How the HELL does your machine know how to interpret the information it has been given? - Surely your computer is somehow decrypting the information in order to do what it needs to?
You may be thinking that the idea of a hash (or digest) is to "encrypt" some information so that it can be "recovered/decrypted" later -- but that's precisely the opposite of what you use a hash to do.
The whole point of using a hash/digest for security purposes is to convert plain text data (like a password), into a (unique) representation such that it's completely impractical for anyone to reverse it and get back the original. The computer never "decrypts" the hash back into the original. That's why we call hashes "one-way functions".
The way you use hashes to handle account passwords is that you hash the plain text password and store the hash. The next time someone enters your password, you again calculate a hash of what they type, and then compare it to what they stored. Now you see why no website has a function that will resend you your password if you forget it -- because they can't do that -- they can't get back what your original plain text password is. That's why they have to send you a link to RESET your password.
Hash functions are used on files for a different purpose, but the idea is the same -- you perform a mathematical hash function on the input, which can be arbitrarily big, and the output is a number (or a short string of bytes). If you want to compare 2 files to see if they are the same, you perform the hash function again on the new file, and compare the output to the result you got when you ran it on the original file. And because the hash function cannot be reversed, people cannot create fake files that generate the same hash as (and therefor might appear to match) the original.
So.. the whole mathematical beauty of hash functions are that the are very carefully designed to be impossible to efficiently reverse (they only go one way) -- and that they have good properties in terms of rarely (but it can happen) mapping different inputs to the same output (collisions), even when the input data is large.
You can't just use any function as your hash function -- it needs to meet these requirements. If it falls short, then bad things can happen. If a hash function turns out to be reversible then malicious people could create fake files or fake passwords that compute to the same hash; If a hash function turns out to have lots of collisions, then you will end up with cases where many files/passwords look the same (according to their hash) when they are different.
Recent Posts
