# generating keys by using only part of longer random number

*From*: Kim Schulz <kimusan@xxxxxxxxx>*Date*: Mon, 9 Jun 2008 23:22:37 -0700 (PDT)

Hi,

If I have a random number generator that e.g. generates 512bit long

keys but I only need 160bits of them. Would the quality of the

randomness then be compromised if I always just uses the first 160bits

of the key?

My feeling is that it should not make it less random if you look at it

from a statistically perspective.

Example using digits for simplicity (3 digit random string S, 3 digit

key K needed):

S=12345

K=123

getting K=123 happens for S=12300 to S=12399

This is 100 times in 100000 draws = once every 1000 draws.

if S was only 3 digits long it would get K=123 only once in 1000 draws

= statistically the same as above.

But is my statistically approach to this correct when working with

cryptography?

.

**Follow-Ups**:**Re: generating keys by using only part of longer random number***From:*Kristian Gjøsteen

- Prev by Date:
**Re: Algorithm For Ensuring p & q Sufficiently Large For n in RSA** - Next by Date:
**Re: generating keys by using only part of longer random number** - Previous by thread:
**Old times and new ciphers** - Next by thread:
**Re: generating keys by using only part of longer random number** - Index(es):