Show HN: Satoshi9000 analog BTC key generator (mechanical)
The Satoshi 9000 demo: https://youtu.be/bJiOia5PoGE
The key value proposition of the machine is that it generates analog randomness in the physical world and converts it into digital (1’s and 0’s) randomness. Seamlessly.
But it occurs to me that it may have other uses beyond crypto keys for your own use, such as: * Randomized clinical trials. Clinical trials need a high degree of transparency for ethical reasons; also, for legal reasons should it come to light after the trial has ended that patient selection and treatment selection was not random or in some way biased (say, by the researchers themselves). The machine described herein can provide that transparency to young and old patients, technical and non- technical. * Non-technical management. Many network engineers in need of security keys have bosses that are non-technical. Such managers might prefer security keys (and their generation) which are easier for them to understand. * Estate planning. Suppose members of a family were to inherit digital assets (such as Bitcoin, for example). Not all members of the family are technical and understand Bitcoin. However, each will still need to generate a secure Bitcoin key to receive their share of the inheritance. The machine described herein might help in that task because its source of randomness is more easily understood by laypeople and each can generate their own private key in private (in isolation with the machine). * Anywhere where the users have to have an intuitive understanding of how the randomness is being created; whether they are 5 years old, or 95 years old, and all ages in between.
I'm curious to know if any of the folks over at HN can think of other use cases?
95 comments
[ 0.34 ms ] story [ 182 ms ] threadI'm not sure what this would add over, for example, entropy derived from a hash of the image of a camera's thermal noise profile.
Those usually don't look and sound like they were made by Doc Brown.
Value proposition: The key value proposition of the machine is that it generates analog randomness in the physical world and converts it into digital (1’s and 0’s) randomness. Especially noteworthy is that it can do so in a visibly, and to a lesser extent audibly (the sound of an agitated coin or die), way that is easily recognized and understood by humans. Other ways of generating digital randomness do not have this characteristic and in some ways have to be considered opaque to the public at large in their method of generating randomness. It comes down to whether people trust their eyes more than a black-box and whether people want this characteristic when generating randomness they are going to use.
I venture to say that anyone, from 5 years old and upwards who saw the machine in operation would understand how it is generating randomness. Dice from prehistory have been used by humans to generate random outcomes, and from the first millennium BC, when coins arose, the same can be said of coins.
Consider a randomized clinical trial. You may have patients that are not technically sophisticated, but must be convinced that the randomized aspects of the trial are done in a way they understand and are willing to give their consent. The same can be said for lawyers.
"I'm not sure what this would add over, for example, entropy derived from a hash of the image of a camera's thermal noise profile." Do you think a 95 year old grandmother will understand the principles by which this type of randomness is created?
Mistrust the machine? Then simply don't use it. ("Don't trust them lying eyes!") What I can say in its favor is it's connected to nothing (air-gapped), you fully control every important aspect of the randomness (fully programmable). Don't like the coins you have? Simply take a quarter from your own pocket and put it in the shaker. Don't like the microcontroller provided, buy (for $4) your own and plug it in. Ditto for the other components. All sensors, motor etc. are commodity parts; replace them. I think this machine is more provably back-door free than any cryptographic machine out there. As I point out in the video, all they important parts used in the generation of randomness walk-away in the palm of your hand -- what I call "walk-away randomness" in the video -- and all that's left is a motor and some wires.
As to the bitrate. Yes, it is not a high bitrate machine, the bit rate of the machine is around 4-bits per minute (time length of tossing/shaking and vigorousness of shaking is wholly under the control of the user - can be longer per shake, faster or slower, or variable during the shake), so for a 256 bit key it takes around an hour. But remember, Bitcoin keys are forever (or the remaining lifetime of the Universe, whichever is shorter), so taking an hour to generate it is short in comparison to its useful lifetime.
I hope the detail, and some background, helps.
So what I typically do is print a warning at the top and bottom of the printout urging the user to transcribe the important parts using archival paper and pen as soon as they can.
Also, if you look at the video, you will see an "Archival Printer" port on the front of the control box because I’m developing a printer that prints the keys (plus QR codes) on metal so they last for decades and perhaps centuries. That may be useful in estate planning where the key may be locked away in a safe, or a lawyers safe, for generations. But transcribing to archival paper and pen is relatively permanent (decades) and is easy and seems to work well (lawyers like it).
“You can pop a lot of trouble in the pop o matic bubble”
Likely thousands of HNers viewed the link. 100+ upvoted. A couple of random accounts (individual(s)? bots?) leave similar comments.
And then somehow those comments are supposed to reflect a wider sentiment of some community or population?
When I showed the machine to my son, Nate, a mechanical engineer, he thought it looked like something from a 1950's sci-fi movie like "Forbidden Planet". Back then, plastics were high-tech and new, and with the acrylic domes, the Satoshi9000 would not look out of place on the set of that movie.
He suggested that every coffee table should have one!
Care to elaborate? Or link?
I mean, everything that is, is just displaced temporarily homogeneous complexity, allowable between the fluctuations of gradients of temperature, allowing the illusion of work to appear more than just energy dissipating into the either of expanding space-time, dragged by the irreconcilability idea of "gravity".
But that doesn't help bake an Apple pie from scratch, as Carl Sagan would put it.
* https://youtu.be/bJiOia5PoGE?si=IEhbNJk0C0-7_2Nj&t=229
* https://youtu.be/bJiOia5PoGE?si=3Se3lYFVAAkElx0w&t=245
This generalizes to a die of N sides. Roll it N times. If you don't get all N distinct results, restart. If you do, then take the first result as your final outcome.
(That may take a lot of trials for large N. It can be broken down by prime factorization, like roll 2-sided and 3-sided objects separately, and combine them for a d6 result.)
https://en.m.wikipedia.org/wiki/Randomness_extractor#Von_Neu...
Someone please (jump?) at the chance to explain this one to me.
(assume i failed 9th grade 3 times)
"The Von Neumann extractor can be shown to produce a uniform output even if the distribution of input bits is not uniform so long as each bit has the same probability of being "one"->[first] and there is no correlation between successive bits.[7]"
As long as the person doesn't favor which of the two bits they chose is "first", then it should appear as random.
But that is self-defeating, as if the person had the capability to unbiased-ly choose between two binaries, they wouldn't need the coin.
But since the only way to determine the variation from expectation is repeatedly increasing sample size, I don't see how doing it twice, and just taking encoding of the bits, then...
Is the magic in the XOR step? To eliminate the most obvious bias (1v5 coin), until all that could had been left was incidental? Then, always taking the first bit, to avoid the prior/a priori requisite of not having a fair coin/choosing between two options?
and it clicked. Rubber duck debugging, chain of thought, etc.
I will actually feel better now.
Say you have a biased coin. It lands heads 55% of the time (but you don't know that.) Then the probabilities are:
HH = (0.55 * 0.55) = 0.3025
TT = (0.45 * 0.45) = 0.2025
HT = (0.55 * 0.45) = 0.2475
TH = (0.45 * 0.55) = 0.2475
If you disregard the HH and TT results then the equal probabilities of HT and TH result in a perfect binary decider using a biased coin. You assign HT to one result and TH to the other.
When imagined, the first result is 99% Heads...until you finally flip a Tails.
We had to do this exact thing in 6th grade, and I picked proving 5%...fml.
I forgot that they are discrete pairs, not continuous (like my head cannon).
The XOR is the magic. Always has been.
Coins and dice and datums (solid objects with detectable outcomes) may, or may not have bias, it depends on how they were made and on manufacturing defects that resulted. But, at a minimum, such bias can oftentimes be side-stepped or bypassed.
Consider this argument from Johnny Von Neuman.
Suppose you have a single biased coin with these outcome probabilities:
A) Heads (1) 60% (Call this probability p.)
B) Tails (0) 40% (The probability of this outcome is q=(1-p), by definition.)
Now let us apply this algorithm to sequential tosses for this coin:
1) Toss the coin twice.
2) If you get heads followed by tails, return 1. (Say this outcome occurs with probability p’.)
3) If you get tails followed by heads, return 0. (The probability of this outcome is q’=(1-p’), by definition.)
4) Otherwise, ignore the outcome and go to step 1.
The bit stream that results is devoid of bias. Here’s why. The probabilities of obtaining (0 and 1) or (1 and 0) after two tosses of the coin are the same, namely p(1-p). On the other hand, if (1 and 1) or (0 and 0) are thrown, those outcomes are ignored and the algorithm loops around with probability 1 – 2p(1-p). So, the probability (p’) of getting a 1 using this algorithm after any sequential two tosses of the coin is p’ = p(1-p) + p’(1-2p(1-p)). The solution of which is p’=1/2, and since q’=(1-p’), then q’=1/2. A fair unbiased toss!
In fact, the example bias numbers given above don’t matter for the argument to hold (note that after solving for p’ it is independent of p). The outcome of the algorithm is a fair toss (in terms of the (0 and 1)-bit stream that results), regardless of the actual bias in the coin for a single toss. All the bias does is have an effect on the efficiency with which the bit stream is created, because each time we toss heads-heads or tails-tails we loop around and those two tosses are thrown away (lost). For an unbiased coin the algorithm is 50% efficient, but now has the guarantee of being unbiased. For a biased coin (or simply unknown bias) the algorithm is less than 50% efficient, but now has the guarantee of being unbiased.
This algorithm is trivial to implement for the Satoshi9000.
This really is a useful idea.
Only holds if no spooky effects change results based on last result. (like a magic die that counts upwards or a magic coin that flips T after H no matter what)
P(TH) = p(T)*p(H) = P(HT)
It's not even really "spooky" - all you need is a flipping apparatus that's biased towards an odd number of rotations, and so then THTH is more common than THHT and you get a bias towards repeating your last result.
P(H|N) != P(T|N)
And
P(H|N) != P(H|N-1) (and visa versa)
Means that
P(HT) = P(H|N-1, T|N) != P(TH)
I suspect that when the user is loading coins or dice in the machine, they would notice any dirt that was significant enough to look as though it might be a problem.
And oil deposits from your fingerprints I would imagine are so minuscule as to be insignificant in creating varying bias.
Even then, in both cases, you could wipe the objects with an alcohol swab before putting them into the shaker cups.
It could be argued, I suppose, that every micro-collision of the coin or die with the cup removes a few atoms, but I would suggest that its effect on the bias of the coin or die over time is again minuscule. Indeed, unmeasurable over a full sequence of cycles (128 for example) of the machine when generating a Bitcoin key.
But an interesting point. Keep 'em coming!
[0] https://csrc.nist.gov/projects/interoperable-randomness-beac...
[1] https://drand.love/
[2] https://blog.cloudflare.com/league-of-entropy/
That's a quite interesting idea. I will put more thought into that.
Thanks!
Presumably they're using ~Dual-EC DRBG~ some kind of quantum randomness generators these days.
https://en.wikipedia.org/wiki/SIGSALY
https://news.ycombinator.com/item?id=20138230
That's sure a hardcore way to run an analogue randomness generator.
https://en.wikipedia.org/wiki/Hardware_random_number_generat...
I think it would be a stretch to think you could pull a random person off the street, point to a wall of lava lamps, and ask "do you see the randomness, how does it work?" Whereas, I think if you pull a random person off the street, let them watch the Satoshi9000 do its thing, and ask "do you see the randomness, how does it work?" you might get an answer that makes sense.
That, in a nutshell, is the value proposition behind the Satoshi9000.
And, as you point out, given it generates randomness by tossing physical objects, it is naturally a low bit-rate machine.
I didn't think it wise for a public demo video to show everyone the private key!
Just like every aspect of the operation of the Satoshi9000, printer output is fully under the control of the user program. I simply put a "PAUSE(hit run to continue)" command between printing the key-pair properly, and printing the key-pair with the private key hashed out (the one visible in the demo video). The "PAUSE(hit run to continue)" appears in the "Log File/Debug" window while the program is paused.
The bit rate of the machine is around 4-bits per minute (time length of tossing/shaking is wholly under the control of the user - can be longer per shake), so for a 256 bit key it takes around an hour. But remember, Bitcoin keys are forever (or the remaining lifetime of the Universe, whichever is shorter), so taking an hour to generate it is short in comparison to its useful lifetime.
I hope that helps.
I suggest that any system like this has the output XOR'ed with another random source. If two random sources are XOR'ed together, then both need to be predictable for the output to be predictable.
Not fully actually:
https://plato.stanford.edu/entries/determinism-causal/#ClaMe...
https://en.wikipedia.org/wiki/Fair_coin#Fair_results_from_a_...
But that only requires you to run the machine with the same coins/dice in the shakers for two consecutive cycles. And to repeatedly do so into you have generated the desired length of 0/1 bit stream.
Bear in mind the machine is fully configurable/programmable. You always choose what goes into the shakers, how many cycles are run, for how long they shake, how vigorously they shake, and what are done with with the 1's and 0's that result.
Implementing Von Neuman's algorithm on the Satoshi9000 is trivial.
what is the point in having a source of randomness if you need to XOR with a random source? relatedly, if you have a source of randomness, (1) please share and (2) well, there's no real need to go down this particular rabbit hole at all, well, is there?
independent of all of that, you seem to be anthropomorphizing the XOR function a bit... sure, there are some contexts where "1"s "mean" something and "0"s don't (sparse coding? and yeah, that is some pretty generous contortion, but hey: we're all friends here, right?), but in the case of "randomness"* the whole point (presumably) is that predicting "1"s and predicting "0"s are both exactly the same thing: Sisyphean.
i'm not sure that the word "random" "means" any thing beyond "distribution that we cannot model". which is a fine definition, given that models are how we attack random number generation...
* mind you i challenge the reader to pray tell what "randomness" even really means in anything other than a pseudorandom context (aka used to justify the randomness of various algorithms). isn't it oddly instructive that we use something that would pass as a proxy for "random"ness as the basis for our official definition for the second? (Cesium-133). the "second" is no more real than it is random. random is defined via a threshold of non-randomness, and all that we value as discrete and integral in the world that exists beyond our minds (-- if it does, in fact, even exist (or even "exist")--) is a house built upon sand. well, worse than that: the universe "works" because Avogadro's Number is a hell of a lot closer to infinity than it is to zero, and that's good enough for me. log(N) < 100.
[0] http://gamesbyemail.com/news/diceomatic
[1] https://www.youtube.com/watch?v=7n8LNxGbZbs
The reason is simple. Humans are terrible sources of randomness. Especially true if money hangs on the outcome!
There are two principal components for bias of a coin or die toss/roll: 1) the coin or die itself (manufacturing defects, etc.), which if it exists is typically minuscule, and 2) the act of tossing or rolling by a human (a twist of the wrist, or a flick of the fingers), whose bias is enormous and which, as I say, is particularly pronounced if money hangs on the outcome.
The Satoshi9000 solves problem 2, the human element, by removing the human from the process altogether. Other than to press the "run" button.
I started my working life at age sixteen as a coal miner at the Deep Navigation Colliery in South Wales.
Today, building useful and interesting machines has a lot in common with coal mining. A lot of hard work, and the perpetual risk of being crushed to death by 1,000 feet of rock above you. (The last part is perhaps a bit of a stretch, but it oftentimes feels that way!)
Would probably need a large engine to power it as well, with careful control because the resisting force would vary along the machine cycles (this could be used as a side channel attack vector to figure out internal state from resisting force).
The control box was a convenience and made the process fully programmable by the user. Which makes the machine far more flexible and useful.
I call it analog randomness because that's what I expect from the real world. For thousands of years, humans have used coins and dice to generate uncertain outcomes. And the fact that they typically generate only one of N outcomes (N=2 for coins, N=6 for common dice) is why humans use them. It is also why the Satoshi9000 uses them, and because its a kind of randomness that humans have an intuitive recognition of.
Now I'm thinking of a random number generator that uses a mechanical calculator printer (but I don't think there are any hex-capable ones easily available) or a typewriter to write password suggestions. The mechanical part would be tricky, because the hammers require some force to be actuated (and I would find it criminal to destroy a vintage typewriter for that).