplete still).

Any way enough for now.

Clive Robinson • December 3, 2020 4:40 AM

@ Cassandra,

Unfortunately, I have had to deal with a domestic embuggerance…

I’m sorry to hear that, hopefully it is both minor and resolvable. As for apologizing, that’s kind of you but not necessary we all get bumps in the road of life every so often. With luck we mostly avoid them but these days I appear to more than stub my toe on them, guess I need better glasses or something (not that I can find them[1]).

With regards the overly dramatically named[2] “Elitzur–Vaidman bomb tester”, it actually has some interesting properties over and above it’s elegance. Think of it as a “observation sensor” that is it detects the collapse of the superposition at the point of measurment, which has implications for examining the Turing Paradox as well as real world sensors.

Speaking of point of measurment sensors on a more down to earth note you mention the infamous Intel paper… What it describes has been described by others as the “Roulette Wheel” or “Waggon Wheel” TRNG as it is in reality a “stroboscopic sensor” and just like the film gate in old movies where the waggon wheels appear slow, stationary or turning backwards on the screen. Thus it likewise has problems, not that Intel invented this type of TRNG though the paper does kind of make you feel like they were implying it. It has been independintly invented by quite a few engineers over the years myself included. As far as I know back a decade and a half or more before the Intel paper in the very early 1980’s[3], I was the first to use a VCO as an analogue to digital converter in a TRNG but these ideas have a habit of “comming of age” thus occur to many people around the same time, and the use of a VCO as a A2D converter was not exactly new then it goes back into the relms of early telephone research.

You note Intel lable the parts in an over encompasing way, almost as though they were writing a first draft patent application. But as a resultvas you note with,

It is interesting to identify the important things not being said, otherwise one could get the impression that everything is hunky-dory.

Or as my father used to dryly observe,

“There is always dust under carpets, though some more than others”.

Such expansiveness by Intel makes for a very large carpet… An observation if I had voiced it back then would have caused “shock and awe” as Intel were to use somebody elses observation “everybodies darling” at the time. Now after more recent events in their public history I suspect that many would just give a wry smile.

But to answer your question of,

I don’t know if the same design continues to be used

The answer is both yes and no, the advantage of the VCO back four decades ago was it had “a lot of advantages”, but it’s big disadvantage was it was “an analog part” thus not realy suitable for integration in digital chips. Also thermal noise sources generate tiny signals and this can cause all sorts of powersupply and other noise issues and be susceptible to subtle feedback effects. So the idea of using “Ring Oscillators” to replace the thermal noise source and VCO came up. Back fourty years ago trying to use 7-10 74LS chips to make a suitable ring oscillator was way to much PCB real estate, slow and caused powersupply issues. You would not use the 4000 series CMOS chips because they were “oh so slow”. Now such ring oscillators are just a tiny tiny fraction of chip real estate so they get used in prefrence, especially as with the devices that make the inverters have upper cut off frequencies above 10GHz means they can “toot” at frequencies that with a few combined will churn out bits in the tens of billions a second… So “quantity not quality” is the game, hence the need for the “magic pixi dust” entropy pools and crypto to hide it all behind. As in the Wizard of Oz you are not supppsed to look behind the curtain… Oh and don’t mention the “pendulum problem” of using multiple ring oscillators having them sing harmoniously is not giving the discord they designers were hoping for.

But if you are not looking for buckets full of questionable bits of quite low “true entropy” but moderate quantities of quite high “true entropy” then a well designed and built “Roulette Wheel” TRBG will be a better bet.

Which brings us onto “what people are looking for”. As you note,

I think one of the problems with the discussion around radioactive decay is the misconception that it you look closely enough at something, you can find the root cause.

Conventional mechanics can only have “work to cross the tipping point” or as some call it the “entropy hump” or as bombs have been mentioned “going high order”. There are reasons for this. Firstly the tenents and axioms of classical physics demands it, secondly is “the need to be master of all you survey” which is in essence what most scientists are all about, though they call it the search for knowledge / understanding. It was after all the reason for the near schism a century ago when quantum mechanics poped it’s head up, and Einstein’s presumption on Gods life style.

You could say “the outcome was ordained” because actually humans can not accept a determanistic universe where everything is “preordained” because that would say at the macro level “there can be no accidents” and more importantly “there can be no criminals” because whilst crime will happen, those committing it were preordained and had no “free will” thus no choice therefore justice and punishment would be of no use to prevent crime.

So we need the concept of choice, free will, and therefore the universe can not be determanistic, thus we can not be masters of all we survey. Thus quantum mechanics or something similar has to exist to give us not just choice but a way to deal with what we can not know except through the veil of statistics and probability.

Speaking of accidents, I’m sorry to hear about your leg such things happen or are “sent to try us” depending on your view point, the sad fact is as my doctor used to point out before she retired is, what appears to bounce off of us when we are young tends to bite more with time.

I know it in my head that as entropy dictates, I am destined to get a little less robust with time, knowing this however does not realy help me, and I suspect many others, be accepting of it. In my head I’m still twenty something even though my mirror lies to me 😉

Thus “do as I say, but not as I do” and take care.

[1] I wear glasses because I have to which is a problem because I have very wide angle vision and can see around the edges which causes head aches and nausea or just “seasickness” in proportion to how much distortion I see. So I can not wear vari-focals and the glasses have to be large and as near rimless as possible. So not just hard to get at a reasonable price but when you think about it “hard to see” when put down even with good eyesight. As my eyesight has got worse over the years the need to take them off to do something and then put them on again to do something else has gone up as I need to switch from close to more distant vision the point at which I have to switch is about two hand widths from my nose which makes working at a desk darn akward as books and things I’m working on are in close, but all computer screens are distant… Anyway I usually remember where I put my glasses down but occasionaly something will cause a sudden change and I will forget where… So, imagine if you will a large aimiable bear with a perplexed look on a face that has been likened to a Klingon, that is thus as hirsute as Carl Marx wandering around myopicaly trying to find it’s glasses. It is almost the definition of the blind leading the hapless, and even occasionally entertaining for onlookers. But I also get the akward helpers… The ones who ask me what I am looking for, and I tell them, then they say something like “They are on the table right in front of you, cann’t you see them”… The implicit question actually being more a sardonic statment… To which the obvious reply of “No that’s why I’m looking for them” is akin to pouring oil on a smoldering fire… First it dampens the flame, then it fuels it vigorously, thus is unwise as a course of action. Just one of life’s little lessons for the unwary 😉

[2] I blaim two things for this over dramatic behaviour, firstly was that man and his “damn cat” which has done more for common literature than physics 😉 And secondly the fact that even at the best of times quantum mechanics is pages of dry maths and knarly symbols, that are less pleasing to the minds eye than snakes crawling around the bottom of a bottom less pit you just know you are going to fall into if you carry on studying it 😉

[3] It came about as a project I did in electronic music and I’d submitted the TRNG design off to a well known electronics magazine of the time for consideration for an article. They expressed thanks and some polite interest for a future article but it never made it to print. I think I still have the corespondence in an old filing cabinate in the garage (if mice have not enhanced the entropy effect by nesting in it).

Clive Robinson • December 3, 2020 3:51 PM

@ SpaceLifeForm,

Will not be easy, but with modern materials, could be better.

It could, but at the end of the day “who gets to build it” it would be nice if it could be all or as much as possible “local labour”. But I’m guesing in todays world it would be contracted out to some company that will import the labour from else where.

There is even the possibility it could be Chinese Labour at engineer and above level, the way things work these days…

But I guess the real question will be “utility” it won’t be cheap and astronomy rarely puts money in the bank. Thus the required ROI will be arguable, and there’s a lot of hands held out for research grants that do have a better chance of putting money in the bank.

The simple fact is it does not matter what your political stripe, purse strings are going to be pulled tight on science especially the long term sciences that are not expected to pay short term dividends.

Thus I would expect that sort of level of funding to go into “energy” related projects that might have military spin off. Which might include high energy densiry storage that can be as rapidly charged as a fuel tank can be.

Clive Robinson • December 4, 2020 5:05 AM

@ Winter,

Both?

I guess MAGA has never looked so honest[1]. Mind you speaking of Japan makes me think, work an N in there somehow and then it would mean “Whimsical Pictures”

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

[1] What has made me smile about it though is the “archaic” definition,

https://www.definder.net/MAGA/

But the page does have some “not suitable for work” “go to the naughty step” words.

Clive Robinson • December 4, 2020 2:14 PM

@ ALL,

It appears that the work in progress that is “The Hitchhikers Guide to Online Anonymity” has been recently updated,

https://anonymousplanet.github.io/thgtoa/guide.html

If people want to scan through and look for errors/omissions and let the author know via Github or just drop them a ThanQ, I think they might appreciate it.

Clive Robinson • December 5, 2020 5:24 AM

@ MarkH, SpaceLifeForm,

As I’ve mentioned the decay closely follows a (1/e)^n curve. Which as I’ve also noted a number of people have been trying to get into politicians heads with regards the growth in a pandemic, which is a pecentage per unit time change and has an inverse half life growth –time to double– and a half life decay that gives us the R value.

The original assumption about isotope decay was that as atoms are very widely spaced appart small particles would mainly pass through a sample. And as had been observed with some high energy particles some small percentage would colide with the atoms and be deflected or as observed bounce back towards the source.

The argument effectively went that if a sample of isotope atoms was in a stream of such particles that were time invarient in intensity their probability of being hit by the particles was small but likewise invariant with time.

But importantly that time invariant behaviour of atoms and stream of particles has a time varient result as an isotope atom takes it’s self out of the game by decaying. That is in each time period you would trigger a small percentage of the atoms to decay, leaving less atoms for the next time period which as the probability had not changed would result in the same percentage –not quantity– change in each successive time period. Which gives you an exponential decay curve.

So two time invariant effects combining to give a time varient result, with the assistance of a little uniform physical randomnes in just one of them.

The result of such thinking is very appealing and still is to some. Hence the reason some people still look for an external trigger. They reason, that the fact that none has been found so far, does not mean it does not exist, so does not deter them looking. Because they reason the fact that something has not been discovered does not mean it does not exist on the old “You can not prove a negative” argument.

Oh one fun implication if it did exist is that stream of particles would have not just a constant rate, they would also need to have a small random spacial distribution with the randomness being uniform or ideal, that is of uniform density by time/frequency which is what many call “White Noise” and is found in most other noise sources and Quantum Mechanics requires and both it and Classical physics both give us[1]…

So yes some people will search the physical universe for a classical physics solution that they believe must be true because so much points to it, whilst most others just sit there and calculate with their Quantum physics mathmatical models irrespective of personal belief.

And there is a funny side to this, Quantum Mechanics effectively gave the universe “free will” but… At the cost of strict determanism in the mathmatical models that requires randomness to make it all work.

The down side of this is the “preordained” argument that alowed man and therefore a god to be master of all they surveyed has been replaced, not with a god that plays dice, but just the dice of randomness themselves… The implication of this is of course evolution is true, thus randomness might be the constant from the impossibly small to the impossibly large.

I suspect Alan Turing had come to this sort of conclusion from various of the things he said –including the nature of spots and stripes on creatures– and did. Not least of which was argue that all computers need a true physical random generator.

[1] Oh another fun thought, you are I assume aware of what the Root Mean Square (RMS) is and what it effectively does? Have you thought about the result of successive applications of it to any signal including a white noise “random” signal? If you take the RMS of a sine wave not only do you halve it’s amplitude, importantly you double the frequency of the sinewave the excess energy becomes an infinite series of harmonics of reducing amplitude. Thus you end up with a an infinite frequency series of small amplitude that successively approaches a “white noise distribution”. As all waveforms can be shown to be made of sinewaves the same thing happens to white noise which is what we assume randomness gives us but as the frequencies both constructively and destructively combine the distribution becomes uniform. So from order we get the chaos of randomness irrespective of if we want it or not.

Clive Robinson • December 5, 2020 8:16 AM

@ FA,

The RMS value of a waveform is the square root of the mean of the square of the waveform. It’s just a number, not a new waveform.

Have a look at the output of any nonlinear square law circuit such as a bridge rectifier. Nearly everything in life follows a power law one way or another. Engineers spend much of their lives trying to work in the part of the curve that’s near linear.

sin(x)^k for increasing integer k converges to a series of narrow impulses of unit ampliture and spaced pi apart. If k is even they are all positive, if k is odd positive and negative pulses alternate.

Now take the second RMS of that including the issues of “sampling” that creates frequency fold over. Then keep going.

The world is part of a universe that is not only quantitized but most definitely not linear.