A cool little app would let us list "problem signs" and we could vote on their seriousness and then assign a weight to each one. Then you could answer it like a survey and get a "Failure Prediction Score".

I'm busy. Someone send me a link when it's done, please.

I do recall there being a study about how internal prediction markets were really good at forecasting software release dates...

Wait a minute, I was hired as a consultant... and I did take the blame! Yup, pretty sure that project failed.

We have 15 of these. Is that bad?

I would say that the second sentence in your comment was probably more significant than the first.