Hello again: there is point about IA i spotted recently.



Sometime an IA's economy/army just snowball, and game become unplayable. The Ia instantly get a giant army and is impossible to stop. All others IA fear them and its impossible to destroy it alone. If this type of IA spawn far in the map, game is done when it start.



Maybe there must add a "stop" on IA snowball, or allow IAs to make an "non official" alliance when an player become too military strong ?



(I was implied in this situation in a multipalyer game. We make a 4 player alliance against a BIG GIANT CRAZY necrophage IA empire. I was on a front line, most of my town fallen, and just a little army of epic Broken Lords defended and avenged their lost lands with the help of the alliance (drakkens, Ardents mages and roving clans). Oh Boy ! That was SO COOL ! That was like play a entire campaign ! )



(I play in serious mode)

I really would like to see some dev feedback on the feasibility of some of the suggested changes. Some ideas might not be applicable, depending on how the ai is coded. Isn't a conversation between players and devs the whole point of G2G?

Diplomacy Suggestion:

Lower the values for some trade items.



O how Necros love their spice.

20 Spices to get a Truce against a stronger opponent AND 2 cities AND ~10 techs AND 60 Titanium? YES PLEASE.

This

Also, the AI never really use the market making some of Roving Clans perks useless



I once played a game over 200 rounds, and in total there had been 10 (approx.) transactions not made by me. There were 5 AI players. I was playing as Roving Clans. It kinda sucked.

Hi

I played the game 10 rounds with easy/normal around 4 AIs recently. It seems that the AI is very aggressive. Most of the time, if the AI start next to each other 1-2 AIs die before turn 20 (especially Drakken never survive).

In one game the AI sieged me at turn ~50. Unlucky for me they had the exact counterunits against mine. With 8 units and + defense upgrades from the first Era i killed the army of 4 + their hero. After that it was quite easy to raid the AI back.



I would like to see an game option at the start of a game to choose the behaviour from the AI - like passive / aggressive or turtle /expand /trading. Or changethe AIs behaviour to be more defensive. Its a bit annoying if 1-2 AIs die every game before turn 20.

But still a good game and i enjoy it

greetings Theraton

Nasarog wrote:
I'm all for the VIPs doing their thing, but sometimes the greater numbers of the larger gaming community notices things that the VIPs don't.

I just wanted to quickly chime in and point out that the VIPs did not decide on or brainstorm for any g2g vote content, ever. That's the devs' stuff, only.



Otherwise, I agree that having a larger dataset from many players is the way to go. The VIPs should test new systems the devs are not yet sure of, as they have done in the past and hopefully will in the future. (Or in plain speech: We get the most buggy builds of them all. )

Army composition could also help a lot to give more teeth to the AI. Look at my sig to get a new file of armycomposition in my AI mod.

Another thing : the defense value of cities is too low. I changed them up to 1000 def and I can say that you will need to siege before attacking a city. And that you won't be able to take multiple cities if AI uses the defense buildings because they'll have a lot more time to send reinforcements.



Another thing in diplomacy that is interesting : I gave AIs more personnalities and that gives them more flesh in what they value most or not. They really feel different when you try to guess what they "need"/"like".

What would be cool is to NOT have the visual meter of their agreement, but get it only once to have already made a deal with them that used th eitems you put in the deal.

For instance :

You need some glassteel, you propose some palladian. But you don't know if it'll succeed. You try and it fails. But the next time you'll try you'll see a "rough" estimation of the possible overcome of the deal. the more you deal palladian/glassteel with them, the mor eprecise will be your guessing. At the moment it's too easy to just get exactly what you need and I think you lose the "bargaining" feel of market.

Some very basic ideas for targeting rules:

• Prioritise units with high damage and low durability. Basically targeting priority = damage/durability. This automatically makes damaged enemies of higher priority.
• Prioritise units that didn't move yet vs those that already did if a) you are able to eliminate them before they move b) if you're melee you can also prioritise the units that have dangerous abilities like circle attack so they don't get to use it.
• Prioritise units you can attack without exposing your units - if attacking unit 1 leaves you open to attack by more units than after attacking unit 2 unit 2 gets higher priority
  
  

Exact weight of these factors should be extensively tested.



A bit more advanced empirical model (still very simple) based on this:

• Each unit is assigned a threatLevel. Base value = damage/durability + abilityMods
• Threat level is decreased if it is not able to reach and attack a unit on its turn
• Threat level is decreased if it has already moved
• Attacking a unit with offensive abilities before it moved decreases its threat_level by abilityMods
• Threat level of a dead unit is 0 (obviously)
• Overall algorithm then should then consider your whole army moves combined(not just unit by unit) to minimise threatLevel
  
  

This is still very simplistic and does not account for stuff like terrain, morale, pathing and possible enemy counter moves, but it's hard fro me to suggest something for covering these without actually trying it out and I do understand that it also should run in reasonable time .



P.S. It would be very cool if there was a battle mode simulator which you could plug your own AI into and test it vs players and other AI. (Something like Google ant challenge or similar AI competitions). I think AI would be then fixed in no time .

Interesting. We need more of these, like 3 or 26 weekly. This is how we get the info to you without airing our dirty laundry for all to see.

We could do both you know. Also, I think this is a great idea! Maybe you could have a thread after this is done to display the results, people's common problems and solutions, and then perhaps how you plan to deal with it? As Nasarog said, this could definitely be expanded to more varied polls, including possible expansion material, balancing issues, and more.

Nasarog wrote:
Interesting. We need more of these, like 3 or 26 weekly. This is how we get the info to you without airing our dirty laundry for all to see.

I think simple statistics gathering can be done in the game client and then sent over to the devs. Personally I wouldn't have any concerns over sharing gameplay data.

the way g2g works now it feels mostly like a gimmick... a real g2g would handle this... the community manager scours the forums once an hour and cuts and pastes the hot button topics, and then the devs look at them once every few days and turn it into some kind of a g2g vote or poll or something..



I know we gamers can be fickle, but if enough people say the same thing over and over, there is usually something behind it.



I'm all for the VIPs doing their thing, but sometimes the greater numbers of the larger gaming community notices things that the VIPs don't.

Well, things like:

• % of wins with each race per difficulty
• How often each hero class/specific hero is used as governor/general per difficulty
• How often certain MF is assimilated with each race per difficulty
• How often certain unit is built per difficulty
• A large portion of questions on this poll
• etc...
  
  

Can be processed without any polls and at a much larger scale with simple in app statistics tracking and would tell a lot about balance if you take data for higher difficulties and MP.



I agree that g2g could be somewhat more involved with the community.

MANoob wrote:
Well, things like:

• % of wins with each race per difficulty
• How often each hero class/specific hero is used as governor/general per difficulty
• How often certain MF is assimilated with each race per difficulty
• How often certain unit is built per difficulty
• A large portion of questions on this poll
• etc...
  
  

Can be processed without any polls and at a much larger scale with simple in app statistics tracking and would tell a lot about balance if you take data for higher difficulties and MP.



I agree that g2g could be somewhat more involved with the community.

Yes to all, except that isn't the case at the moment.



I haven't played too many games since I play each game to completion and really try to play each faction to the end.

Nasarog wrote:
Yes to all, except that isn't the case at the moment.

Well maybe it can change depending on whether the devs think it's worth it. It would be also much more useful than polls in many aspects since it will give more data and more insights. Open polls are often voted on by newbie players who have probably only seen a small percentage of the game features. This is not reflected in the poll but it can be reflected in game data. I can't wait for Ardent Mages and Wild Walkers units getting buffed based on the results of the military poll. 5 range rangers and whole map aoe for zealots maybe?

MANoob wrote:
I can't wait for Ardent Mages and Wild Walkers units getting buffed based on the results of the military poll. 5 range rangers and whole map aoe for zealots maybe?

4 should be max, and only if firing from the woods. That's how I'd do it.



Don't even get me started on the zealots.

Personally I find AI personalities that are just a bit on the wack side are a lot more fun to play against. Better to double the strengths of having AI then try to make them more player-ish; it's impossible to emulate humans afterall.

Hello everyone!



As we said in our recent dev blog post "Endless Legend Post Launch Plan", we want to improve the AI. And for this, we need your help! We would like to gather more feedback about the AI in different category, so we have created surveys:

• War Management - Defense
  
  
• War Management - Attack
  
  
• Diplomacy - Economical deals
  
  
• Battle - Targeting
  
  

Feel free to develop your thoughts and your feedback about the surveys in this thread.



Thank you a lot for your participation!





We will also publish the results of the surveys.

I feel like I'm cheating every time that I surprise attack an AI by walking an army right up to their city before declaring war. I realize that the AI is suspicious because they usually (but not always) send me a warning like, "are you here as tourists?"



I really feel like they should take armies seriously, depending on how many of your units are in their territory. Without even warning you, they should close their borders on you to kick you out. Consider how Civ 5 handles borders. You can't even move over a border without declaring war or making an open-border treaty.

IonDragonX wrote:
I feel like I'm cheating every time that I surprise attack an AI by walking an army right up to their city before declaring war. I realize that the AI is suspicious because they usually (but not always) send me a warning like, "are you here as tourists?"



I really feel like they should take armies seriously, depending on how many of your units are in their territory. Without even warning you, they should close their borders on you to kick you out. Consider how Civ 5 handles borders. You can't even move over a border without declaring war or making an open-border treaty.

They have a question that I think is a warning that they see what you're doing, but it needs to be more stern than just that. When I do that, the next turn they usually close their borders.