What I have experienced is despite having shortcomings and bad implementation, if it has gained too much network effect of both investors and developers using that network, it tends to keep going for long, as these investors and developers feels too much invested into it, they feel like just 20+% more before our project works perfectly, and they don't leave it soon enough. Don't remember the term for that in psychology, but that's seems like happening in BTC and ETH. As as time passes by, no negative news affect to divert the capital into better architecture cryptos in industry.
It's much like TCP/IP v4, the industry is just too much invested in to v4 architecture and is giving less importance to move to v6, until it gets really really unable to move with old version.
with such understanding, I expect ETH might not get less valued in coming years, but much possibly will gain far more than past years it has ever had.