It's time to upgrade when your stuff is so old that he "early adopters" no longer laugh at your rig, but are amazed that those antiques can be useful.
This also helps resist supporting the Evil Empire!
I'm not a cryptologist, just a geeky engineer, but it seems to me that we're missing the point. Everyone seems to be making the assumption that you can use hashing to absolutely assure that the input data was what we thought it was.
Since the hash contains less data than the original message (again, within the assumptions of the orginal article), we're dealing in probabilities. It seems to me that two factors come into account:
1. That the probability of any message having the same hash as any other is sufficiently unlikely that it's useful (i.e., the above 2-bit digest value example)
2. That the probability of a false message passing a hash check falls within our level of comfort for a given system.
While I'd like for the hash my bank uses to be 100% reliable (at least when the error wouldn't be in my favor), I'm willing to accept that things will go wrong from time to time. We societally accept these risks.
If we can't accept collision probabilities, then we should forget digesting altogether.
Slashdot Mirror
Perhaps back to an abacus? Remember that great GUI?
It's time to upgrade when your stuff is so old that he "early adopters" no longer laugh at your rig, but are amazed that those antiques can be useful. This also helps resist supporting the Evil Empire!
I'm not a cryptologist, just a geeky engineer, but it seems to me that we're missing the point. Everyone seems to be making the assumption that you can use hashing to absolutely assure that the input data was what we thought it was.
Since the hash contains less data than the original message (again, within the assumptions of the orginal article), we're dealing in probabilities. It seems to me that two factors come into account:
1. That the probability of any message having the same hash as any other is sufficiently unlikely that it's useful (i.e., the above 2-bit digest value example)
2. That the probability of a false message passing a hash check falls within our level of comfort for a given system.
While I'd like for the hash my bank uses to be 100% reliable (at least when the error wouldn't be in my favor), I'm willing to accept that things will go wrong from time to time. We societally accept these risks.
If we can't accept collision probabilities, then we should forget digesting altogether.