Accelerators are actually a terrible business because you only make money if you get a big hit (which most won't), and even then it can take 10 years or more to realize the gains. I predict most will disappear in the next 5 years.
What's your advice for a new accelerator (or any investor for that matter) in measuring if they are doing well? The problem is one of retarded measurement of success. It feels like you are doing well because you are deploying money, you are being highly selective, etc. But you can still be slowly dying without noticing. What are some good predictors? You guys at YC have some heuristics for identifying good founders, but I believe it would be hard for new accelerators/investors to use those, because they are highly subjective and I would argue most new accelerators/investors will not know how to assess those founders characteristics when they are getting started. Would you advice people to experiment for a few years (literally being willing to loose all the money invested) just to develop that ability to measure, and be ready to be a good accelerator/investor a few years down the line?