So, I understand why most commonly used password policies are used.
N character minimum? No low hanging fruit left for brute force attacks.
N character maximum? The database field can only be so long.
No unicode? The underlying database doesn't support it.
Include at least one letter, special character, number, and heiroglyph? Adds a few extra required bits.
The one I can't for the life of me understand though is the rule where no more than n consecutive characters (commonly two from what I've run into) may exist in the password. I can't think of a technical reason to do this, and it actually reduces the available password pool, thus making a brute force attack take less time. This practice seems too wide spread to not have some justification, I just can't think of what it might be.