Radio has been taken out in C3C as well. There was a slight flaw in implementation (4 turns for a tech) that AlanH caught when looking things over for C3C, but that was taken care of. In the "huge" tech rate game things were off as well. Other than that the best dates are as intended, but are limited in the factors they take into account.
Ideally all factors could be quantified and taken into account in a manner which accurately represents their effect on the game. Number of Scientific AI's and quality of initial landmass is outside the scope of the scoring system.
Back when the free tech was truely randomized in one of the PtW patches cracker and I discussed the change and IIRC the decision was that the number of Scientific civs was going to be kept to 1 or 2 so that the difference between 1.29f and PtW tech rates would stay closer to the same. I forgot about this decision and I don't think that it was ever conveyed to Ainwood. It is certainly a valid concern and a bit of variation could be taken out by returning to the stable number of Scientific civs.
That said, because of the modding to the rules and maps, the scoring system was reworked to be much less dependant on the best dates. Date is the more faulty side of things. Max score is relatively stable and so the scoring system is mainly a comparison of base score against the predicted "milking" curve. At 2050AD the date bonus is +0% to the score increasing linearly by turn to the best date at +15% (not sure on this, might just be 10%). So the difference of 20 turns around the best date on Spacerace (usually with around 300 turns remaining) should equate to ~1% Jason score. (assuming that base score maintains the same relation to the predicted curve during that time)
My initial goal was to get a system that would bring all the victory conditions within 5% max scoring potential from map to map (un-modded in across a single version). With the mods and various versions supported it's 5-10%, which is as good as I could have hoped for. I would like to improve the scoring system, but at this point I'm afraid adding factors would simply make it more complex and/or add further variation. For instance, if you account for the extra Scientific civs, but they end up all getting duplicate techs, the problem would only be reversed. Same thing with techs from Huts.
---------------
I still think that many of the "fast tech" (quickly to Cavalry) Domination/Conquest victories we see scoring at the top can score a bit more by building up to a Diplomatic/Spacerace victory. There have been examples of just about every victory condition hitting near 11-12k.
In this thread I played SirPleb's Gold Medal GOTM19 Domination (Jason 11787) out to a Spacerace victory (Jason 12229) to help illustrate this.
Some maps will favor one victory condition or the other due to factors the scoring system isn't designed to account for, but the only consistant advantage Domination/Conquest has (at least in 10k+ games) is that it requires less playtime. That is the reason I think most of the top scores end up Domination. To get a top score with another condition you still need to get to the domination limit ASAP, but then have to do the "milking" to get to the victory condition. (It has to be this way, otherwise we'd be right back to milking, as the most time intensive victory condition, being the only way to score high.)
