1) Your assumption that there are no starbases really understates the problem. 5 economic starbases is easy to achieve. +res/manf techs add to it. And relics... etc. 600% is really easy to get.
Try reading this part of the last paragraph of my previous post again:
... if we relax the assumption about starbases so that the effect of the starbase is to increase the bonus to each output type by b (i.e. the starbase bonus is b*100% and this is additive with the bonuses from existing structures on the planet), then X = (1 + BR + b ) and Y = 0.25*(1 + Bm + b )*(1 + Br + b ), which under the assumptions that Br = Bm and BR = Bm + Br gives z = (3 - b + sqrt(12 - 4b )) as the lower bound for Z s.t. Y > X for all B > Z.
I chose to go with the assumption of no starbases for the numeric example because you actually get a number out of it without having to make further assumptions and because it's a bit easier to work out, not because I think starbase effects are negligible or anything like that. If your relic bonuses, civilization bonuses, etc also apply evenly, then they can be tossed into b as well. You're more than welcome to work out a formula for a generic setup; it'd start from this equality:
1 + N_labs_p*B_labs + N_lablvls_p*B_lablvls + B_starbase_r + B_planet_r + B_civ_r + B_uniquebuildings_r + B_uniquebuildings_r_lvl*N_uniquebuilding_lvls_p
= 0.25*(1 + N_labs_m*B_labs + N_lablvls_m*B_lablvls + B_starbase_r + B_planet_r + B_civ_r + B_uniquebuildings_r + B_uniquebuildings_r_lvl*N_uniquebuilding_r_lvls_m)*(1 + N_facts*B_facts + N_factlvls*B_factlvls + B_starbase_m + B_planet_m + B_civ_m + B_uniquebuildings_m + B_uniquebuildings_m_lvl*N_uniquebuilding_m_lvls)
You could probably collapse several of the terms together, and I may have forgotten a term or two, but the general case is not something particularly easy to solve.
Furthermore, you're already assuming that you have access to a fairly significant resource base to work off of. If you want it to be strictly better to use a mixed world rather than a 'pure' world for research or wealth, then you require b = 3 (a 300% bonus to both manufacturing and the output type you wish to produce), assuming equal non-factory and non-lab/market bonus sources. Even if b = 2.9 (+290% to manufacturing and +290% to weath/research), you still need to come up with about a 74% bonus to each of manufacturing and the output type you want. Early on, this simply is not going to happen, and it's early on when the rewards from perfect optimization are at their greatest. It doesn't matter that much if you're "wasting" 10% or 20% of your theoretical maximum output if you're running an empire of several dozen planets; you still produce enough to get on competitively with your neighbors. It's when your empire is small and undeveloped that getting the most out of what developed worlds you have matters most, and at that point in time you're probably not fielding 5 fully upgraded economic starbases around every planet with enormous civilization-wide bonuses from relics, special events, technologies, etc.
2) Factories have all sorts of other bonuses - for example, at turn 30, you will have a much more developed world if you build 50% factories 50% labs, compared to if you went 100% labs from the start. I dont think this should be underestimated. It takes forever to build a 100% specialised world, and by the time you do, you are near the 600% point where it was all a waste of time anyways!
You could always build factories and then replace them with labs, in whatever build order results in an optimal time for such an operation.
2) If those pesky Drengin decide you look tasty, being able to instantly convert your research specialised worlds instantly to ship-producers can be life-saving.
Shouldn't this be #3?
