RE: strong passwords (was Radius/MS ISA stuff)

["Ben Nagy" <ben () iagu net>]

> -----Original Message-----
> From: firewall-wizards-admin () honor icsalabs com 
> [mailto:firewall-wizards-admin () honor icsalabs com] On Behalf 
> Of Paul Robertson
[...]
[This is Paul]
> > > IMO, strong passwords are dead- dictionaries are too good
> > > now,[...]
[This is me]
> > I can't buy that without being shown more numbers[...]
[Paul] 
> I don't have great numbers[...]
[Me]
> > I'm not good at this sort of stuff

This is sounding fantastic, isn't it? Firewall Wizards - where the gurus
hang out. ;)

> > but for the space 
> > required for the 
> > md5sums of typeable passwords at 12 characters I get 5.94e24 bytes, 
>
> I'm assuming that's 116 printable characters?

No, I just looked at my laptop keyboard and went "1,2,3, um...4....", so
it's only 94.
 
> > For time,with the 4.1e6 ops/second figure 
> > you quoted elsewhere for md5, I took a million processors 
> and [probably had my hand down my pants]

[lots of performance stuff snipped]

> In '94 the estimates for finding a collison in MD5 were 24 
> days for a $10M 
> custom-built machine.

I think that finding any MD5 collision is not a useful work comparison
to guessing a specific password. Also, we already know that the
collision thing (birthday attack) is the area of MD5 operation that
crypto geeks are most worried about.

[...]
> Until then, I'd appreciate any other insights people have.

Let's look at it upside down (I should have approached it this way from
the start).

For a completely random hex password it's a pure 4 bits of entropy per
byte.[1]

Completely random typeables comes out at 6.55 something bits for my 94
character keyboard. 

Let's say that order 2^64 is still "safe" for work attacks (that's an
arbitrary figure I Just Made Up. I get to do that because it's my
email.).

So, we need 16 random hex characters, or 10 random typeables.

The trouble is that memorable or, worse, dictionary passwords have
waaaaaaay less entropy than that. I've heard english language quoted as
~1.3b/b (so we need about 50 characters in our passphrase). Even
passwords that people _think_ are random "because they just made them up
at random" I'd guess would be under 4b/b.

So, basically, Paul was pretty much right at the start in saying that
strong passwords are "dead", because I'm prepared to bet heavily that
very few people select truly random passwords of that length in
practice. (Although I do routinely use md5sums of random things for VPN
shared secrets).

> Paul

Whee.

I should be less flippant, but, oh well.

[1] For those to whom this is confusing, I'll explain. There are 16 hex
characters, right? And, like, 16 is 2^4, right? So, for this one
character there are 16 possibilities, 2^4, ie "4 bits worth of entropy".
You're allowed to just add these numbers up as you add more characters,
because of mathy exponential goodness. There you go - now you can write
"information theory" on your resume.[2]

[2] I have no point here, I just like footnotes.

--
Ben Nagy
Delirious Sick Fool
Mb: TBA  PGP Key ID: 0x1A86E304 

_______________________________________________
firewall-wizards mailing list
firewall-wizards () honor icsalabs com
http://honor.icsalabs.com/mailman/listinfo/firewall-wizards