I just came up with a faster version of ASCII on the subway this morning...
Since string parsing usually does a lot of changing case or case-insensitive comparisons, wouldn't it have been better if small and capital letters would have been exactly the same bits except for something like the highest bit? That way if you want to convert between upper & lower case, all you need to do is flip a bit instead of having an if statement check if the char value is > x && < y...
I know it probably wouldn't make a huge performance impact, but every little bit helps.
Now I just need a DeLorean so I can go back in time and change the ASCII table...