My plan for expansion and science is to do neither.
Everything is a trade-off right? The snowball effect can take multiple forms. Early investment in growth, culture & production leads to science. Early investment in units leads to domination. The quicker you can take cities, the less cities there are to take, the less units you face, the less time they have to build walls, the less time they have to fill in the gaps.
Aside from 1-turn worker repair (2 policies + pyramids) Liberty is only worthwhile if you intend to have a lot of cities. Tradition is only worthwhile if you intend to have a high population capital.
If you concede that you don't need more than a few social policies or technologies to win, then virtually all you need to build is units. The faster you build units, the faster you take cities, and thus the faster you win.
So, if you want a super-fast finish, play it like you only get 3-4 policies. Assume you won't even get past medieval techs. Plan your tech order around the things that immediately benefit early conquest. Now, Poland changes things a bit. You're going to get enough policies to complete a tree, or to go hybrid and partly invest in two trees.
So, you can get real creative. Maximize your production, gold, and unit count. Don't be afraid to be running with a negative economy, or negative happiness, or a -70% production penalty from being over the unit cap.
But, above all remember my rule of thumb for domination:
Movement rate is *everything*!