[noize]

If it's randomly generated it's not even a hash. A hash is a one way algorithm, but for each hash, we cannot have more than one plain text.

I mean, if it is randomly generated, "plain" and "text" might get to be the same hash.

[...] your argument is not true. Every hash function has input values that result in the same output.
The reason is, that hash functions have a bigger data set as input than as output (input can be every string of every length, the output set has exactly 2^hash_length_in_bit values). It is impossible to not have collisions. You can just take care to make it hard enough for others to find them.

Yes, you're actually right, my bad.