Re: Why unhashing is not possible?

Unruh wrote:


For a limited set of inputs, this is very easy.

Yes. then it is not a hash. It may be an encryption, or a translation.


A hash is a function of A^* -> B^m for a fixed value of m. Nothing else.

A very interesting example would be: Take the first 12 bytes of the input (pad with zero if necessary), concatenate it with the SHA1 checksum of the input. This hash has a length of 256 bit, is cryptographically collision-resistant, yet leaks information and for any message up to 12 byte
is trivially invertible.

This is only true for cryptographic hashes.
That was what he said, in the parts you erased.

You were the first to talk about cryptographic hashes. The OP just talked about hashes, and hashes serve a wide variety of purposes other than cryptography.
.

Relevant Pages

  • RE: [7.8.2002 44916] Notice of Copyright Infringement]
    ... Appending a single bit onto the end of the file makes a different hash. ... and you no longer match the hashes. ... The only way to prove you're breaking copyright is to download at ... |"real" warezed version of whatever movie. ...
    (Vuln-Dev)
  • Re: ACCEPT and the SCREEN SECTION.
    ... General-purpose hashes, for hash tables and similar data structures, ... In short, cryptographic hashes like MD5 are used to verify data, ... if a client asks for and receives a chunk ...
    (comp.lang.cobol)
  • Re: what is probability to create two equal hashes for md5 algorithm
    ... Other than that, if the hash is statistically good, the longer the hash, the ... few cases the hashes match. ... md5 and crc32)? ... How much does it cost to compare two hashes? ...
    (sci.crypt)
  • Re: Parsing problem
    ... > I get into hashes I get a headache. ... First off a 'hash' is simply an 'associative array'. ... They are in turn a reference to a hash. ... a punchcard column formated cobol wingDingDingDing style ...
    (perl.beginners)
  • Re: Is MD5 outdated ?
    ... >> produce identical hashes, never mind hashes differing by 1 bit? ... > bit will themselves differ in roughly half their bits, ... > in one crucial way and yet hash to the same digest is far easier said ... > desired way and yet produce the same hash. ...
    (sci.crypt)