[Cryptography] Very best practice for RSA key generation

Jon Callas jon at callas.org
Thu Nov 7 17:10:43 EST 2019



> On Nov 6, 2019, at 3:31 PM, jamesd at echeque.com wrote:
> 
> One obvious solution is to have the permitted word list, and have some algorithm that picks the nearest word in the word list, and then displays the autocorrected passphrase.  (Which the user can retype if the "correction" picked the wrong word, as is notoriously apt to happen.)

Yeah, and to talk out of the other side of my mouth, you could always make them type it exactly. Certainly I'd code it up that way and then add in the error correction later.

> 
> Which requires a word list, which I don't have, and algorithm to pick the nearest word in that word list, which I don't have and do not particularly want to write.

Outsource that issue to someone else, then. There might be correction packages around or even OS functions you could use.

> 
> I would think that there should be a pile of open source word lists and a pile of such open source algorithms around somewhere, but I cannot immediately find them, and do not want to re-invent the wheel.

I found this:

http://www.paulnoll.com/Books/Clear-English/3000-words-order.html

Three thousand most common English words, listed in order, from newspaper statistics. Take the first 2048 of them, and then poof, you have an eleven-bit-workfactor/word list right for the taking. Or use them all, and then you have 11.55 bits per word.

	Jon




More information about the cryptography mailing list