The reason you can't have nice numbers and other considerations.

asciilifeform Entropy generator now wired to high-speed sampler, about to do some classical tests. In other news, adding an (unpopulated) SOIC8 pad to the board, for SPI RAM, for if/when we roll in McEliece.
mircea_popescu McEliece ?

asciilifeform The lattice-based pubkey algo
mircea_popescu A a. Ya.

asciilifeform That laughs at 'quantum', etc. With 64k keys.
mircea_popescu Well, laughs at quantum as far as we currently know. Imo the biggest threat of quantum isn't that it runs our current algos faster,

asciilifeform Mostly Shor's.
mircea_popescu but that it might unlock new algo paradigms. Who's to know, you know ? It may come up with algos we don't even comprehend, but we can prove (on quantum) that are correct.

asciilifeform By the same argument, someone could always come up with a cruncher based on fermenting goat shit that makes P = NP. Not impossible to rule out.
mircea_popescu Exactly. You'll grant me this sort of weird is more likely once we allow quantum in the first place, which quantum is nonsense otherwise.

asciilifeform If P != NP, there are some hard thermodynamic limits on what can be diddled how fast.
mircea_popescu Yeah. I think that's the case, TBH.

asciilifeform Not exactly an uncommon view. If P == NP, time for onetimepad.
mircea_popescu Not even. If "quantum" is shown to work why wouldn't it break onetime pads ?

asciilifeform Think about it for a minute.
mircea_popescu I am.

asciilifeform The whole concept of 'breaking' onetimepad is nonsense.
mircea_popescu Why ?

asciilifeform When key is same length as message.
mircea_popescu So ?

asciilifeform There is no way to distinguish a 'correct' key from any other.
mircea_popescu Maybe the key and the message are magically (oops I mean quantumically) entangled in some magical way you don't know about.

asciilifeform Then you're not breaking an algo, you're performing successful voodoo.
mircea_popescu Which is what quantum computing IS.

asciilifeform Might as well blast the enemy with a telepathic death ray.
mircea_popescu Exactly. My point is basically that on the current understanding we can't logically speculate what will resist once "quantum computing" appears. Only AFTER it has appeared and we understand how it works can we cross that bridge. You see ?

asciilifeform Yeah mebbe Asimov's 'chronoscope.' Enemy peeks into the past of your living room and snarfs out the key.
mircea_popescu Right-o. Why not. "Quantum".

asciilifeform Once somebody coughs up the magic runes, I'll learn magic. For now, I'll stick with mathematics.
mircea_popescu Exactly. I think I'll use that perfectly condensed form to explain my stance on qc from now on.

asciilifeform Ever read Feynman's "Lectures on Computation" ?
mircea_popescu Yes.

asciilifeform Beautiful little thing
mircea_popescu I'm a great Feynman fan.

asciilifeform I've fed it to otherwise nontechnical people with some success.
mircea_popescu I was gonna say, perhaps it could be said he's more of a "science for managers" sort. But still, he's very good at it.

asciilifeform The man dealt with bureaucrats for a living, and survived. That's talent.
mircea_popescu Yea.

asciilifeform Unshielded RNG coughed a little when shot in the face with cell phone. Otherwise quite stable.
mircea_popescu Define a little.

asciilifeform Pisses out a little periodic. I'd be surprised if it hadn't. 'Do not operate in working microwave oven.'
mircea_popescu Yeah. How far away was the cell ?

asciilifeform 1 cm.
mircea_popescu So what's that like 0.1W/cm2 ?

asciilifeform Approx. right.
mircea_popescu So what will we say, do not use in em storms delivering in excess of 1kW/m^2 ?

asciilifeform Nah, we're shielding it, remember?
mircea_popescu But even so.

asciilifeform With shield, anything that bumps the RNG should also scramble the digital half and render the unit inoperable.
mircea_popescu Listen, EVEN IF the unit would rather melt than let itself be influenced, we gotta have a value. We're pros over here. 1kW/m^2.

asciilifeform From my preliminary test, the result of shooting rng in the face isn't a direct pickup of the signal, but induced self-oscillation in the amp. So, eminently detectable by the entropy tester.
mircea_popescu Ah. That's nice. We still thinking shield be copper ?

asciilifeform Copper ought to work; commercial shield plate (like the one found on the gsm module from p#2) also ought to work. I intend to compare them. Plate looks much nicer.