It wasn't until modern times that people actually directed their research at a specific targets. In fact, most research is done 'post-discovery' meaning that there's a period of general research in the area which is "blind" and then there's a sudden insight, and thereafter, the research is directed/guided so as to be applicable. Most often, you don't actually theoretically know about something before you discover it, you formulate the theory after the experimentation based on the hypotheses/educated guess.
eg. gunpowder/guncotton were somewhat accidental discoveries which once discovered, were researched further into applications and uses for society.

so I would favour this form of semi-blind research, but also on PARALLEL paths. Just because I'm researching steam engines doesn't mean I can't be simultaneously researching improved irrigation techniques or something like that. You could use a slider bar which allocate to different categories, with semi-random chance of breakthrough in the general blind research and then once a breakthrough is made, hasten the application/completion by funding that particular tech.

Can't remember exactly but MOO had something like this? SMAC's was a decent system too, except for the actual techs themselves.

It was an "interesting" way to play the game, as I was forced to be more flexible w/ my strategies. I could no longer play the same tried-and-true strategies, but I had to adjust to whatever technologies I was stuck with next.

It was quite fun, and I think it ultimately made me a better Civ player

I can't tell you how many times the well being of the whole of my civ was based on a particular science advance that I was waiting.

I think that knowing exactly what you are researching adds to the strategic element. Having said that, blind research offers more unpredictability and thus.. playability