Ninakoru wrote: 

I think the AI needs still lots of work. I didn't even bother about it because I expected to be lots of changes in the AI department.

I usually play on impossible and I find it not difficult to win. I needs still lots of refinement, I have high hopes for it, as many players consider AI to be crucial in these games.

Out of curiosity how did you win most of your games?

I play on endless. I win the game mainly throw finding a solid weapon early on, through ananomolies and after that I just conquer the home planets of all races.

I play united empire or voydani.

wilbefast wrote:

Some good data here, thanks  Though do bear in mind that tweaking and balancing is still ongoing, so this may not be true at release

What I'm really curious about though is whether you unlock new hulls ASAP, regardless of your production of strategic resources, or whether you wait until you've secured a source of the required strategic resources first. I ask because the AI currently will not include a design in its "task force composition" if said design requires any resource that the AI is not yet producing. This prevents the AI from getting into dead-locks trying to build designs that it can't afford, but could also creates situations where the AI does not build craft it can afford (based on stock or the marketplace) simply because it doesn't own a system which produces one of the resources it needs. But this is only a blind-spot when the AI has enough money to buy the required resource from the market or if they have a massive stock of the required resource already. hence my question: how often do you, as players, build craft that you can only afford temporarily thanks to a surplus of dust or resources in stock?

I will unlock new hulls as soon as practical, with consideration to the ratio of expansion to defense. 

I will make sure that it doesn't interfere with my expansion, and will try not to expand further than I can defend, but I don't consider my resource stockpiles when deciding whether or not to unlock the tech. I can get the resources through the market or quests, that's not something to be concerned about.

What I'm really curious about though is whether you unlock new hulls ASAP, regardless of your production of strategic resources, or whether you wait until you've secured a source of the required strategic resources first.

To get the technolgy of medium hulls, and to get the needed ressources for it, are two very independant and changing timings. So i work independently on both, without trying to make it coincide. I don't wait to have ressources to go for the tech. I don't wait to have the tech to try to get ressources. I try a lot to get everything i need to build the ships, and i build it asap. The only condtion I give to me to search for medium hulls is to have already a good economy (industry/science/dust), because the investment and the ships are very costly. Most of the time the medium hulls are the 1st T3 tech i search for.

how often do you, as players, build craft that you can only afford temporarily thanks to a surplus of dust or resources in stock?

That's heavily depend of the military pressure i have to endure. If i don't feel safe, i improvise units craft only destined to be as powerfull as possible spending all ressources i have. If i feel safe, i keep the ressources, maybe I will find a good way to use it later.

Also, I most of the time build two versions of the same unit to solve this kind of question : 

1. A coslty version, that need a lot of ressources i produce, and very strong. Those units are destined to be the elites and the main strengh of my army, and i produce it as much as i can.
2. A cheap and weaker version, that don't need any strategic ressources (or only ressoucres i possess almost infinitly). I produce it when i cannot build more elite units but i still need to increase my military power.

Maybe this is a idea to unlock IA's units production. 

I made that everygame in endless legend, and it oppend a lot of cool strategies ! 

With forgotten, the elites units are mostly invaders lead by the strongest heroes, and the cheap units are raiders, explorers, and sentinels, giving vision in order to open ways for spies.

With wild walkers, elites armies are the main and unbreakable military power, meanwhile cheap units are mostly hunters, destined to die in fight but not without making a lot of damage.

With necrophages, the main horde is destined to captures strongest cities, meanwhile cheap units are super mobiles and can swarm and sack the lands, and capture little towns

Etc ! 

I didnt check in the AI tool, but I play on endless and every planet I have conquer was on Level 1.

How to encode the stradegy?

I think the answer gonna simple, but I think if you do a quick interview with the player, which sent you the save game. He/ She will be able to explain his/her goals. I think its important to solve the main Quest of your race and at the first 20 turns or at least set the pace for achieving those. 

I might be wrong, but I think there is defentily a solid built order, which will work on every position in the galaxy. 

I mainly play united empire. For example I always start with +10 science and +10 Dust building --> Scout with 4 probes --> Settler or the production building you get from the xenology tech. 

So my decision tree just starts at turn 7. 

To be honest I have no clue about AI develpoment, but I think the Priotiries for the UE for the first 20 turn should be simple.

1. colonize

2. try to solve the main quest 1 and 2 until turn 20.

3. scout

4. if enough ressources avaible go for Settlements Level 2 Science. 

I could be totally wrong, but my hypothesis is that for the first 20 turns or mabye just 15, there is no AI needed and I think the gameplay doenst differ to much at those turns even by human players.

to be honest I think the starting strategy is mostly hardcoded, if you are human player or an AI. As human player you will adapt your starting stradegy over time, but you will find an optimal goal, which you adjust, if outside factors influence the stradegy. Like pirates or not a lot of worlds to colonize.

And I just mean like for the first 15-20 rounds. 

Negative for that kind of Approach is: for every patch or update, the starting stradegy has te be reviewed.  

Ifoldforweed wrote:

I played a bit on the weekend.

I have noticed the AI doenst upgrade there Settlements, they stay at level 1, which is a huge disadvantage.

Wouldnt it be possible to take that into their core priorities? an maybe even buy ressources on the open market!

other questions would it be possible to take some save games of succesful players and take like 2-3 core starting startegys for the AI? we humans dont do any different. surely we adjust the early game stradegy, but I think most players have a set of early built orders from which they wont diviate to much?

While I would be wary of doing this as a hard coded behavior, I do like the idea of AI learning things from human players. For instance, in Fallen Enchantress: Legendary Heroes the AI will actually use your unit designs from past games in their own armies. Having some sort of mechanism in the game where the AI take some cues from successful human players would make things interesting. I just don't know who extensive that system could reasonably be made. 

I played a bit on the weekend.

I have noticed the AI doenst upgrade there Settlements, they stay at level 1, which is a huge disadvantage.

Wouldnt it be possible to take that into their core priorities? an maybe even buy ressources on the open market!

other questions would it be possible to take some save games of succesful players and take like 2-3 core starting startegys for the AI? we humans dont do any different. surely we adjust the early game stradegy, but I think most players have a set of early built orders from which they wont diviate to much?

SuperMarloWorld wrote:

By avoid do you mean retreat? We can look at the retreat heuristics perhaps...

If am not wrong, IA don't engage fight, but retreat if the opponent engage it. I am realising I was not clear at all on this point, because IA don't look to try to avoid stronger enemy fleet "on the map", but refuse fight by retreating. 

Hmmm... yeah we don't have any specific fleet-avoidance behaviours, so a weaker AI fleet will currently just ignore strong enemy fleets. However the AI shouldn't send inferior fleets until enemy territory where they will have no hope of fighting someone at their power level.

I'm not sure what you mean by rebooted  

I mean they "won" the colonization phase. By "rebooted" i mean they send me back to my system, me and my vainglory, so i have to take what i deserve by force and not by cold ward and settlers anymore  I don"t know why i was sure this word mean defeated/repulsed. 

Rebuté in French maybe? I get it - military should have an up in the next update, again it's one of the main priorities. Hopefully it'll be enough to keep the AI on par in terms of strength as well as economy.

For these kinds of rare issues only saves are going to help I'm afraid - I can try to guess what the AI is doing but at least 70% of the time these guesses will be wrong.

My bad. I need to get the reflex to save games when i see something very weird. 

Yes please sir! We do love saves

NagohShan wrote:

wilbefast wrote:

There were changes to the generation of strategic resources lately which penalised the AI in various ways, but notably preventrf them from unlocking better hulls until quite like as the AI wants to secure access to all a hull's necessary resources before unlocking it. I'm actually curious how many here play this way: do you ever source strategics purely from the market and quests? That is, do you ever unlock hulls based on your current stock of resources rather than your production?

The stronger hulls are just too good, so there's no better use for your strategics at the moment. If I have no access to antimatter or adamantium generation, but I have stockpiles, I'm still going to build the hulls with some upgrades, because a 12 CP flotilla of T2's will wipe a 22 CP flotilla of T1's comfortably. T1 hulls become obsolete for anything other than filling the last CP of a flotilla that isn't divisable by 3. Which is unfortunate.

Three Carriers, a T2, and a T1 feels like the objective best flotilla composition. Earlier Hulls just fall off too hard due to lack of modules and stats. The AI would feel much more difficult if they didn't lose out of Tech obsolescence. 

I love that there is progression in the sense that your strategics buy better hulls. Honestly strategics are used better in this game than EL all around, but the earlier hulls are just too weak. If the mutually exclusive upgrade techs for the T1 ships added a strategic cost to the T1s, but upgraded both, and the mutually exclusive choice was if the strategic cost was Titanium or hyperium, that would provide a hook to keep the ships relevant. Pick the Titanium option for a defensive skewed stat boost to the hulls or the Hyperium for more of an aggressive option.

Some good data here, thanks  Though do bear in mind that tweaking and balancing is still ongoing, so this may not be true at release

What I'm really curious about though is whether you unlock new hulls ASAP, regardless of your production of strategic resources, or whether you wait until you've secured a source of the required strategic resources first. I ask because the AI currently will not include a design in its "task force composition" if said design requires any resource that the AI is not yet producing. This prevents the AI from getting into dead-locks trying to build designs that it can't afford, but could also creates situations where the AI does not build craft it can afford (based on stock or the marketplace) simply because it doesn't own a system which produces one of the resources it needs. But this is only a blind-spot when the AI has enough money to buy the required resource from the market or if they have a massive stock of the required resource already. hence my question: how often do you, as players, build craft that you can only afford temporarily thanks to a surplus of dust or resources in stock?

Ninakoru wrote:

I think the AI needs still lots of work. I didn't even bother about it because I expected to be lots of changes in the AI department.

I usually play on impossible and I find it not difficult to win. I needs still lots of refinement, I have high hopes for it, as many players consider AI to be crucial in these games.

Out of curiosity how did you win most of your games?

I think AI seems dumb because players will always out-tech their ships.  Once AI either pushes to better ships, or teir 1 ships become a threat then the AI will feel smarter because players wouldn't be able to wade through an entire empire's military with a single flotilla.

Yeah, we have a couple of bug reports from QA that we're working on where hulls are concerned - we're aware of the issue.

Buzzkillington1990 wrote:

Bit of an anecdote, but I think the AI is unable to play Vodyani on the lowest difficulty setting.

I was playing a VERY casual game, going for 200+ turns... and the Vodyani had not expanded beyond their Homeworld. Even if it's the absolutely weakest AI setting, that seems silly.

Hmm... Yeah we have regression issues with this weirdos every once and a while - at once stage they would sometimes undock and then just sit there. I'll be sure to included them in all my test-runs from now until release.

Buzzkillington1990 wrote:

Bit of an anecdote, but I think the AI is unable to play Vodyani on the lowest difficulty setting.

I was playing a VERY casual game, going for 200+ turns... and the Vodyani had not expanded beyond their Homeworld. Even if it's the absolutely weakest AI setting, that seems silly.

Same as higher difficulty level for my part (maybe it has changed since last patch). I remove Vodyanis from my IAs games for the moment. I guess it will be fixed at the release. I personnaly hope they will work a lot of vodyanis beaucause they are absolutly wonderful but don"t look finished at all even in humans hands. 

Bit of an anecdote, but I think the AI is unable to play Vodyani on the lowest difficulty setting.

I was playing a VERY casual game, going for 200+ turns... and the Vodyani had not expanded beyond their Homeworld. Even if it's the absolutely weakest AI setting, that seems silly.

Well I'm a little bit more optimistic about the AI now and hopefully it will just get better from here. Thanks to Wilbefast and everyone else who answered.

wilbefast wrote:

Out of curiosity how did you win most of your games?

Most of the time with economic victory, and not going specially for it. Once the trade routes start rolling with some bonuses it goes out of control by itself. Anyways by the time I win I'm usually 1st in most of the scores.

I'm not specially aggressive until I need to expand further or I'm being pressed by a close neighbor. I think I played with all races in impossible saving united empire. Not based on exploits or anything, just trying to take the best decissions all the way.

wilbefast wrote:

There were changes to the generation of strategic resources lately which penalised the AI in various ways, but notably preventrf them from unlocking better hulls until quite like as the AI wants to secure access to all a hull's necessary resources before unlocking it. I'm actually curious how many here play this way: do you ever source strategics purely from the market and quests? That is, do you ever unlock hulls based on your current stock of resources rather than your production?

The stronger hulls are just too good, so there's no better use for your strategics at the moment. If I have no access to antimatter or adamantium generation, but I have stockpiles, I'm still going to build the hulls with some upgrades, because a 12 CP flotilla of T2's will wipe a 22 CP flotilla of T1's comfortably. T1 hulls become obsolete for anything other than filling the last CP of a flotilla that isn't divisable by 3. Which is unfortunate.

Three Carriers, a T2, and a T1 feels like the objective best flotilla composition. Earlier Hulls just fall off too hard due to lack of modules and stats. The AI would feel much more difficult if they didn't lose out of Tech obsolescence. 

I love that there is progression in the sense that your strategics buy better hulls. Honestly strategics are used better in this game than EL all around, but the earlier hulls are just too weak. If the mutually exclusive upgrade techs for the T1 ships added a strategic cost to the T1s, but upgraded both, and the mutually exclusive choice was if the strategic cost was Titanium or hyperium, that would provide a hook to keep the ships relevant. Pick the Titanium option for a defensive skewed stat boost to the hulls or the Hyperium for more of an aggressive option.

Ninakoru wrote:

I think the AI needs still lots of work. I didn't even bother about it because I expected to be lots of changes in the AI department.

I usually play on impossible and I find it not difficult to win. I needs still lots of refinement, I have high hopes for it, as many players consider AI to be crucial in these games.

Out of curiosity how did you win most of your games?

I think AI seems dumb because players will always out-tech their ships.  Once AI either pushes to better ships, or teir 1 ships become a threat then the AI will feel smarter because players wouldn't be able to wade through an entire empire's military with a single flotilla.

So I was wondering how effective the AI is considering that ES2 was built from the ground up taking AI into account?

This was a lesson learned from EL so I'm hopeful for a strong AI even on default difficulty.

Slaunyeh wrote:

I was also referringto the Strategic aspect of the game.

But i agree with you.

Update.

I finished my impossible game.

diference in AI. 

It build bigger ships.

It builds more ships

It tries unsuccefully to find ship build to counter your ship config

period.

Nothing new in this galaxy.

What i noted as far as AI goes:

Passive AI,

defensive AI

Expansive AI

Even on high dificulty AI will only bother you IF you are way weaker than it.

Ai is unable to form alliance

Ai is unable to vulture you or another AI(when you start crumbling under an assault, no Other Ai try to take some piece of you.)

haven t seen Alliance AI defend you. (maybe not needed in my games.) It seem logical as Diplomacy is... quite restrictive.

Sorry if I didn't fully understand what you meant. But thanks for the update. And sadly yes...that seems kinda meh for AI efforts. Looks like I'll be putting this on hold for a while.

Slashman wrote:

Slaunyeh wrote:

The Ai has its share of problem. 

But i would say you can t really judge until you play impossible and beyond.

I'd say that's highly inaccurate. When you move to the point where the AI is getting massive bonuses and thus have to do very little to maintain their economy, you have ceased to really play against the AI and are palyin against bonuses. Unless the AI behaves significantly differently at higher levels algorithmically, then you can certainly judge when playing at lower levels. 

I never had to play EL or ES1 at impossible to find out where AI weakness is...quite the opposite.

I was also referringto the Strategic aspect of the game.

But i agree with you.

Update.

I finished my impossible game.

diference in AI. 

It build bigger ships.

It builds more ships

It tries unsuccefully to find ship build to counter your ship config

period.

Nothing new in this galaxy.

What i noted as far as AI goes:

Passive AI,

defensive AI

Expansive AI

Even on high dificulty AI will only bother you IF you are way weaker than it.

Ai is unable to form alliance

Ai is unable to vulture you or another AI(when you start crumbling under an assault, no Other Ai try to take some piece of you.)

haven t seen Alliance AI defend you. (maybe not needed in my games.) It seem logical as Diplomacy is... quite restrictive.