Hindsight comes into play when people promise to deliver something for 50 YEARS !!! and fail to deliver but keep saying they just need more money to git er done.
They promised to deliver said something in 50 years given increased funding. Instead, the funding steadily decreased. Well guess what, if you keep giving them less money, the schedule is going to stretch out further and further. So that's your "it's always in 20 years" effect, completely self-fulfilled.
Not to mention that theoretical science and advanced engineering like this is very hard to estimate accurately - you can't really estimate how long it'll take for the breakthroughs necessary, you can only roughly gauge the amount of experiments you might need to run. This isn't Starcraft where you click on a button to research something and get a progress bar.
If it were that easy, then everyone in the field would have been expecting Skunkworks people to come up with the first working thing. But few people bet on them, even despite the fact that the details behind this project were available for quite a while. To me, this indicates that the subject is much less obvious than you make it be.