I know this is testing AI, but maybe it's also revealing as to some imbalance between the different factions? Procipinee and Markin seem pretty underpowered to me whereas the Empires are a bit too much so.

I haven't seen Markin be underpowered in my games.  They have superior weapons and more hit points, and golems.  They and Tarth are the ones on Kingdom side that give me the hardest problem.

I did not know that.  One lives and learns.

Then 'Cruel' is the uncontested winner of 'Weakness that only a masochist would take" contest.  I played Karavox recently, and I can attest the penalty is Empire wide, and very significant.

Tarth , hmm master scouts op much? At the absolute least they need to make it 2 points. Ythril, Gilden, and Kraxis definetly due to those sovs having excellent early game traits. Bad early game for the ai is almost a death sentence it still reaches a kind of shutdown point where it can't expand and doesn't know what to do, I refer to this as "stalling out".

A big factor in this is the MinDistanceFromStartingPointsin the coregoodiehuts.xml not being adjustable per map size, i.e. on bigger maps an ai that gets hemmed in by high level monsters at the default 12-16 avg distance gets crushed by ais that are able to spread, there is no catching up. If SD doesn't change it I will in my mod but it will have to be a tailored coregoodiehuts.xml for a specific map size. It would be really nice if they put in ...

<GoodieHutType>

 <MapSize>

<Medium>

<MinDistanceFromStartingPoints>12</MinDistanceFromStartingPoints>

</Medium>

<Large>

<MinDistanceFromStartingPoints>16</MinDistanceFromStartingPoints>

</Large>

</MapSize>

</GoodieHutType>

Medium/Large/etc would be the Internal name from CoreMaps (so you could add custom map sizes) and any attributes you place in that element only apply for that particular map size. And it would be nice to know the general starting position algorithm so we know the average grid distance between players so we could set the minimum spawn distances accordingly.

From what I remember, those values don't apply to the AI anyway.

Golems... now there's something that has a lot of potential for version 1.1. One game Markin found a hammer that cast shrink on hit (or does he always have it? I only fought him once). Since he could take a hit no problem, he could just tap everyone in your army with his shrink-hammer and then clean up afterwards. Of course, the AI didn't figure out to do that but rather kept him bashing away on the same unit until it died, it wasn't a game-ender for me. But it still took me some figuring out to beat him.

-1Hp per level is a major penalty. Pariden no longer needs Inefficient to be balanced. The weak troops are enough. And Magnar's slaves need only -1Hp per level. That plus -10 weight capacity and less Initiative is why you will never see a slaver win Elemental.

More results, more results!

Ug if thats true then thats actually the problem. You can watch the ai play and see what happens if they get stuck in an area due to high level monsters then the ai  almost completly stalls out. Its no wonder why Tarth can get through these monster chokes and the other factions are the ones with sov's geared to early game melee. They need to either push back the high level monsters or give "wizard" something to help them in the early game.

Is that wad? Otherwise whats the point? Sounds like inefficient is supposed to be empire wide.

Without hundreds (or thousands) of games with a detailed look at the replays to see why some are winning and others are not... I don't see how this is useful.

How long does it take to run an AI vs AI game with all the graphics disabled? 

Then it sounds broke to me. Does not make much sense if its not empire wide.

inefficient tool tip states that it applies for the sov when stationed in a town/city only, and this has been so since WOM, SO

when the sov looses a battle and hoofs it back to the nearest town to heal, THEN that town slows down the production until the healed and the sov leaves the town and as the town is often out at the front lines this can be significant when battling other sides.

harpo

Its a nice list, seeing at what turn it happened, and who they won against would be nice .
Good to  see you doing this

Sincerely
~ Kongdej

Yes it should cost 2 points, but they should also remove the "Monsters ignore this player"-part of the master scout trait. As someone else pointed out in another post somewhere, you miss like 1/3 of the game if you take monsters out of the equation and that simply isnt fun.

(Btw, i will still take this trait 100% of the time because movement is everything).

So Brad and Derek, make it happen BEFORE release!

Well Im gonna pick this everytime! 99% of the time it will not affect me!

Agreed I modded that out of the pick it is way too powerful!

I'm pretty sure Derek made some changes to this.

Specifically, I think in faction creation you now get more total points but that some of the really big faction abilities cost 2 or 3 points or something.

I'll definitely be limiting my contribution in the beta forums to constructive bug reports from here on out.  It appears that the game we're currently playing and the one you have in store for the world next week are versions apart in themselves.  Your team is astonishingly quick.

This is so the right thing to do!

quick easy fix

As a player I would of course always want more interesting choices, but this will help even out the choices.

Sincerely
~ Kongdej

Great to hear. The built-in factions seemed to have an upper-hand against the custom ones.

How cool would that be to play against a randomly generated AI sovereign, instead of just the built-in ones? Probably not practical, but it would be cool.

Would be cool if you could train henchmen as fallen race. As of now I cannot train evil henchmen I guess that goes along with the bloodlines. How about traitors? Altarian blood but evil?

Edit: I take that back I just discovered you can woohoo!

Whut?? I think you got it the other way around

Sincerely
~ Kongdej

Are we going to see more testing? I would like to see the AI results even without the other information.