Koshling - Not sure where to post this, but I've attached a save game (Settings - Large/Epic, Immortal to Diety on Turn 65) in which at least one AI beat me to Sedentary Lifestyle (Egypt on Turn 145, whereas I reached it on Turn 156). That a fabulous AI performance - that's the fasted I've ever gotten to SL with these settings, & the AI still easily beat me. The logs are attached too in case they may be of some use. Two quick things:
1. There seemed to be fewer AI Trackers/Hunters than previously, though if this is the 'cost' of better AI teching, that's a good compromise.
2. I can't remember exactly, but I think Rome researched a few non-essential (at least to getting to SL) aquatic techs even though they are landlocked.
Thanks for the logs. There are a few interesting points:
1) There are likely to actually be MORE hunters abroad than previously with the recent changes. The logs also back this up, showing quite a high population of scouting and hunting units, so I guess you just didn't see the high density areas so much
2) The tech path Rome took shows a few interesting points that may be worth tweaking. The path was this (just the targets - it will beeline between them as needed so you need to interpolate a bit):
Oral Tradition
Hard Hammer
Persistence Hunting
Cave Dwelling
Controlled Fire
Tracking
Cultural Identity
Shelter Building
Basketry
Trapping
Naturopathy
Natural Pigments
Simple Wood working
Trap fishing
Tattoos
Binding
Bead Making
Sewing
Wood Working
Carpentry
Shamanism
Barter
Warfare
Canine Domestication
Tribalism
Portable Shelters
Animal Husbandry
Chiefdom
Boat Fishing
Ideograms
Although this isn't quite the path I'd take, it doesn't seem too terribly flawed. The interesting (and possibly questionable) elements seem to me to be:
1) Should Tribalism be a bit more prioritized? - I notice that the AI usually goes for canine domestication (which IS an excellent tech) beforehand
2) The reason it diverts down the boat fishing track is that that tech reveals lots of bonuses, and bonus reveal value is considered independent of what territory is available (other factors relating to things like buildable units ARE sensitive to this). This is the way it is mostly because the XML doesn't say anything about where bonuses can occur, so the AI doesn't know that fish can only occur in water. In principal one could infer this (from bonus find the set of builds that provide it, from those find the set of units that can build them, from those find the domain), but it's a fair bit of processing (cacheable though at asset-load time) and you also need to bear in mind that bonuses can be traded once you know about them, so knowledge of fish is worth something even if you have no water access. I am inclined to say it should do the calculation to determine the domain the bonus can occur in and then adjust the value as follows:
2.1) Not a map bonus - divide by 2
2.2) Land map bonus - full value
2.3) Water map bonus - Sliding scale between 50 and 100% value proportionate to how many of your cities are coastal
3) The reason it targets Ideograms at the end (which includes Sedentary Lifestyle on the path to it, incidentally) is that it has a HUGE value attributed to 'Library of Ninevah', a Wonder that the tech enables. The value is clearly too large (it's 160000 for the component for that building alone, where the average entire tech is worth circa 4000 or so for most techs). I'll look into why this is and make some appropriate scaling adjustments.