"So don’t join an accelerator unless you can win it." So we're basically in a situation in which self-selection rules now. In the future, if you made it through an accelerator and "won" it you were already in a position to do so. That means that the benefit of actually winning the round from the accelerator was minimal since you most likely would have succeeded anyway.

Given the massive difference in funding between an accelerator seed (tens of thousands) and the first raised round (millions) it seems like the accelerators disproportionately gain from having these self selected success stories.

Accelerators, more recently, really seem to provide business direction and a stream of well documented CEO advice which, for a lot of these startups, solves zero of their engineering or customer problems. It's not that the mentorship provided at the labs is not useful but rather that it seems to solve a less important part of the overall problems early companies have.

And to top it off, like the article mentions, the effort rewarded during an accelerator is often related to pizzaz, flash, and cool demos rather than core product. I hope the accelerators look to strengthen their offerings to seed companies and really incubate them rather than slip further towards the "gateway to VCs" which, at times, seems like the predominant movement.