Tactical battles look really fun!

Maybe it is possible to emulate human thinking with probabilities? Order the list of decisions based on what is good for the AI (make some sort of metric for this). When the computer orders this list based on your metric, and assigns a number to each option, normalize it so they become probabilities. You would get something likes "Defend city from strong player town: 60%,Attack weakest city : 30%, others: 10%". Then make the computer randomly decide upon the top two, based on their probabilities. This might emulate the decision intuition that the human mind has. Maybe make a threshold, so that if the first option is 90%, the rest of the other options aren't chosen.

The difficulty with probabilities is you have to have some randomization of the probabilities from game to game or the AI remains predictable.

Also Brad doesn't like scripting in AI's.  (I think scripting makes some sense in 100% no-brainer situations, but it does hurt modding)

What time does Game #30 start?

There's a catch to that. If starting locations are very valuable, then it adds more emphasis on rushing. You'd need that extra bonus for yourself if you want to win.

I'm all for giving extra sight distance or other invisible bonuses. I wouldn't give the AI full access to the map, because that could lead to non-fun events. It's nice to meet AI explorers instead of meeting a huge army on the doorstep of the single colony you left with weak defenses at your rear areas.

That's ok on the hardest difficulty level imo. I want FE to kick my ass! Agree on lower difficulty levels players would not want that...

Thanks for the update.

2 Questions:

Question - Is there a single AI or will AI players have a chance to use a different AI each game? I.e. AI 1 and AI 2 use the standard melee focused AI, while AI 3 is guided by a magic focused AI in game two. In game 2; AI 1 follows a mount strategy AI while AI 2 and AI 3 use the standard melee AI.

Question - Will the player be able to design new fraction (Sovereign) specific AIs?

The game is looking really nice. Thanks for sharing the development process with us like this.

My BIG concern is still the bloody balance. The champions and sov's look like insta-win all over again. Please please please make their effect important by all means - but not game-breaking which is what it looks like now. I would have thought a boost to adjacent troops would be good. And by boost I'm talking 10 to 20 percent - not 300 percent which is what would have happened in WoM. But having champions destroying whole armies is just going to kill this game before it even begins. What is the point of recruiting troops if champions can dish out that sort of damage?

As I said at the start, the game itself is looking good but seeing this imbalance is really leaving me worried - very worried! You said in a recent post that one of the big lessons you have learned is that 'less is more'. Please take heed of that with regards champions. It doesn't look like the 'less is more' has made it to champions and sovereigns.

Really hope I'm wrong about this.

Historically, if an army commander/king dies untrained levies tended to break and run. If a sovereign falls in battle will his troops be affected? Will they panic, will they run, will their combat strength be affected or will their initiative be impaired for the duration of the battle?

@Das - Like I said in the blog, I felt champions / sovereigns were OP'd.

The issue IMO is not so much how much damage they can do but more like how much damage they take.

The cause is pretty obvious - lots of level up upgrades include constitution bonuses which are just too powerful.

BTW, in November, the issue was the opposite - the champions were pointless.  It's just a matter of balance. That's why we iterate and iterate.

Gonna take me a bit of time to put in all the changes.

There's still so much to do.  One of the biggest changes to FE over anything I've worked on before is the sheer number of treaties.  Rather than monkeying around endlessly with AI on "trading" I've instead put that time into coming up with treaties that players can have with one another.  (and before anyone ask, no, I'm not going to screw around with resource trading or tech trading, it ain't  happening, it's a black hole that could be spent on lots of interesting diplomacy stuff and smarter AI overall).

Having lots of different treaties in the game will give players an incentive to be on good terms with other players as well as allow other players to effectively gang up on whoever they consider a huge threat.

Thanks for the clarification, Brad.

One thing I don't understand is how you can develop and reach conclusions about the AI before all the AI is actually working and hooked up (AITraits)?

The traits are likely weights that make one type of AI more likely to choose one option in front of the others. What he's developing is the basic AI from which the other AIs will come. So rather than making 10 different AIs, he's making 1 Ai that can act in 10 different ways.

Hi AlLan!

Definitely want to get that ass-kicking from the AI too 

I really appreciate seeing the time and effort Brad is able to commit to the game. Makes me feel warm inside.. probably the fever starting to rise already.. I still haven't even got the game and the disease is hitting, omg it's that bad!!

Agreed.  This seems like it might be an important step in the process.

Perhaps, an intelligence sharing treaty? So that my sovereign sees enemy units spotted by a friendly sovereign?

Perhaps, gaining the services of an allied champion for x months (turns)? Break the treaty and this champion leaves your service early. Lose a battle and this champion returns to his original lord.

Perhaps, loaning the services of your Kingdom's champion to an ally for x turns, for so long as the ally only battles forces of the Empire. (A way to indirectly oppose a fraction, without going to war against the fraction).

Perhaps, loaning a unit of veteran warriors (not untrained militia) to an ally for x turns. (another way to oppose a fraction without going to war against the fraction).

Well I'm a huge fan of resource trading and economics but if it really takes a lot of effort to implement right then you are probably better off spending the time on something else. The inability of AIs in WoM to work together to take down powerful players was a huge problem and would definitely be a priority.

  If you don't want to have active resource trading between players what about more automatic stuff. When it comes to trading the caravan system in Elemental is marvelous because it is so simple yet realistic. I can recommend a simple little change to caravans that would add a lot of depth without requiring practically any AI work.

I like this & the trade city idea above

I think trading with resources would be simple and useful, if not necessary. 

IE: I need metal and I have plenty of gold.  My ally needs gold and has an abundance of metal.  We trade one for the other.  Why is this so difficult or unsavory?

I think he was referring to over-time trade treaties like "I will give you five gold per turn for 30 turns" and such. I don't think they will be removing the existing ability to trade resources, since metal is so important that would be weird.

I believe what he means in general is that you won't be able to do complex trades involving cities, blackmailing, champions, techs, etc. Instead there will be generic treaties that accomplish similar things. ie you can't trade techs but you can enter into research treaties. Perhaps you can't blackmail the AI for 90 gold per turn  but there is a "surrender" treaty that forces them to pay a set amount of their income. Things like that.

Most treaties will probably work like in Sins, so that they give a multiplier to both players allowing players with good diplomacy to have tangible benefits. It also offers incentives to players to seek alliances even if they are not military. Its a good system if done right. 

Love it!  My eternal desire for all 4X games since I played Civ 1 is to have more intricate diplomacy.

"Need" is a 3 month coding process, minimum otherwise you'd just have players exploiting it.  I've written this kind of AI for decades and it's a time sink, and for what? It's not even very fun.  

By contrast, treaties ARE fun because there's a lot of interesting choices (you can only have 1 treaty in effect, you can't have the same treaty with every player).  They affect your relations, they give both sides benefits, they are just great tools to make the game more interesting.

Hmm, you can't have the same treaty with every player?  Will Diplomacy tech upgrade the value of those treaties?   I always felt the GC2 treaties were a good idea that needed more work, but didn't work that well in WOM, so curious to see what the changes are.

I really would like to see a dev diary on this stuff and how it works.

One thing I would like to see is tech diffusion via treaties.  Having treaties with an empire makes it easier to research techs they know.