I also wonder if it is a victim of the just-one-more-try problem you see in big code projects: you get funded for a year or more and the lead researcher honestl

I also wonder if it is a victim of the just-one-more-try problem you see in big code projects: you get funded for a year or more and the lead researcher honestly thinks he is - almost there - if only for that one more try. This means all resources are diverted, and only really late in the game you have to admit you were wrong, which gets harder and harder to do.