Large Primes

[Peter Fairbrother]

> Adrian Midgley wrote:

> How many bits should the key for a substantial narrative or summary piece of
> a medical record be, in order to reasonably hope it will not be read within
> the lifetime of the patient or the doctor concerned?
> 
> I know, string: length of... but I'm reminded of the passage in Neal
> Stephenson's novel Cryptonomicon where Randy is asked to generate a new 4096
> bit key for what is a thinly disguised GPG with the avowed aim of maintaining
> a secret communication "as long as men have the capacity to do evil".
> 
> 

Hmmm, a 4096-bit symmetric key _might_ work that long. I doubt a 4096-bit
assymetric key would.


If quantum computers are introduced you would need about 384-512 bits for a
symmetric cypher, according to current thinking, although that might be
overkill. In 100 years large QC's will likely either be working or proved
impossible (or the world as we know it will have ended). Use 512 bits if you
can, and if it doesn't cost too much.

Better, use as many key bits as you can afford. Then use two completely
different ciphers that don't interact, with different keys. There are ways
of getting encrypted information that don't involve keylength, but choosing
too short a length or a duff cipher is going to make you feel silly later
on.

Forget assymetric cyphers, if QC's come in they're gone. This means not
using them for anything but short-term signatures/authentication, not key
exchange.

An otp is probably good for 100 years, but impractical.

Just my opinion, and Brian's point about something completely unknown
turning up is very relevant. I for one can't predict the future! We're all
guessing here.

Another possibility is that lifetimes may be extended to well beyond 100
years in 100 tears time....

-- Peter