An ICO is a horribly inefficient method for raising capital.  (I also believe most run afoul of U.S. securities laws, even if the SEC has not been aggressive at enforcement.)

In conventional venture capital, as bad as it may be in many ways, money invested is usually proportional to company progress.  A company gets a little investment to start-- a few million dollars, usually, or even less.  Then when it has success building a product or finding product-market fit, it can raise more money--- maybe tens of millions.  Finally, when the company is growing quickly and has proved its value, it can raise hundreds of millions in an IPO, or from large private equity investors.

ICOs almost uniformly squash all these tranches together. The founders get all the money up front. (And why wouldn't they?) They raise entirely on promise, not on results.  So if the idea doesn't work, or the token doesn't find an market, there has been enormous waste, compared to incremental funding.

There are other flaws, but I think this is the key one: fundraising happens all at once, without any proof of product-market fit.

(I wrote a more in-depth analysis of Filecoin on Steemit previously: https://steemit.com/filecoin/@markgritter/should-you-pay-for-storage-with-filecoin-probably-not)