Pseudorandom number generator

One, it was done back in 97, 98.

I don't know that anybody's actually audited that that's, you know, PRNG.

and, and, and decided that it is absolutely like it's valid random numbers or random number generators and, and number I'm somewhat, I don't know, whatever.

Sorry.

Claude tried a different angle, anticipating how the software that generated the curves must have worked.

Claude guessed the software used a pseudo-random number generator," a PRNG that we've talked about ad nauseam on the podcast in years past, "that shipped with 1990s C libraries.

These old PRNGs are weak enough that their output can be recognized.

Recognizing that output would reveal which PRNG was used and possibly allow recovery of its initialization state." Because just to digress for a second, you, you start a PRNG with a certain state, and it is a very simple algorithm.

Typically, it's you, you add a number and you multiply a, a different number and you throw away the overflow, keeping just the lower bits.

And that follows a very recognizable pattern based on what those two numbers are.

But it also starts at something, and that's the or- the, the original seed, the, the, its initialization state.

So then they said, "Claude, having been asked where did these nu- where, where weird-looking numbers come from, having no idea whatsoever, says, 'Well, probably, how about maybe it was a, a PRNG back from then, like the 1990s, and people were using C.

... or you do anything like that, you're always gonna get something that we completely control.

Now, a lot of software will read dev random and time, and then they'll like mix the bits, and then they'll initialize like a PRNG, for example, to be able to have faster random number generation, which most runtimes will do.

When you're running in Antithesis, we actually recommend that you don't do this.

So if you do use a PRNG in your software, you actually replace the PRNG with direct reads to system randomness.

And there's a reason for that, and that reason is the second ingredient of DST.

So the first ingredient is determinism, so the ability to be deterministic, so the same series of events occur.

The second ingredient of DST is simulation.

And the reason why we want to be able to control randomness at every moment, so you not having a PRNG that's preceded, that has this, a particular series of events that's gonna happen, but instead you're always asking us for randomness, is that in DST, we want to be able to branch the simulation.

Because of this, they are extraordinarily bad at creating true randomness.

To simulate chaos, they actually need external help, and this is the difference between a pseudo-random number generator, a PRNG, and a true random number generator, or TRNG.

We covered the ColdCard vulnerability in a recent episode, and that bug was directly tied to this exact distinction, wasn't it?

It was.

But what about the cryptographically secure ones?

Right.

A cryptographically secure PRNG makes its outputs computationally infeasible to predict without knowing its internal state.

But, and this is key, it cannot create additional entropy out of thin air.

Okay, okay.

PRNG was leaking its output.

Wow.

Okay.

Now, one of the tests that he did was to, uh, hook it up to the plant, but then he, he had, uh, shrimp, and live shrimp, uh, in a bowl and water.

A- and he r- had a release mechanism that was on a, uh, they call it a pseudo, pseudorandom number generator, a PRNG.

And, uh, these, this random number generator, when it would w- uh, it would dictate when it, when the release was gonna happen, and so nobody knew the exact moment.

And he closed the door to his office.

Um,

That is interesting to think about because I wrote an implementation of PRNG last week, which sounds like a very difficult and challenging task, but really what that means is that I read an implementation that already exists and I wrote it over again in a different language.

that's programming that is like a significant chunk of programming yeah and it's just like you know oh it's literally just doing stupid shit to numbers so that you get a sequence of numbers that kind of vaguely looks like it's random even though it absolutely is not

is that for liar is

And then we can't ever have true random numbers, so that's why they're pseudo random.

So we need a pseudo random number generator, a PRNG.

So how do we get that? And so we said that, well, there are some of these that have been built for us, like inside Excel, RAND.

They can do that for us.

We've gone through the so-called correlation links And so now I don't see that so much, but I just want to highlight that, that if you happen to do, you know, if you were to search for a runs test, make sure that you're learning about the right test because the test we deal is the runs above and below the mean, but there are other runs tests.

One of them is the up and down and they're both very common, but I just think this is the easier one to use.

Now, if you go to this little PRNG widget that I built for you, I also put in what's called a lattice plot or a Poincaré plot or return map.

And so it turns out if you plot the numbers against a previous number against the next number, and you look at that scatter plot for really bad random number generators, you can eventually kind of rotate this and start seeing a lattice form.