Md5 Question

[Merlin33069]

A friend of mine an i were wondering the following

If there is really a one in 20 million chance of and two files having the same MD5 hash

in theory, would it not be possible to recreate a file from the MD5 hash itself?

I understand the time involved in doing this through the "rainbow tables" method, not to mention the processing power.

but with new processors coming out with 8 cores or so....

you see where this is going...

[jbennet]

in theory, would it not be possible to recreate a file from the MD5 hash itself?

No. Hashing cant do that.

[Merlin33069]

ay, still would be cool :P

[jbennet]

its not possible though

thats not what a hash is. A hash is a totally arbitrary thing, its meaningless without a context with which to apply it to.

if we are going to be really technical, an MD5Sum isnt a type of hash at all, its a checksum, but the same thinking applies

[thoughtcoder]

Um, yes, it is a hash. Is hash not a synonym of checksum?

[jbennet]

no hashes have a completely different meaning when used in the context of databases

http://en.wikipedia.org/wiki/Hash_function

The first few paragraphs of this explain it

[thoughtcoder]

The differences explained there are a pedantic wikipedian classification system. Hashes, checksums, and cryptographic hashes are all the same thing and are all designed for the same general purpose. The only difference between a "good hash function" and a "good checksum function" and such is that different traits are valued more intensely.