I never thought MD5 was supposed to be used to authenticate the content of a file, only as a quick check for simple bit errors.
I'm not buying any of this discussion about coming up with a new hashing algorithm. MD5 works great when it is used the way it is supposed to be. The trouble with coming up with something better is that you are trying to represent a large body of data with a smaller body of data. The ONLY situation in which your smaller body of data will absolutely be uniquely coupled with the original large body of data is in the case of lossless data compression such as LZW. Any other form of lossy data compression or tokenization will run the risk of duplicates matching the same fingerprint.
Someone who comes up with a magical fingerprint that can be tied to one and only one large body of data will have discovered the Holy Grail of data compression.
Slashdot Mirror
I never thought MD5 was supposed to be used to authenticate the content of a file, only as a quick check for simple bit errors. I'm not buying any of this discussion about coming up with a new hashing algorithm. MD5 works great when it is used the way it is supposed to be. The trouble with coming up with something better is that you are trying to represent a large body of data with a smaller body of data. The ONLY situation in which your smaller body of data will absolutely be uniquely coupled with the original large body of data is in the case of lossless data compression such as LZW. Any other form of lossy data compression or tokenization will run the risk of duplicates matching the same fingerprint. Someone who comes up with a magical fingerprint that can be tied to one and only one large body of data will have discovered the Holy Grail of data compression.