One thing I always feel I want to do while watching a cast is to zoom around and watch stuff.
If you have 2 screens you can do that, watching a cast on one screen and running the rerplay on the other. But the problem is syncing the replay with the cast
So I was thinking if you had a script to use Optical character recognition on time of both the youtube video and the replay, and compare them
And based on whoever is ahead, make the game either +1 or -1 in speed(or more but one inclination every check)
I found a Possible candidate for OCR engine, it is based on autohotkey so it should be possible to do some tricks with autohotkey
http://capture2text.sourceforge.net/ I did a quick test with it and it seems to be working, I dont know if it works flawlessly every time however(did not check multiple times)
It has option to output to Clipboard, not sure how easy it is to use that... But I assume it is possible to act on it directly instead, if we look into the source code(a autohotkey script)
I might just try this on my own because it seems like a fun challenge to get it working. But I am afraid I might be lacking in coding skills to be able to get it to work.
As far as I see the problem will be to easily collect the 2 values that need to be collected, the comparison of the two and the following actions to preform should be fairly trivial in autohotkey. And if it is to be used by more people it needs some kind of UI to say what part of the screen to be collected(for my own needs it can be hard coded ).
If anyone that is a competent coder wants to do this please feel free to do so I will probably try myself if no one voulenteers