AI does not value +diplo techs highly enough

This is the same strategy I use and it does make it quite easy. I've been playing on tough with a medium sized galaxy and am now going to have to ramp up the difficulty. The computer AI can't compete with a human influence/tech broker

I usually only have to forcefully invade 1 or 2 opponents the rest will ally up quickly for an easy win. One thing I also do is boost up trade routes and economic starbases to further raise income. Even if I get invaded earlier than I want, my really strong economy can usally handle it. The only time i've lost using this strategy is when I tried to side with the Drengi and they eventually turned on me and were able to take out 2 key planets before I could get defense ready.

I've tried going back to a pure military style, but the economic/diplomatic route is so much fun

If the play is getting formulaic, try turning tech trading off. If you can tech trade well, and it sounds like you can, it's easy to step on the AI. With tech trading off, well, it's still not exactly hard, but you don't get free money.

yeah I used this exact same strat in almost all of my games before 1.1, when I tried to not do it, the AI would end up getting ahead using tech trading so it would basically force me into it, and it was so easy to just get free money and cripple the enemy economy at the same time I really couldn't help myself sometimes. Now I just switch tech trading off and the game has a lot more depth to it.

@hampsta: I do admit this style of play is very fun, but it will probably "get old" for me very soon. Alliance victories are a cinch with this strat, I'm trying to turn them into militaries now because I hear you get a better Metaverse score. The Drengin should never have backstabbed you: just buy all their good ships, they will never bother you again.

@Veblen: Funny you should mention the no tech trading option, that's actually how I started playing in order to learn the game more effectively. Once I could beat Masochistic without tech trading I switched it on to compare - what a shock!

@Wargazmo: The game indeed has more depth with tech trading off. But as far as I know there's no "reward" for it in Metaverse scoring (which, of course, is not everything).

Just by the way, the point of my post is not to complain about tech trading per se, but rather that the AI's ignorance of diplomacy techs unbalances the trading component of the game.

I always thought diplomacy was far too powerful, which is why I mostly play with tech trading off. For my first metaverse game I made a small map on crippling and went diplomatic since my character (The Space Pope) is to be my good character. Anyway, I got a horrible start and lost out on all the possible colony locations as my 1st 2 star picks happened to be the homeworlds of other races. (I hate starting Terran) I barely made it back to Sol in time to grab Mars from the AI and was still able to play them with diplomacy to stay on top and pull a political victory out of my rear.

Only thing I can suggest is to play differently or turn tech trading off if you don't want to exploit the poor AI for now. =/

It's not just tech trading, but the nice thing about diplomacy techs is that it can prevent AI players from going to war with you, even if they are otherwise very hostile. I mean this seems obvious right, but I didn't expect how effective it can be!

Version 1.1 I was playing this modified race with bonuses on economy + pop growth, so early strategy involved quick land grabs + focusing on economy techs so I could operate at 100% spending for most of the game. Opponents were humans and Tolarains set to intelligent.

With no military at all besides a flag ship and several scouts (no weapons) , i continued to crank out fast colony ships (not much range just daisy chain colonies) and constructors to mine all the resources (economy ....yum tum) took over about 50%-60% of the map,

This was the point where usually i would start researching military techs, as well as spending resources for building up a token military (just to ramp up the total attack score), to remove the 'conquer me' sign as some guy wrote. I was probably cutting it a bit close in fact.

But this time i decided to experiment with diplomacy related techs, so I focused on those, and influence related tech. Yes, my influence was crazy, to the extent, that the humans and tolarians found my influence 'alarming'. but still they didn't attack.

So i continued cranking colony ships colonising the smaller worlds, as well as the last few planets lurking at the edges. By now I colonised 70% of the map, with my influence over 75%, 1st in almost every area except military (last) and approval. I could basically win any way I wanted.

It seems if your diplomacy is at the stage where you can 'do the jedi mind trick', even if they hate your guts they won't attack. I mean they had 3 reasons to hate me, I was evil, I had no military, I had an alarming influence and the humans were miliartistic which made 4, and yet my diplomatic skills alone kept them off.

I suppose i would have won anyway (getting 50%-60% of the map in a 3 way fight is a big advantage), if i didn't do the diplomacy trick, but they might have made me sweat a little and slowed down my win while i caught up with them in military.

From the point of view of computer AI, i can see why they shouldn't value diplomacy techs too highly, because after all
I'm not going to give the AI a better deal just because it is 'majestic'. But it seems necessary as a defensive measure, to prevent humans from taking them for a ride.

Tech trading is the most obvious area for exploitation and this has being addressed, but I think it should also be more prepared to go to war, even if the opponent has godlike dipolmacy skills if the pros for going to war are high enough.

From the point of view of computer AI, i can see why they shouldn't value diplomacy techs too highly, because after all I'm not going to give the AI a better deal just because it is 'majestic'. But it seems necessary as a defensive measure, to prevent humans from taking them for a ride.

That's a good point. However, the AIs would still get diplomacy bonuses when dealing with each other (I assume), so neglecting the diplomacy branch for such a long time does not make for a very realistic/holistic AI (i.e. one that fully knows how to play the game, not just how prevent the human player from winning).

That's a good point. However, the AIs would still get diplomacy bonuses when dealing with each other (I assume), so neglecting the diplomacy branch for such a long time does not make for a very realistic/holistic AI (i.e. one that fully knows how to play the game, not just how prevent the human player from winning).

I don't think the AI is designed to gang up on the human to prevent him from winning. In fact the AI plays in character.

That said , well the thing is, from the point of view of AI vs AI, if they all start valuing diplomacy techs more , they would be roughly at the same level with each other and it cancels out. .

I'm just pointing out that for these techs there's an asymmetry here, where the AI gains less from diplomacy techs than the human.

As for the point about getting a real AI that plays exactly like a human, I think that isn't one of the goals of the game. If that was really the goal of the game, there would be no such thing as diplomacy techs, because they can't affect humans! Ditto with all those specific messages complaining about things the AI sends you, while you can't return the favour.

The AI doesn't cheat but it isn't exactly the same as a human player.

Yes, certain AI:s, terran for example, would benefit from valuing diplomatic techs alot more. While others, for example drengin, should keep ignoring them like they already do and focus on what's IMO important, cooler looking guns and bigger hulls.

From the point of view of computer AI, i can see why they shouldn't value diplomacy techs too highly, because after all I'm not going to give the AI a better deal just because it is 'majestic'. But it seems necessary as a defensive measure, to prevent humans from taking them for a ride. Tech trading is the most obvious area for exploitation and this has being addressed, but I think it should also be more prepared to go to war, even if the opponent has godlike dipolmacy skills if the pros for going to war are high enough.

This explains a few things. Ever since I read that 1.1 beta 4b included the new "gang up on the aggressor" protocol. I've played a couple games where I would have a military just big enough to be left alone and then aggressively "attack" with influence star bases. To date, not one AI has defended against the ISB's. Of course, before I go spamming them, I've researched through the diplomacy tree to have all modules available and have the highest diplomacy.

There's a line from "Catch Me If You Can" ...... ...... "You know why the Yankees are so good? It's not because they have the best players, it's because the other teams are so busy looking at their pin-stripes."

I'm still going to finish my 1.1 final experiment, but it's nice to see some explanation for my observations. Then I can go back to actually building fleets that can do something.

This explains a few things. Ever since I read that 1.1 beta 4b included the new "gang up on the aggressor" protocol. I've played a couple games where I would have a military just big enough to be left alone and then aggressively "attack" with influence star bases. To date, not one AI has defended against the ISB's. Of course, before I go spamming them, I've researched through the diplomacy tree to have all modules available and have the highest diplomacy.

Well I've seen them try to defend from my overwhelming influence by building influence starbases though...... But they don't go to war over it. Altough it shows up as a negative factor in diplomacy...

I don't think the AI is designed to gang up on the human to prevent him from winning. In fact the AI plays in character.

I'm not saying the AI gangs up the human player. But if it negelcts diplomacy techs simply because they are not useful against the human, it's not quite as "in character" as in could be, right? If the Humans are supposed to be the "diplomacy AI", they should research diplomacy techs.

That said , well the thing is, from the point of view of AI vs AI, if they all start valuing diplomacy techs more , they would be roughly at the same level with each other and it cancels out.

That's like saying that if they all prioritize military then their militaries cancel each other out. I'm not saying they should all prioritize diplomacy, but some should: Humans, Drath, etc. In my games all the AIs consistently refuse to research the diplomacy techs.

As for the point about getting a real AI that plays exactly like a human.

Note sure if you're quoting me here, but I never said that. I said the AI should know how to play with all the difference aspects of the game, in order to give the human player a well-rounded challenge. If you think about it, that's distincly non human-like.

That's like saying that if they all prioritize military then their militaries cancel each other out.

That's true in a sense, except again, military forces work equally well against the humans. Your point was I believe about computer AI exploiting other AIs using diplomacy, and that obviously doesn't work if they all start using it more something which I think was implied in the thread, which says AIs does not value dip tech.. This seems to say all AIs should benefit from valuing diplomacy tech more, something I agree with.

I'm not saying they should all prioritize diplomacy, but some should: Humans, Drath, etc. In my games all the AIs consistently refuse to research the diplomacy techs.

Actually, I (and other people in this thread apparently) are saying they should ALL value diplomacy more, independent on whether some should prioritize it more. The thread here talks about AI in general after all.

I think you overstate the case a little though, they *do* research such techs, just not a big priority, and certainly they don't try to reduce the diplomacy gap, something they should do just as they try to counter players trying to kill them with influence or with certain military tech.

I do agree that none of the AI personalities use diplomacy as a weapon , but I think in general the problem is not
whether such AI should exist, but that all of them should notice diplomacy being used as a weapon and act to counter it. Right now they notice influence but not diplomacy even though in theory the 2 come hand in hand a lot of the time.

Note sure if you're quoting me here, but I never said that. I said the AI should know how to play with all the difference aspects of the game, in order to give the human player a well-rounded challenge.

No offense but I think the implication was there. when you talked about a an AI that wasn't centered around trying to beat the human.

If you think about it, that's distincly non human-like.

A well rounded player who can play with different aspects of the game is non human-like? Could have fooled me.

Yes, certain AI:s, terran for example, would benefit from valuing diplomatic techs alot more. While others, for example drengin, should keep ignoring them like they already do and focus on what's IMO important, cooler looking guns and bigger hulls. [Ensign] [Yor] [Tough]

From the gaming point of view, if the drengin don't care about diplomacy at all, i can keep them from attacking me, regardless of their cool guns and hulls...

The humans would be less affected by my diplomacy skills, but they don't carry a big stick.... LOL.

From the gaming point of view, if the drengin don't care about diplomacy at all, i can keep them from attacking me, regardless of their cool guns and hulls...

Well, if that's the case, it's stupid and should be changed. Diplomacy shouldn't have such a huge effect on relations that it makes all the other things that affect relations meaningless. Drengin are both evil and a warrior-race, if they see an opportunity to grab some planets, they should take it, regardless of the big + from diplomacy.

That's true in a sense, except again, military forces work equally well against the humans. Your point was I believe about computer AI exploiting other AIs using diplomacy, and that obviously doesn't work if they all start using it more something which I think was implied in the thread, which says AIs does not value dip tech.. This seems to say all AIs should benefit from valuing diplomacy tech more, something I agree with.

If none of the AIs research diplomacy techs, the human player cleans up. If they all prioritize diplomacy techs more by a fixed amount this little "exploit" is fixed, but then the races all feel the same and the solution is just code intended to beat the human player. That's my objection. There should be some races that like the diplomacy branch more than others, but none of them should ignore it. On average they should all prioritze diplomacy techs more (see the topic of my post).

The only reason I mentioned the "races should be different" idea is in response to your "diplomacy bonuses cancel each other out" statement. They would not cancel each other out if some races felt more strongly about diplomacy, some less - but all, of course, more than at the present time.

This is great, we're arguing the same point. Your turn!

A well rounded player who can play with different aspects of the game is non human-like? Could have fooled me.

Of course: anything that humans do is human-like. You got me...

My point is that, in general, human players tend to specialize their game very quickly as they master it. Just look at how many players on this board favor pop.growth/morale/economy as their custom traits. This does not do for the AI, which must present a well-rounded challenge in order for the game to be fun. So in this sense, no, the AI is not coded to "play like the human". It has to come prepared AND be entertaining!!

You guys could experiment with changing the AI values of the diplo techs in the XML file. Advanced Diplomacy has and AI value of 3, Expert Diplomacy has an AI value of 1!

Yeah, with all the effort put into the AI making better decisions, its odd that they trade away their dip bonuses so easily. Like many others, I never *ever* trade a dip bonus tech. That gives you a pretty major advantage.

From the point of view of computer AI, i can see why they shouldn't value diplomacy techs too highly, because after all I'm not going to give the AI a better deal just because it is 'majestic'.

This is one of the three biggest asymmetries between a human player and an AI player: a human player can always click "Reject." The second big asymmetry is that a human player will tend to trade, and especially tech trade, constantly, while the AI does it only occasionally. The "No Tech Trade" button eliminates the disparity to a certain extent, but not all the way. The third big asymmetry is that the player is unconstrained by his diplomatic relationship with other races. The only way to bring the AI significantly closer to the level of a human player is to (a) place the same restrictions on the player that are placed on the AI, or (b) eliminate some of the restrictions the AI faces.

Option A:
The AI cannot declare war on someone it does not have poor relations with; neither can you. The AI cannot refuse a trade which is "balanced" according to relative diplomatic skills and relevant values; neither can you. In order to make this work, a few AI values are tweaked. You value diplomacy techs because they let you take advantage of others with regards to tech trading and relations; so do certain AIs to varying extents (Terrans especially, I would imagine). You tech trade every few weeks to gain advantage; so do certain AIs to varying extents(again, Terran especially).

This option makes the game more interesting in that the AI can take advantage of you if you don't boost diplomacy, but forces you to play like a robot. The playing field is leveled at the cost of creativity, because now you are playing by the same rules as the AI.

Option B:
You attack whomever you feel like, regardless of relations; so do sneaky bastard AIs (Drengin and evil guys especially), and other AIs add "noise" to their friendship calculations, making them less predictable friends. If you don't like a deal, you tell the AI to forget it; unless the AI needs what you're peddling (e.g. it's in a war with someone who uses mass drivers and you have high level armor), it'll tell you to buzz off. The AI never cared about diplomacy, and now that you're playing against enemies that aren't constrained by your smooth-talking ways, you don't care either.

Option B levels the playing field, but at the cost of enjoyable game mechanics. Now the AI is playing the same game as you, or closer to it, so you can't take advantage of diplomacy any more.

I prefer option C: constrain the player with regards to making deals (either by eliminating choice or adding substantial penalties to relations with the offering race or all races when you refuse goods deals), boost the amount by which the diplomatic AIs value diplomacy, boost the frequency with which those AIs offer tech deals to you and each other, and let the evil AIs attack their friends while maintaining a preference for attacking those with lower relations. That way the Terrans actually act like diplomatic power brokers and you can't sweet talk the murderous rampaging robots into leaving you alone all the time.

Thoughts?

This is one of the three biggest asymmetries between a human player and an AI player: a human player can always click "Reject." The second big asymmetry is that a human player will tend to trade, and especially tech trade, constantly, while the AI does it only occasionally.

I disagree that human player tending to trade frequently is a asymmetry, because the AI can be designed to do that too. In fact, in pre1.1 versions they tend to trade everything, so if you traded a unique tech to one of the Comp AI, soon, all of the AIs would have them.

The third big asymmetry is that the player is unconstrained by his diplomatic relationship with other races. s.

Agreed. I would say the 2 main asymmetry for diplomacy techs are

1) Human player does not take into account diplomacy when deciding to go to war.
2) Human player does not take into account diplomacy when accepting deals.

I'm not too worried about #2.

#1 is pretty serious because while the AI is generally predictable just by looking at the diplomacy score card, humans aren't.

I would say a way around the problem is for AIs to have personalities. Some would be 'chaotic', 'unstable' etc, in that
they can declare war at a drop a hat, regards of relations. Or they might offer peace, even though diplomacy wise it makes no sense. Others would be lawful and tend to be 'rational' so act like the current AI.

The only way to bring the AI significantly closer to the level of a human player is to (a) place the same restrictions on the player that are placed on the AI, or (b) eliminate some of the restrictions the AI face

Here's a crazy idea.

Before the game begins you can choose to be a chaotic personality or a lawful personality.

If you choose the former, you can do whatever you like regardless of diplomacy, you can declare wars at the drop of a hat, reject tech trades you don't like even though a computer AI in your position would , basically what galciv 2 is now . On the negative side, because the computer Ai knows you are chaotic, your own diplomacy techs also don't influence their behavior when you are dealing with them. But their techs would be fully functional when dealing with 'lawful' personalities.

I mean fair's fair right? If the computer AI doesn't gain from diplomacy techs against you, you shouldn't gain the advanatages of the tech either against them.

A lawful personality would take away choices from the human player. If the AI offers a trade and it's fair, you can't decline. If you want to go to war, you might not be allowed if it is too 'crazy'.

E.g 'You intend to go to war, but the humans diplomacy skills, plus your close trading links convince you not to, for the next X turns'.

Very restrictive, but it's not unheard of already, since if you use none-imperial forms of government, you can be overruled at times if you want to go to war.

A third option would be between both extremes. Diplomacy techs work, but to a lesser degree. Similarly, you are occasionally constrained to do things, but most of the time you are allowed to do what you want.

Just wondering, but has anyone won a masochistic game with all of the AI's starting the game in a state of war with you?

I myself only bother to play the game on tough, so the question is just academic.

END COMMUNICATION

Before the game begins you can choose to be a chaotic personality or a lawful personality. If you choose the former, you can do whatever you like regardless of diplomacy, you can declare wars at the drop of a hat, reject tech trades you don't like even though a computer AI in your position would , basically what galciv 2 is now . On the negative side, because the computer Ai knows you are chaotic, your own diplomacy techs also don't influence their behavior when you are dealing with them. But their techs would be fully functional when dealing with 'lawful' personalities. I mean fair's fair right? If the computer AI doesn't gain from diplomacy techs against you, you shouldn't gain the advanatages of the tech either against them. A lawful personality would take away choices from the human player. If the AI offers a trade and it's fair, you can't decline. If you want to go to war, you might not be allowed if it is too 'crazy'.

Your solution more or less takes what I was saying to its logical conclusion in a way that permits the player to retain a significant degree of choice over the kind of game he plays, but it feels unsatisfying for a game that prides itself so much on its AI. After thinking about it for a while, what I'd really like to see is something that deals with the second problem:

2) Human player does not take into account diplomacy when accepting deals. I'm not too worried about #2.

It's easy not to be, because it doesn't seem to be game-breaking, but it's the big reason that tech trading has been such an issue, to the point that there's a No Tech Trade button in 1.1. If you fix this problem and force the human player to accept deals, you might start seeing what it's like to be the AI facing a human. Imagine that the AI had a high natural diplomacy while your racial bonuses were focused on population and morale and econ. It focuses on diplomacy tech, and gets Diplomatic Translators, and won't trade them away. All of a sudden it starts offering you the kind of trades that you used to offer it. Suddenly you're giving Photon Torpedos and cash for Habitat Improvement, and the AI is siphoning your money and technology away. This is a messed up situation, but I didn't quite nail the solution in my earlier post; I think I'm closer now.

The problem is with the AI, and I think it's not that tough of a fix. Time intensive for the developers? Maybe. But definitely doable, maybe even for 1.2. The problem is that the AI evalutates value, but not need. If I want an iPod, I value that iPod at $300, or whatever the price is these days. There's the market price for services and IF I WANT SOMETHING I'll pay it. I don't want an iPod, so even though I know what it's worth, I won't necessarily pay $300 for it just because the offer comes along.

The AI needs a secondary layer of value built in so that it can evaluate whether it wants/needs something. The very first check the AI should make before accepting a deal is whether it wants/needs the thing it's getting; if yes, proceed. The second check should be a comparison check of whether the thing it's giving up is something it needs more than what it's getting. The final check should be whether the deal is for fair value, which is the only check it currently does. Note that those last two could be combined, with the AI discounting or marking up value based on need, but I think the flat evaluation is better; just because I need a car to get to work doesn't mean I'd trade my home, which I need to live in, for 5 Aston Martin Vanquishes. I'd rather work for the money, then buy the car myself, then have BOTH things I need.

In principle, that sounds like a tall order, but the AI already makes the hard judgments underlying need/want all the time. The AI evaluates what it wants to research in a given situation, whom it wants to attack, what it wants to build, what boosts it wants from trade goods, etc. It shouldn't be terribly difficult (I imagine, but what the hell do I know) to divert the values the AI checks to decide these issues into programming that decides what the AI is willing to trade for. There would need to be a few extra values added, and some new decisions to be made (for example, what percentage of its money is the AI willing to spend to buy new tech?), but given the quality of the AI that Stardock has put out thus far, I can't imagine that this sort of thing is beyond their abilities.

As far as AI wars of aggression go, I'd like to see more variance and backstabbing based on alignment, with less emphasis on relationship status by the AI. If I'm playing as evil, I'm clearly not to be trusted, so our "historic friendship" shouldn't keep the AIs from watching their backs. The whole point of the alignment system, in my mind, should not just to give interesting bonuses, but to give the AI some clues as to who can be trusted, and who can't, as well as to give the AI some "moral" flexibility.

Edit: in fact, going back to Richrf's notion regarding players choosing their level of freedom: choosing alignment should do that. Good folk should never be able to attack people who are not already their enemies or at low relations with. Neutral folk shouldn't be able to attack people when their relations are better than neutral. Evil folk should be able to attack anyone (except maybe allies). The counterpoint to this sacrifice of freedom would be that Good civs would get boosts to diplomacy and even bigger boosts to relations with other civs than they do now. The extent of the bonuses would, naturally, have to be determined by playtesting.

As far as AI wars of aggression go, I'd like to see more variance and backstabbing based on alignment, with less emphasis on relationship status by the AI. If I'm playing as evil, I'm clearly not to be trusted, so our "historic friendship" shouldn't keep the AIs from watching their backs. The whole point of the alignment system, in my mind, should not just to give interesting bonuses, but to give the AI some clues as to who can be trusted, and who can't, as well as to give the AI some "moral" flexibility.

I was thinking of something along the same lines:

---

1. Good civs get full diplomatic ability bonuses when trading with other Good civs.
Evil civs get full diplomatic ability bonuses when trading with other Evil civs.

2. Diplomatic ability bonuses count as negative when trading with civs of the opposite alignment.

3. Diplomatic ability is halved when a Neutral civ is involved.

---

That's a bit of a departure from the current implementation, but still I think within the intent of the game's design.

The idea is that tech trading "cliques" are formed along ethical lines, with Neutral civs being able to trade with everyone but never getting the best deals because they are never fully trusted. The AI would still have to prioritize the diplomacy techs more, but (roughly) half the problem of unrestricted tech trading would be eliminated.

Note that this rule would only apply to trading. Getting a high diplomacy bonus would still make sense in order to avoid being attacked, and whatever else it applies to.

It would still be possible to trade with civs of the opposite alignment, of course, just a lot harder to get what you want. If the goody-goody Altarians request Photon Torpedoes from the Drengin they might get it, but not without coughing up Harmony Crystals and Micro Repair Bots first (or whatever).

Prioritizing the diplomacy techs would still be a viable strategy: you would get better deals from the civs of your alignment and Neutral civs than they get from you.

The main thing I'm worried about with this idea is that it might make Neutral the no-brainer alignment. It's already the arguably the most powerful. It might need some tweaking in other areas.

Thoughts?