I'm working on fine tuning a very fast online search algorithm I invented, provisionally called HashChain.

It's already faster than all other algorithms for most data, pattern lengths and alphabets, but it also has a lot of different ways to parameterise it. So I'm exploring the parameter space, so the algorithm can auto tune itself.

Not sure if anyone really cares about this sort of thing anymore. Most people still seem to think Boyer Moore is the best, but that is positively ancient and long superceded by others.