My preferred approach (other than going back to the drawing board and building a better balanced economic system from the ground up, which isn't going to happen) would be to increase the modifiers associated with researching ahead of era and behind era techs and civics.
I'd double the value of science/culture contributed to researching behind era techs/civics.
I'd half the value of science/culture contributed to research techs/civics that are one era ahead of the current world era, and cut it to 25% for techs/civics that are two or more eras ahead.
The benefits of this approach are expected to be:
- Damaged AI civs won't fall as far behind the rest of the world.
- After a certain point, adding one more campus/theatre square will offer dramatically lower returns. This would nerf pure specialization strategies (campus in every town) and infinite city spam strategies.
This system is also somewhat self-correcting, unlike straight "double tech cost" style solutions, to future tech/civic additions or changes to base science/culture yields, as well as to different player playstyles. If you generate insufficient science to get all current era techs before the world era moves on, you'll be able to catch up. If you generate too much, the game will slow you down. You'll still be better off to be ahead of the AI and not behind, but you won't get as far ahead, and if you're a new player you won't fall so far behind.