So assuming the basic theories stuff I've proposed are used, what's left is to fill out the tree with applications and refinements, and to (perhaps) add some effects to the theories. IMO, a good way start this would be to establish what the difference between a theory, application or refinement is.
I figure the idea should be that theories themselves generally do nothing; they only unlock applications and refinements.
Applications would be broken into two basic types: "direct effect" applications, which would themselves have effects that do something interesting (eg. bonuses and penalties), and "unlocking" applications, which themselves do nothing, but which are basically only there to unlock something the player can build or some game feature.
The unlocking applications would be named specifically for the thing they unlock, so for example, if there was an "Imperial Bank" wonder, then there would be an associated "Imperial Bank" application technology under some economics theory. The application tech would just unlock the appropriate building, which the player would then gain access to. Ship parts would be unlocked essentially the same way. Other unlocking applications would unlock features, rather than buildings or ship parts or other "buildable" things. This would let the player pick a new government, sign a new type of treaty, do a new spy mission, use a new type of biological plague, etc. The degree of distinction between "buildable" and these latter examples is hard to say at this point, as we haven't really designed any of the features yet, but I'm trying to look ahead.
The "effect" applications would just have effects that do something interesting, like boosting some meters. There's not really much we can do with this type of application at this point, as the v0.3 effects system is quite limited. In future I hope there'll be a larger selection of interesting effects that can do interesting things, making for an array of interesting effect applications.
For v0.3 in general, there's probably not much point in putting a whole lot of time and effort into designing application techs. Unlike the theories, which presumably could be carried as they are until v1.0, the application techs we currently have the tools to create are too limited to be anything resembling a finalized version. This applies to both the effect applications, which are limited by the set of effects we can add, and the unlocking ones, which can't even be implemented at all yet, as there's no way to build anything in the game, and even if there was, the buildings are basically limited to the same effects as techs are...
SO... my suggestion is not to worry too much about v0.3 buildings and application techs and such. In the event the buildings system gets implemented, I'd planned to put out a slew of filler buildings, building-unlocking applications, and direct effect applications to basically test the system. Until we get more interseting effects, that's pretty much all that's worth doing, IMO.
As for refinement techs, they can currently just add effects groups to buildings. I've asked a few times, and desparately hope to get, the ability to remove (or deactivate) effects groups as well, as IMO this would be about 47 times more useful. In any case, the effects groups that can be added by refinements are just as limited as the effects groups that can be on buildings in the first place, so the same suggestion applies as it did for unlocking application techs: just put some filler testing stuff in v0.3, and worry about interesting stuff later.
From a design standpoint, we might want to figure out what other sorts of things refinements could / should do besides adding (and removing?) effects groups from building types. Presumably they'll change ship part stats in some way, though that can wait until ships stuff is designed... But are there any other general / broad types of things refinements could do that doesn't break down to just adding effects groups? If there are applications that unlock things like governments, espionage, etc, presumably these things could also be "refined" in some way... Can we, and should we, figure out what sorts of things would be involved now? (or wait until those things are designed?)