Brad Wardell's site for talking about the customization of Windows.
Published on February 8, 2013 By Frogboy In FE Sneak Peeks

Lest you think Legendary Heroes is going to get all the fun, plain old Fallen Enchantress is getting a ton of love too.  We’re working on 1.3 and wow. I don’t want to oversell it but gameplay wise, this will be the biggest change since the pre-release builds IMO.

Where do I start? Let’s start with Diplomacy.

Diplomacy

IMO, the biggest weakness of Fallen Enchantress is the diplomacy system.  There’s still a lot of stuff that could be done here but my biggest beef has been the way the AIs interact with players and you.  One minute they’re your buddy and the next they’re declaring war.

image

Now, the schizophrenic behavior of AI players is nothing new. Heck, multiplayer games are much worse in that regard.  But it’s one area that single player games can be much better at. 

I tinkered around the edges in 1.1 and 1.2 but ultimately decided it had to be tossed out.  Sometimes, you have to nuke the system from orbit.  And so the 1.3 diplomacy system won’t really seem different at first glance but under the covers it’s a whole new world.  First, the AI no longer “calculates” relations. Instead, it fills a bucket of love (or hate) that gradually changes over time.  Because it’s now a bucket (that means your history with them matters) there’s a lot of interesting behind the scenes “stuff” I can do. 

Unfortunately, most users won’t even be aware of these changes but I think the expert players will.  For instance, AI players will be more inclined to offer peace even if they’re stronger. That’s because it’s not calculation relations, it’s looking at your history together and the history of others.  So it’ll look at what is in its best interests overall.

Smarter

The AI is quite a bit smarter.  The problem with these games is that it’s hard to have the AI be “smart” if the designer isn’t that good at the game.  I’ve gotten a lot better at the game thanks to those players who post YouTube videos of their strategies. I incorporate those into the AI (and use them myself).  For example, you very well may get creamed by the AI getting the Forge of the Overlord (spell of mastery victory) as a lot of players win that way and the AI focused too much on conquest.

Balance

We’re also doing a lot on the balance front.  One of my pet peeves is how many “empty” turns there are.  It’s a habit we brought over from GalCiv II that we’re curing (or starting to cure) with v1.3 of FE.  Things will definite “move” a bit faster.

For instance, the differences between Towns, Conclaves, and Fortresses has been significantly increased.  Towns provide a lot more gold. Conclaves a lot more research. And Fortresses a lot more production.  The AI has the advantage here because it’s able to calculate the right balance based on looking at other players (you can look at other players too but most human players don’t – but it’s fair game to the AI since we let you look at the other player stats).

World Generation

1.3 is the first version with procedural tile yield support.   That basically means that the map generator will look at a lot of other factors when deciding the tile yield of a given tile is.  The result is a lot more variance.

image

FUN!

Generally speaking, FE 1.3 is just a lot more fun.  That said, the best way to make the game more enjoyable is to get more units for each side.  The AI doesn’t design its own units. It uses what you create. The more units you give a faction, the smarter the AI gets.

Why?

So why are we putting so much effort (and hence $$$) into making Fallen Enchantress even better? Well first, it’s important that Legendary Heroes be distinct. Legendary Heroes is an expansion pack. It’s not supposed to make the game funner by “fixing” things in FE.  It’s about expanding on concepts already there and adding new ones. FE 1.3 is about taking an already fun game, taking feedback and what we’ve learned and polishing it. 

I’m pretty excited for 1.3 to get out. I don’t have a date yet. My guess is probably in 2 weeks.

Spell of Making: DEFEAT

image

Ultimately in my test game, Pariden won by casting the spell of making.  I’m requesting some tweaks so that there’s more warning when the various factions are constructing the towers leading up to it since this caught me by surprise and I only had 10 moves to do something about it.


Comments (Page 3)
7 Pages1 2 3 4 5  Last
on Feb 10, 2013

Kongdej
Quoting Chehalden, reply 30I would use my better resources to build a fleet of ships designed to counter everything he had. Since the AI focused on number of ship more than quality it was easy to build ships basically immortal and stop everything at my border while i continued my empire building. I didn't always feel like being forced to absorb more territory. Then things would go like this:

Me: Hey i've killed literally like a 100-200 ships of yours i haven't lost a single unit. you want peace?

Funny, while you are half-right, half-wrong in my opinion, I also think the mechanics are wrong when wars can be fought with no losses (it also happens in elemental, and I don't find it suiting).
Besides that, the AI SHOULD re-think the strategy, and try to either gang up on you, or sue for peace, the problem is, if you can focus research, production and economy peacefully each game you will always have a cheap win, it NEEDS to try and punish you for approaching this goal while having no military, if nothing atleast it would force you to build up a quick military before going into peace again, if you can already wipe out the enemy without much stress you basically won you are just waiting for the "win" screen.
The AI should try to figure out if opponents are trying to out-tech them and either run a tech race or a war of conquest upon those opponents, because if they are set to idle they surely will win after 50 turns of ultra-teching up to mega-destroyers or whatever they get after several turns of awesome teching

Sincerely
~ Kongdej

 

I really don't like the concept of the AI ganging up on you. Was just playing GalvcivII the other day and that very thing happened. Took a lone transport and took over a minor race. Next turn the AI has a secret meeting and determines that you're over-agressive behaviour will no longer be tolerated and EVERY major race declared war on me. . Wasn't right OR fair cause just the turn before another major AI player did the same against a minor race. Why wasn't I invited to the secret meeting for wiping HIM out?!?

While I do see the sense in ganging up to bring down the leader, it should be a progressive change. In my next game that I played (am currently playing), one of the races...Korath...was wiping out race after race. They were leaving me along cause I had a good fleet, but no one else could stand against them. It came down to the last 4 races grouping together to fight him off....they asked if I wanted to join in. THIS...is awesome and what I expect out of AI behaviour. So, I don't join in...cause they're gettting their butts kicked and Korath would do the same to me....cause they had Photon Torpedo II and I had no Missle defence...I had to rush the tech and I'm glad I did, cause they shot really far ahead all of a sudden as they ramped up production (had something to do with everyone sharing 3 unique techs)..oops.

Also had a REALLY lucky Ranger Class ship found and WOW that thing is powerful, so I declared war, splitting Korath's forces in half and the 5 of us chewed him apart.

Now they're not liking me....

Anyways, point is that AI isn't simple to code....but it has been done (in galcivII) and has been done well; there just needs to be more consistency.

Keep it up SD!

 

on Feb 10, 2013

@GFireflYE
I did not mean for every single AI opponent to gang up on you, but if you are running the tech race, and one AI see's you as a main threat but doesn't feel strong enough to beat you on his own, he should seek an alliance with 1 other AI or so, to not just give in to your racing the tech tree, and to also have a chance of winning.
The AI should NOT play distinctively different against the player compared to other AI players, and the AI should not all uniformely follow the same diplomatic code, since that would lead to big gang-ups against whoever the code says to be wary about.
I was merely saying that if an AI is getting to the extent in a war where he refuses to surrender, he should atleast seek out an ally in this war as to avoid just "losing" because the AI is being stubborn and silly, if the AI cannot find an ally in said circumstance (AI diplomacy should consider this as high an offer as if a player in a similar circumstance offered an alliance).

Seeing it from the point of view from the other side of the alliance, if he thinks you are a threat he might get some help beating you, and I think that if a player pursues pure technology, the AI shouldn't just leave you alone in a corner, because that would spell sure doom for themselves...

I do remember the AI in "Starcraft" (the first one), if it COULD freely choose alliances, all the AI players would always instantly team up right at the get-go to get the biggest chance of winning, I don't want that in Elemental, because that's just annoying to play against.

Sincerely
~ Kongdej

on Feb 10, 2013


good things to hear sounds like good progress.

On AI; my offer to help write improved ai still stands; and the biggest way to easily make significant improvements in the ai is through genetic algorithms and data mining of the games played by all the players; that allows for much better races and units quite readily.

on Feb 10, 2013

zlefin

good things to hear sounds like good progress.

On AI; my offer to help write improved ai still stands; and the biggest way to easily make significant improvements in the ai is through genetic algorithms and data mining of the games played by all the players; that allows for much better races and units quite readily.

I appreciate the offer to help.

But unless you have ever actually written a computer game AI in a shipping game I don't think you (or anyone in a similar place) is in a position to understand what it takes to write good game AI.

I've been familiar with expert systems, neural nets, and genetic algorithms for over two decades now.  I might be wrong but I'm not familiar with anyone who has been writing computer game AI as long as I have (maybe someone reading this could check).  The OS/2 version of Galactic Civilizations used to save bits of data to disk that when the player played future games the AI could gather what type of player he or she was. I even tried to implement a primitive expert system (took it out for performance reasons).

The Windows version of Galactic Civilizations I would record build order, tech tree research order, etc. that could then be used by the AI to improve on its research.  Galactic Civilizations II added the ability to run, in a thread, simulated turns (substituting computer AI for the human) to get a guess on what is likely to happen an then "react" to that future even in the present.

And while those things are useful, they are extremely time consuming to write and you can do all that only to discover that none of it matters if the player has figured out they can simply win the game by buying 20 salted porks for their mounted units and kite to victory.  Or aim straight to archers, build a large force of them (to get the larger maps) and just massacre every army they play. Or combine spamming henchmen with buying quests and build an uber army.

Game design is always a much bigger effect on how effective computer players are than any sort of theory craft on AI development.  It's all about bang for the buck.  

Your interest in AI development is commendable. I think you would be better served putting that interest to use in your own game so that you can see what works and what doesn't when it comes into contact with human beings.

on Feb 10, 2013

Frogboy
Game design is always a much bigger effect on how effective computer players are than any sort of theory craft on AI development. It's all about bang for the buck.

Dont break our dreams frogboy, deep inside we are all smarter than the ones next to us!...

That said, I do agree with this statement.

I mean, sometimes I feel like the AI could be smarter, but I know that I would not have the ability to do it myself, and looking at ALL the games out there, I guess it is rather complicated, difficult, or night impossible.
but... Shaddap and write "true" intelligence, because if you can do that, of course you would spend it on a TBS game

Oh, and this is written late at night, so please read it with a grin on your face

Sincerely
~ Kongdej

on Feb 10, 2013

Kongdej
@GFireflYE
I did not mean for every single AI opponent to gang up on you, but if you are running the tech race, and one AI see's you as a main threat but doesn't feel strong enough to beat you on his own, he should seek an alliance with 1 other AI or so, to not just give in to your racing the tech tree, and to also have a chance of winning.
The AI should NOT play distinctively different against the player compared to other AI players, and the AI should not all uniformely follow the same diplomatic code, since that would lead to big gang-ups against whoever the code says to be wary about.
I was merely saying that if an AI is getting to the extent in a war where he refuses to surrender, he should atleast seek out an ally in this war as to avoid just "losing" because the AI is being stubborn and silly, if the AI cannot find an ally in said circumstance (AI diplomacy should consider this as high an offer as if a player in a similar circumstance offered an alliance).

Seeing it from the point of view from the other side of the alliance, if he thinks you are a threat he might get some help beating you, and I think that if a player pursues pure technology, the AI shouldn't just leave you alone in a corner, because that would spell sure doom for themselves...

I do remember the AI in "Starcraft" (the first one), if it COULD freely choose alliances, all the AI players would always instantly team up right at the get-go to get the biggest chance of winning, I don't want that in Elemental, because that's just annoying to play against.

Sincerely
~ Kongdej

True dat. I think we're saying the same thing....just from opposite sides of the coin.

on Feb 10, 2013

Frogboy


I've been familiar with expert systems, neural nets, and genetic algorithms for over two decades now. I might be wrong but I'm not familiar with anyone who has been writing computer game AI as long as I have (maybe someone reading this could check). The OS/2 version of Galactic Civilizations used to save bits of data to disk that when the player played future games the AI could gather what type of player he or she was. I even tried to implement a primitive expert system (took it out for performance reasons).

 

expert system= saving those bits of data to disk?

 

I really think the AI in a game like this does need to be partially adaptive (but randomly switch things up)   Why is it such a performance drain?  Would this be something 64-bit would allow?

 

I kinda wish LH was 64-bit, or that you'd do a DLC that made it 64-bit with improvements as an experiment.  Might be too expensive of an expeiment, especially since you have a 2nd project that's coming down the pike (which I'm hoping is 64-bit)

 

 

Sounds like some of the issue is henchmen being not completely designed, or mass archery being too powerful- I know you did something in LH to counter mass archery.

 

The quick battles are a major problem- you really need to copy what AOW3 is claiming they're doing, which is to make quick battles tactical battles that run in the background, even if it slows things down some.  It's bad when you can cheese dark wizards that way.

 

 

 

on Feb 10, 2013

Alstein

Quoting Frogboy, reply 35

I've been familiar with expert systems, neural nets, and genetic algorithms for over two decades now. I might be wrong but I'm not familiar with anyone who has been writing computer game AI as long as I have (maybe someone reading this could check). The OS/2 version of Galactic Civilizations used to save bits of data to disk that when the player played future games the AI could gather what type of player he or she was. I even tried to implement a primitive expert system (took it out for performance reasons).

 

expert system= saving those bits of data to disk?

 

I really think the AI in a game like this does need to be partially adaptive (but randomly switch things up)   Why is it such a performance drain?  Would this be something 64-bit would allow?

 

I kinda wish LH was 64-bit, or that you'd do a DLC that made it 64-bit with improvements as an experiment.  Might be too expensive of an expeiment, especially since you have a 2nd project that's coming down the pike (which I'm hoping is 64-bit)

 

 

I could write one now without much of a hit. You could even have the game upload user strategies and have it analysed to come up with much better city improvement patterns, unit designs, tech research.  The problem is time.  It' takes a lot of work to do that and it wouldn't address the ways most players win the game.

people best the ai by playing the game in interesting ways that we didn't imagine.

I have a great deal of appreciation and gratitude to anyone willing to volunteer their time to help. It makes working on the game worth it. The computer opponents will get better over time but it won't be due to implementing advanced artificial intelligence techniques.

on Feb 11, 2013

Frogboy
people best the ai by playing the game in interesting ways that we didn't imagine.

Fun fact about Meeee. I prefer to play games like this
going at each different mechanic from an odd angle.

I still am excited about the range in LH, not because I like the melee part, but because I don't like the random part, it would be nice to see the current FE AI try to use melee units with the "Charge" trait when the enemy go out with many archers (that is, the AI should favour high-move units, or other "counters" like high-armour, or archer units themselves)
FE tactical is at its worst AI wise due to heroes being such a major factor, but only if levelled correctly... Many many things I would see fine tuning could help upon. (like derek working to fine-tune the dodge bonusses and penalties)

Sincerely
~ Kongdej

on Feb 11, 2013

Frogboy
Game design is always a much bigger effect on how effective computer players are than any sort of theory craft on AI development.

So true...
I've made a mod for Civ 3 to improve the AI without ever touching the actual AI itself.
I analyzed what parts of the game the AI has trouble with and changed them.

Here's some examples:

AI had serious trouble with the corruption and properly fighting it.
=> I removed corruption.

AI for some reason hardly built factories.
=> I found out that the pollution was the cause to make it a no-go for them and thus removed it.

AI would try to build wonders in very underdeveloped cities and thus waste massive amount of production that they could have used to develop their empires.
=> I made a new building "Forge" with the same effect as in Civ 4 and made the Wonders require it as prerequisite. So AIs would only build Wonders in good productive cities.
I also made later wonders require factories.

AI would build way too many defensive units.
=> Changed the stats for most units to be good at defense and offense and all gave them the "offensive Unit"-Flag.

AI handled Units with only 1 move very poorly.
=> Gave every unit at least 2 move-points.

AI would build way too many Jet-Fighters.
=> Removed their ability to bombard which drastically reduced the value for the AI because of an ability they weren't good at anyways.

AI often build ineffective field-improvements.
=> Made it so that every field allows only one type of improvement so that it was impossible to pick the less effective one.

These changes made the game significantly harder!

on Feb 11, 2013

I like the steps forward. You guys are listening, and it makes sense. 

Though, are you stopping DLC's? I think some spell based (adding spells to the game) ones with scenarios/maps could have a lot of potential. I would pay just for more diverse/interesting spells, and I think most would if you add scenarios as well in the DLC. 

And finally, please add Legendary starts, or customizable starts. It would go a long way to make going "tall" a good option, not counting the xpac changes, and everybody loves a powerful capital. (atm, capital is a mediocre city at best, no benefits at all)

on Feb 11, 2013

Edwin99

I like the changes that you are making to further differentiate - conclaves, fortresses and towns as currently I find that I can with just Fortresses, and looking forward to the AI changes.

 

I would like some differentiation, though one bit of differentiation I'd like to see disappear is limiting walls to fortresses. 

 

I'm also curious about the status of the other DLC's- are they being rolled into LH, or are they still in production?  I was really interested in more items/more quests.

 

 

on Feb 11, 2013

Kongdej

Quoting Frogboy, reply 39people best the ai by playing the game in interesting ways that we didn't imagine.

Fun fact about Meeee. I prefer to play games like this
going at each different mechanic from an odd angle.

I still am excited about the range in LH, not because I like the melee part, but because I don't like the random part, it would be nice to see the current FE AI try to use melee units with the "Charge" trait when the enemy go out with many archers (that is, the AI should favour high-move units, or other "counters" like high-armour, or archer units themselves)
FE tactical is at its worst AI wise due to heroes being such a major factor, but only if levelled correctly... Many many things I would see fine tuning could help upon. (like derek working to fine-tune the dodge bonusses and penalties)

Sincerely
~ Kongdej

Ultimately, the way this sort of AI needs to be done is to tie the AI into the object itself via scripting. 

So for instance, every spell should have some Python/Lua script attached to it that tells the game how "valuable" that spell would be given the conditions fed into it.  Unfortunately, we don't have that kind of system in FE/LH.  But it's the direction we want to take things in.

on Feb 11, 2013

sjaminei
I like the steps forward. You guys are listening, and it makes sense. 

Though, are you stopping DLC's? I think some spell based (adding spells to the game) ones with scenarios/maps could have a lot of potential. I would pay just for more diverse/interesting spells, and I think most would if you add scenarios as well in the DLC. 

And finally, please add Legendary starts, or customizable starts. It would go a long way to make going "tall" a good option, not counting the xpac changes, and everybody loves a powerful capital. (atm, capital is a mediocre city at best, no benefits at all)

We plan to do more DLC.  Here's some of the DLC I woulld like to see made:

  • Spell Pack (bunch of magical spells)
  • Quest Pack (bunch of new quests and events)
  • Loot Pack (bunch of new loot and equipment)

They're a bit time consuming to do but I think a lot of people would appreciate what they do AND understand why they're DLC and not part of the base game (which is half the battle on DLC -- you dont' want people thinking that the developers just cheaped out).

on Feb 11, 2013

Frogboy
Unfortunately, we don't have that kind of system in FE/LH. But it's the direction we want to take things in.

As an update to FE/LH or as a part of a future product in the FE line?

7 Pages1 2 3 4 5  Last