[Cryptography] let's kill md5sum!

[Zooko Wilcox-OHearn]

Folks:

Several people replied pointing out how sticky data integrity
algorithms and values are. I totally agree. I think the best shot we
have of stamping out MD5 needs to have the following properties:

1. It has to be faster than MD5. Maybe this is irrational of these
users, or maybe it isn't, but in any case I don't believe we're ever
going to get the world to upgrade from MD5 to a slower algorithm. My
evidence is that we've already tried this for the last 20 years with
SHA-1, SHA-2, and now SHA-3. Last time I checked, even NIST, the
sponsor of the SHA-3 competition, was still recommending MD5 for data
identification/forensics.

2. It has to be a "drop in" replacement.

3. There is only one successor. No alternatives, no optional features.
And it has to be spelled the same way on all systems. System
administrators and programmers who don't know anything about anything
need to be able to say to one another: "b2sum is the new successor to
md5sum.", and that's that.

Regards,

Zooko Wilcox-O'Hearn

Founder, CEO, and Customer Support Rep
https://LeastAuthority.com — Freedom matters.