I've been pondering for a time on the use of a research system a lot like the taming knowledge that dwarves collect through animal interaction.
It's most likely impossible or just not feasible at the current time, but I'll give some of my thoughts anyway.
It would be neat if alchemical tasks adding slowly to the overall alchemical knowledge of certain elements or forces, say you get lucky and get 2 ethereal reactions right off the bat, your alchemists would learn more about the process of pulling ether from the void and you might have better luck with it, perhaps at enough research poured in, the ability to specifically use a reaction to search for ether.
The same could be for other tenets of dwarven research, from the production of bayonets to the magnetic dart launchers, or delicious, Delicious robotics. If they were randomly generated with the civilizations, albeit with a cap (Nobody who knew exactly how to make the things moved on, only those who may have been involved in the process or used them); Knowledge of their existence is present but not much about making them.
That gives a bit of headway into the pursuit of science, at least for one or two areas, without having to cause some sort of breakthrough to gain access to them (Like the slabs of hastily chiseled, etc) which could be specifically to unlock research areas entirely untouched.
The idea would be that alchemy experiments and the like produce results, which are filed away until a breakthrough is made, the breakthrough officially adds that section of Science to the research pool, and additional dabblings in that area slowly raise your fortress' knowledge of it. Robotics might rise up to General familiarity (They stopped trying to kill us) and you as a result, have better luck building them or possibly access to more designs.
Those are my initial thoughts anyway, I might add more to it later, if everyone doesn't completely hate it. :3