Since last time I have spent a while thinking a bit more about the two Boring World test cases. Each was designed to be the most boring according to each metric for suspense I had come up with, the idea being to try to judge the relative weighting of each factor.
But the two results have what feels like a pretty important similarity. The country coming first never changes. It occurred to me that this seems like a much simpler, better measurement.
Suspense = Number of times the country coming first changes.
This isn't perfect, because this system is likely to produce many ties for most suspenseful. I am not sure that it would be worth the complexity to avoid that.
I do have a couple of ideas for tie-breakers, so the full metric will probably be:
Number of times First changed, highest round for the last change.
This still leaves the problem of actual ties, 2+ countries coming equal first, depending on how we sort this may be static despite being pretty suspenseful. I think that just looking at the very top slot will still work though. Will need to ensure that I test this so we will need to include Eurovision 1969 as one of our early tests of the algorithm.
So we finally have a fitness function. The next step will be tacking the search space. There are 37! (That is 37 factorial, i.e. 1.3763753 * 1043) different ways of ordering the current Eurovision voting nations. Definitely can't brute force it.