Requests/Suggestions Thread

[pavelk]

Hello again,
I am sorry for not coming back to you earlier. We have been really busy with the coming update and therefore despite my presence on forums I was answering only questions related to technical difficulties or updates.

There is a significant and long lasting relationship bonus that is granted to two factions fighting on the same side of a war against another faction. This relationship bonus is hard coded into the game and has a large impact on AI behavior. I suspect a similar coding practice exists for Aggressors: Ancient Rome. Overuse of the "declare war to curry favor" strategy can be a bit annoying from the human player's perspective, though.

I agree with you NKlein1553. I haven't said that we are 100% happy with the current diplomacy and that is also the reason why I am happy to investigate an AI diplomacy behavior which might look odd. There is now a bit less time for bigger changes but tweaking AI is definitely still a pretty big priority.

I'm wondering if you have any plans, Pavel, to impliment some kind of a casus belli system for Aggressors: Ancient Rome at any point.

In our tracking system is similar thing already logged and not only that. There are things which would bring the diplomacy to a totally different level. I cannot promise it anytime soon but what is still in plan is some kind of "strategy planning" together with AI (or another player). Basically you (or the opponent) propose a joint attack (on political map) how would you synchronize attacks and what would be taken by you and what would be taken by the opponent. The actual goal is to have a system where you can plan something like "dividing Poland between Germany and USSR in 1939". It will be there, I can say that, I just dont have time for it now. We are getting more and more busy with upcoming extension where is already number of new features which are blast on their own!

We still cannot afford to hire another developer which would be a huge help, that is also the reason why any kind of support (visibility or reviews which would increase sales) would make it easier in the future.

It is true that during the second world war most of South America declared war against Germany to please the USA even thought only a couple sent so much as a handful of troops to the front. But that is not the situations I described.

Yes I agree that in the situation you described it simply doesnt make much sense. I am happy to investigate it further if you have an end-turn save right before that happened.

I suspect part of the problem is that the AI isn't able to distinguish between a real war in which there are real clashes and a phony war but, even allowing for that, the above makes no senses.

That is actually not true. There are mechanisms with which AI is capable of understanding how bad/good the situation is to go to another war. Despite that I absolutely agree that there is a place for improvements for sure.

[dejvid2]

We still cannot afford to hire another developer which would be a huge help, that is also the reason why any kind of support (visibility or reviews which would increase sales) would make it easier in the future.

Point taken. I do recognize that getting AI human like would be monumental task. But being aware of where it falls short can be helpful in improving things.

It is true that during the second world war most of South America declared war against Germany to please the USA even thought only a couple sent so much as a handful of troops to the front. But that is not the situations I described.

Yes I agree that in the situation you described it simply doesnt make much sense. I am happy to investigate it further if you have an end-turn save right before that happened.

I don't think I can retrieve it now. Next time I will immediately rename the game save from the turn before. (Should there not be a next time then it is something that occurs too rarely to be important).

I suspect part of the problem is that the AI isn't able to distinguish between a real war in which there are real clashes and a phony war but, even allowing for that, the above makes no senses.

That is actually not true. There are mechanisms with which AI is capable of understanding how bad/good the situation is to go to another war. Despite that I absolutely agree that there is a place for improvements for sure.

I don't think that's what I am talking about.

You are saying that the AI makes an assessment of how powerful its enemies are and whether it can afford to take on a new enemy.
Of course, if the AI makes that assessment on the basis of what troops appear on its border the Seleucid Empire in my game might have assumed that all that was left of the Ptolemaic empire was the south of Palestine where as any human player would know that Ptolemy controls all of Egypt and Kyrene.

But that's not what I was talking about.

Lets say the Gauls are at war with Athens. They are at war because that is the default state and neither have got round to agreeing to a peace. Is the AI able to grasp that as none of their troops have crossed swords with Athens they have no reason to care whether or not Pergamon (for example) is at war with Athens or not.

Again lets say that in the example m current game that the reason the Seleucids declared war against me was because the Gauls asked them to. [I don't know what the Gauls were doing in the actual game but lets say that the Gauls are at war with both the Ptolemaic Empire and Pergamon]

Are the Seleucids able to say "The Gauls may technically be at war with the Ptolemaic Empire but they haven't lifted so much as a finger against them while Pergamon is doing some actual fighting so Pergamon is the more useful ally.

In the actual Ancient World, the powers would have only a vague idea of what was going on beyond their borders but they would have been aware whether or not there was actual fighting between the different powers not simply whether or not there was a formal state of peace or war.

[dejvid2]

As I understand it, it is planned to enable players to mod the AI by writing their own dll files.
I also gather that for the moment, improvements in the AI are likely to be tweaks, in which one or other factor is given extra weight.
(And given there are a lot of things on the to-do list this is likely to remain the case unless there is a dramatic increase in sales allowing another developer to be hired)
Would it not be an idea to do a few alternative mixes of factors and give players the option of choosing these alternative AIs on the understanding that they would be guinea pigs as a tweak intended to solve one problem might well produce an odd effect in another area.

[pavelk]

Point taken. I do recognize that getting AI human like would be monumental task. But being aware of where it falls short can be helpful in improving things.

Yes and I totally agree. Everytime there is something weird or suspicious I try investigate that particular decision (ideally when end-turn save is available) and tweak it.
The decisions related to war are not easy to tweak. The number of factors taken into consideration is really high and I need to be sure that it will not break the "balance" that much. The only way to do this is to have time for a long run testing to see everything is fine or even better.

I don't think I can retrieve it now. Next time I will immediately rename the game save from the turn before. (Should there not be a next time then it is something that occurs too rarely to be important).

Thank you.

Lets say the Gauls are at war with Athens. They are at war because that is the default state and neither have got round to agreeing to a peace. Is the AI able to grasp that as none of their troops have crossed swords with Athens they have no reason to care whether or not Pergamon (for example) is at war with Athens or not.

Right now it doesnt take this into account and it is a very good idea. I will add it to our system and once there is some time I will look into it more.

As I understand it, it is planned to enable players to mod the AI by writing their own dll files.

Player's are already able to build their own AI behaviors. The problem is that this is not a matter of days and till someone with a very good knowledge of C# programming shows up, I doubt there will be any tries to create new AI players.

Would it not be an idea to do a few alternative mixes of factors and give players the option of choosing these alternative AIs on the understanding that they would be guinea pigs as a tweak intended to solve one problem might well produce an odd effect in another area.

It sounds like a great idea but to be honest, it is not as simple as you probably think. It s not a matter of few numbers to change (like from 1.2 to 1.1, etc.), it is a very complex decision making based on pretty high number of factors and weights. If there is no very detailed description of what is each of these weights for and what it affects, it would be really just a waste of time (ours and also the modders).
I guess that the AI development will probably stay on our side, we will try to promote modding and scripting more and if we succeed in that, it would already be a great step forward.

[dejvid2]

Lets say the Gauls are at war with Athens. They are at war because that is the default state and neither have got round to agreeing to a peace. Is the AI able to grasp that as none of their troops have crossed swords with Athens they have no reason to care whether or not Pergamon (for example) is at war with Athens or not.

Right now it doesnt take this into account and it is a very good idea. I will add it to our system and once there is some time I will look into it more.

That's good to know (And I do realize that there will be other things with a higher priority)

Would it not be an idea to do a few alternative mixes of factors and give players the option of choosing these alternative AIs on the understanding that they would be guinea pigs as a tweak intended to solve one problem might well produce an odd effect in another area.

It sounds like a great idea but to be honest, it is not as simple as you probably think. It s not a matter of few numbers to change (like from 1.2 to 1.1, etc.), it is a very complex decision making based on pretty high number of factors and weights. If there is no very detailed description of what is each of these weights for and what it affects, it would be really just a waste of time (ours and also the modders).
I guess that the AI development will probably stay on our side, we will try to promote modding and scripting more and if we succeed in that, it would already be a great step forward.

I didn't mean that you would enable players to adjust a list of factors but more that when you are able to do a tweak you issue it as a beta AI to get some early feedback on how the AI behaves in practice.

[stormbringer3]

When I cycle through units to check who needs rebuilding, if the unit is fortified it won't come up. It would be great if fortified units would also show.

[pavelk]

When I cycle through units to check who needs rebuilding, if the unit is fortified it won't come up. It would be great if fortified units would also show.

Hi stormbringer3,
we dont want the fortified units to be part of the loop (usually you fortify them for a reason) however when you use alert (alarm action) they are added to the loop. To see alert action in UI, enable "extended UI" player option.