It depends what era you compare to.
If you think of ICOs as a world which is only 9 years old, then we are merely just past the dotcom crash now. Yet, the ICO world is even not 9 years old.
Additionally, what do you call success here? Amount of raised capital or actually dApps which achieve longevity.
Lastly, what is the table of comparison since many a startup receives pre-seed, or angel funding, but never manages to close an A-round and are never reported either.
Currently funds raised from ICOs are close to achieve parity with funds raised from more traditional VC, but the ICO graveyard is rather full already as well. Beginning this year Bitcoin news reported that almost half of the ICO raised in 2017 had folded already, or gone MIA, by the end of the year.
Generally, one can say that ICOs will achieve a similar success rate as startups: 1 every 15/16 will succeed and achieve scale sufficient to continue, and “succeed”. That compares to “success” rates of large startup fund programs such as TechStars and 500Startups. That even though nowadays “lean” is fully
But there are more inherent problems with BC technology, problems which we can not necessarily relate to yet and are as such an unknown quantity.
DApps can often follow the book of the lean methodology, but the more “waterfall style” blockchain at the core has an own governance, governance which may make decisions which can affect the evolution and possible success of dApps. Pivots may also be much more difficult in a world of ICOs, especially if at some point regulatory frameworks are created and a WhitePaper becomes possibly a rule book (for SEC).
TL;DR: Just like in this online tech Millenium, expect many fails. And almost as many flukes. But the amount of capital raised by projects will soon be baffling, much more baffling than Sandhill Road throwing money round. EOS’ ICO is a perfect example of that with a multiple billion valuation already even before actual release.
Until last year the success rate of ICOs were good but this year bad.