Rushing AIs... Are you turning on Aggressive AI?

[futurehermit]

It's been my general experience that the AI is fairly unwilling to wage an early war itself (unless it is against a poorly defended human player), to expand, grab more land, and remain competitive against the player's growing empire. I almost never see an AI eliminate another AI, unless there was somehow an extremely large disparity in their empire size to begin with (European nations forced to play OCC due to start situation in the Earth map come to mind) All too often, the AI's play nice, allow themselves to be boxed into a tiny area, and in the midgame you see a bunch of smaller nations coexisting on the other continent, while you have monopolized yours. If the AI were more willing to eliminate rival AIs so it could keep up a landmass parity with the player, then maybe the early rush strategy wouldn't create such an overwhelming advantage for the player in the mid to lategame.

An absolutely brilliant explanation of what is wrong with the AI in this game imo.

EDIT: That's not to say that I think every game should end up 1-vs-1. In theory, it would be nice if the AIs could have a reasonable chance of winning with a small empire as well. But, in general, it is extremely hard for small-territory AIs to compete with a human player who has control of their continent.

 

[aelf]

I just happened to stumble on this thread. I haven't been playing much, plus my memory of the details of the old debate is hazy. However, I think one of the big concerns at the time was the de facto recognition of warmongering, and not simply rushing, as the 'optimal' strategy in Civ4. There are a few points about this.

While early rushing might be considered the equilibrium strategy for the old regular AI because it yielded the greatest payoff, gaming, especially for something as multi-dimensional as Civ, should not require one to play the optimal strategy. There's often a bit of a role-playing element, so, even if the "warmongering is optimal" notion is/was true, some people would still prefer to be builders. Aggressive AI was seen as pushing people to make the 'equilibrium' choice of warmongering, greatly reducing the multi-dimensional aspect of Civ.

Even if you could survive by spamming units to deter attack, this essentially negated the role of the builder, since you have to continuously build large numbers of units, which are sitting around and being paid for without actually being put to use. So, on one hand the builder has to accommodate his strategy to the new warmongering demands, and on the other, he's all the more clearly not playing optimally because all those units sitting around amount to a large waste of hammers and gold if they are not used for - guess what - warmongering. Naturally, the debate evolved gradually into the issue of unit-spamming and how it was possibly the way to play BTS given a challenging difficulty, whether you chose to be a warmonger (a clearly superior choice) or not.

Also, at that time, and before the debate turned to unit-spamming, I pointed out that warmongering was not necessarily the optimal strategy. As it happened in one of my public games, I clearly lost because of too much warmongering, which left me behind the largest Immortal AI, whom I could not reach and take down before he grew too powerful. Basically, the conclusion was that the optimal way of playing the game was not to war all the time, but knowing when to and when not to.

I don't know how much light this history can shed on the game today, in its current form. Heck, my current ongoing public game (which I've had trouble finding time for) may even be behind with the patches by now. But I hope it's interesting nonetheless

 

[PieceOfMind]

I just happened to stumble on this thread. I haven't been playing much, plus my memory of the details of the old debate is hazy. However, I think one of the big concerns at the time was the de facto recognition of warmongering, and not simply rushing, as the 'optimal' strategy in Civ4. There are a few points about this.

While early rushing might be considered the equilibrium strategy for the old regular AI because it yielded the greatest payoff, gaming, especially for something as multi-dimensional as Civ, should not require one to play the optimal strategy. There's often a bit of a role-playing element, so, even if the "warmongering is optimal" notion is/was true, some people would still prefer to be builders. Aggressive AI was seen as pushing people to make the 'equilibrium' choice of warmongering, greatly reducing the multi-dimensional aspect of Civ.

Even if you could survive by spamming units to deter attack, this essentially negated the role of the builder, since you have to continuously build large numbers of units, which are sitting around and being paid for without actually being put to use. So, on one hand the builder has to accommodate his strategy to the new warmongering demands, and on the other, he's all the more clearly not playing optimally because all those units sitting around amount to a large waste of hammers and gold if they are not used for - guess what - warmongering. Naturally, the debate evolved gradually into the issue of unit-spamming and how it was possibly the way to play BTS given a challenging difficulty, whether you chose to be a warmonger (a clearly superior choice) or not.

Also, at that time, and before the debate turned to unit-spamming, I pointed out that warmongering was not necessarily the optimal strategy. As it happened in one of my public games, I clearly lost because of too much warmongering, which left me behind the largest Immortal AI, whom I could not reach and take down before he grew too powerful. Basically, the conclusion was that the optimal way of playing the game was not to war all the time, but knowing when to and when not to.

I don't know how much light this history can shed on the game today, in its current form. Heck, my current ongoing public game (which I've had trouble finding time for) may even be behind with the patches by now. But I hope it's interesting nonetheless

Thanks for your thoughtsful post aelf. I remember you were very involved in the first discussion and you had some of the strongest points against the intended changes to Agg AI.

I wanted to clearify this but Andvare has already pretty nicely covered the subject. You made an assumption that is not agreeable (rush always best), so the discussion was pretty much guaranted to strive in that direction.

I agree with what you write but you are drawing the discussion away from the point. I can appreciate you want to defend your position but can you please drop that issue now, because it is more a distraction?

NEW QUESTIONS... THINK OF THIS LIKE A NEW FIRST POST.

The argument is... if you do rush, do you use Agg AI? I do not want anyone answering this question by saying either (a) optimal play often means not rushing, or (b) in Agg AI games you can't always rush. Basically because these two responses are stating obvious things and missing the point of the question, drawing the discussion away from the actual question. For (b), if you start an Agg AI game and you can't possibly rush anyone (maybe isolated start?) then I'm saying you should probably regenerate the map.

Another question that is more relevant to the discussion I'm trying to start is this...

Is it considered fair to rush an AI on the regular AI setting? In answering this question, don't make the comparison with doing the same thing with Agg AI - forget Agg AI even exists if you must.

Having answered that question, then try the following question...

How could Agg AI be made different from regular AI so that the game is a greater challenge for the human player who usually (not always!) rushes?

If you wished, you could instead respond to this question by stating your case for why Agg AI should not be turned into a setting for rushers. That is an entirely reasonable position to take IMO, but in that case how else do you think Agg AI should be programmed to be different from regular AI?

If anyone still positing here could please offer their opinions on these questions, I'd be very grateful!

 

[Andvare]

I'd like to see the AI gather information, and react to what it's opponents do.
Either through espionage (let it use the relative power rating better) or through scouting. Let the AI count the enemy units it sees, and let it compare that with the distance to the enemy and it's own number/strength of units.

I would also like to see the AI gang up on the leader, be it an AI or a human, and I think this would fit nicely with agg AI.
And the AI should go for the kill. Every human player can tell you that taking one or two cities in a war is not enough, you end up loosing a lot of your units (the first few cities are often the best defended) and gain little. Eliminating other players should be part of the agg AI. Just like what the person living alone in the future quoted above.
I don't mind that the AI is banned from rushing, as that would absolutely flatten the player, I would like it to be a selectable setting though, as I like to play on lower-than-my-skill level, and it would be very entertaining to see Monty rush me (when I have a chance of stopping the rush). This could also be part of the agg AI, though I'd prefer it to be a separate setting.
Agg AI should be more opportunistic and join in the slaughter of a neighbour more often.
Agg AI shouldn't just be about more units (it feels that way ATM IMHO, though I haven't tried better AI mod), but more about more wars and more complete wars, not just this silly border skirmish that characterize the AI as it is nowadays.

I realize that some of this is probably outside the scope of a mod, but, hey, one can dream while waiting for CivV .

 

[lilnev]

"Fair"? Fair to whom?

Maybe some of the confusion here is that other people are making positive statements -- statements about facts, such as "rushing is often the optimal strategy," or "rushing against Agg AI is harder than rushing the regular AI." Positive statements may be true or false, but they don't even attempt to speak to whether something is good or bad.

Meanwhile, you're looking for a debate on normative terms. A normative statement is one about how things should be, and necessarily speaks to whether something is good or bad.

OK, here's my normative statement: "Players should choose the settings that they enjoy, and play in the manner that they enjoy." If you enjoy rushing against regular AI, go for it. You're not being "unfair" to anyone, you're enjoying your game. If you prefer to play on a level/settings where you're unlikely to ever lose, have fun.

 

[Skallagrimson]

OK, here's my normative statement: "Players should choose the settings that they enjoy, and play in the manner that they enjoy." If you enjoy rushing against regular AI, go for it. You're not being "unfair" to anyone, you're enjoying your game. If you prefer to play on a level/settings where you're unlikely to ever lose, have fun.

Very well-put, IMO. I play at a range of levels because sometimes I'm more in a mood for a fairly guaranteed win (using the game to blow off steam), and sometimes I want a supreme intellectual challenge with just a shred of hope of winning. And sometimes more in-between when it's basically the start position and the leader I random-rolled, that decides.

 

[SnowlyWhite]

The argument is... if you do rush, do you use Agg AI?

No; agg AI never convinced me... it's just dumbing down the game imho.

Is it considered fair to rush an AI on the regular AI setting?

Yes...

How could Agg AI be made different from regular AI so that the game is a greater challenge for the human player who usually (not always!) rushes?

The reason why the ai sucks compared to human is that it doesn't conquer. Imho that's the sole reason. A competent AI shouldn't vassalize another AI after getting 2 cities, but should try(as long as it's economically logic) to grab all the land from the weaker AI. Furthermore, when you bribe them to peace in order to let the weakling survive, it should be costlier(it costs around 3 turns of 0% slider later in the game - 1 turn on normal speed which is ridiculously cheap) and be more prone to redeclare asap. Anyway, issue is with AI acquiring land(most probably from another ai - there are 5 other ais on a std. map and only 1 human).

But even on marathon I find rush to be optimal ~50% of the cases; the other time it usually just leaves room for another ai to settle and become a monster...

p.s - other aspects badly needing improvement would be worker turns and city placement(though that can hardly go in a setting named aggAI ) but I admit those are much harder to implement...

 

[PieceOfMind]

"Fair"? Fair to whom?

Fair to the AI.

Maybe some of the confusion here is that other people are making positive statements -- statements about facts, such as "rushing is often the optimal strategy," or "rushing against Agg AI is harder than rushing the regular AI." Positive statements may be true or false, but they don't even attempt to speak to whether something is good or bad.

Meanwhile, you're looking for a debate on normative terms. A normative statement is one about how things should be, and necessarily speaks to whether something is good or bad.

OK, here's my normative statement: "Players should choose the settings that they enjoy, and play in the manner that they enjoy." If you enjoy rushing against regular AI, go for it. You're not being "unfair" to anyone, you're enjoying your game. If you prefer to play on a level/settings where you're unlikely to ever lose, have fun.

I absolutely agree that players should choose the settings they have the most fun with, and I've said this many times. For many players I suspect they would take away more enjoyment from the game the more they feel they have beaten a worthy opponent. If players have to play Deity where the bonuses are ridiculous to the AI to have to get a challenge, I would suspect it's not as satisfying as playing a much more capable AI at a lower difficulty. This is a normative statement but based on some people's fascination with Better AI and the general forumer's obsession with demanding a more competent AI, I think it's a reasonable statement for me to make.

If an AI cannot rush you, but you can rush the AI (and quite decisively at that, usually), is it fair? In particular, think about the game chances for that AI you just crippled. No one really gives the AI an opportunity in these questions of "fairness" because it is after all just code. Machines are not there for us to please, they're there to please us (videogames) or serve as tools for various things (business etc.), so it is IMO also reasonable to say slaughtering the AI by taking full advantage of its weaknesses is nothing reprehensible. We simply do not see any need whatsoever to be fair to computers.

The 'positive' statement I am making is that for people who prefer to play in a way they feel is fair to the AI, they may not think twice about whether it is fair to rush a regular AI. Regular AI is after all part of the default game.

Perhaps you think I am personifying the AI too much. Perhaps I am.

The reason why the ai sucks compared to human is that it doesn't conquer. Imho that's the sole reason. A competent AI shouldn't vassalize another AI after getting 2 cities, but should try(as long as it's economically logic) to grab all the land from the weaker AI. Furthermore, when you bribe them to peace in order to let the weakling survive, it should be costlier(it costs around 3 turns of 0% slider later in the game - 1 turn on normal speed which is ridiculously cheap) and be more prone to redeclare asap. Anyway, issue is with AI acquiring land(most probably from another ai - there are 5 other ais on a std. map and only 1 human).

I know I keep bringing up this mod but in Better BtS AI, AIs have actually won Domination victories in several games now. They definitely "conquer" a lot better than in the regular game - they especially mount naval invasions much more effectively and they also use air warfare much more effectively.

 

[lilnev]

OK, I think I understand where this idea is going, or perhaps where it started. For you (PieceOfMind) to enjoy the game, you have to feel that you're playing "fairly". This is quite reasonable, and probably pretty common, so long as we recognize that each person's definition of "fair" is a personal one. It's useful to describe what conduct of play you enjoy; it's not useful to try to hold other players to your personal standard. (Talking SP of course; and competitions like HOF, publicly played games, and boasting all presume some basics like "no WorldBuilding or reloading to a previous save after you've played on.")

OK, my personal standards for fairness include:
No WB, with very rare exceptions to correct legitimate mis-clicks.
No save-reloads.
No exploiting what seem to be code bugs (I believe the PRO-Stone-Walls overflow "trick" to be a bug, and thus don't use it).
No odd corner cases that seem out of the spirit of the relevant game mechanics (chain-transporting; farming siege XPs off a barbarian city).

Beyond that, I play to win, and choose settings hard enough that often won't. I only enjoy the game while the outcome remains in doubt, and my settings (posted previously) make it unlikely that I'll reach a "winning position" early (I also generally play for domination; on large maps, that can mean modern wars against X and Y under time pressure from Z's spaceship).

I could understand, though I don't share, the idea that it's unfair to act in a way that the AI is specifically programmed not to -- declaring before 1000 BC, or declaring at Friendly, for example. Maybe this is because I see the challenge as the game overall, rather than a one-to-one competition with each of the other civs. I'll axe-rush if it seems best, and Asoka's game may be over, but my challenge certainly isn't.

Overall, we can't expect "fair" to mean "symmetrical". Unless you plan to play on Noble and lots of really bad decisions, it's going to be an asymmetrical game. The AI gets bonuses. I'm smarter and get fewer constraints on my actions.

And back to the original topic (rushing vs. Agg or regular AI). A case could be made that's the opposite of what you seem to be thinking. If you enjoy rushing and you want a challenging game, you should play the version of the game in which rushing is more frequently optimal (i.e. regular AI), and turn up the difficulty to compensate. If you enjoy trying to out-tech the AIs while periodically defending yourself, use Agg AI and turn up the difficulty to compensate. This way, you're playing the game that you enjoy at a level that is challenging.

 

[PieceOfMind]

OK, I think I understand where this idea is going, or perhaps where it started. For you (PieceOfMind) to enjoy the game, you have to feel that you're playing "fairly". This is quite reasonable, and probably pretty common, so long as we recognize that each person's definition of "fair" is a personal one. It's useful to describe what conduct of play you enjoy; it's not useful to try to hold other players to your personal standard. (Talking SP of course; and competitions like HOF, publicly played games, and boasting all presume some basics like "no WorldBuilding or reloading to a previous save after you've played on.")

OK, my personal standards for fairness include:
No WB, with very rare exceptions to correct legitimate mis-clicks.
No save-reloads.
No exploiting what seem to be code bugs (I believe the PRO-Stone-Walls overflow "trick" to be a bug, and thus don't use it).
No odd corner cases that seem out of the spirit of the relevant game mechanics (chain-transporting; farming siege XPs off a barbarian city).

Beyond that, I play to win, and choose settings hard enough that often won't. I only enjoy the game while the outcome remains in doubt, and my settings (posted previously) make it unlikely that I'll reach a "winning position" early (I also generally play for domination; on large maps, that can mean modern wars against X and Y under time pressure from Z's spaceship).

I could understand, though I don't share, the idea that it's unfair to act in a way that the AI is specifically programmed not to -- declaring before 1000 BC, or declaring at Friendly, for example. Maybe this is because I see the challenge as the game overall, rather than a one-to-one competition with each of the other civs. I'll axe-rush if it seems best, and Asoka's game may be over, but my challenge certainly isn't.

Overall, we can't expect "fair" to mean "symmetrical". Unless you plan to play on Noble and lots of really bad decisions, it's going to be an asymmetrical game. The AI gets bonuses. I'm smarter and get fewer constraints on my actions.

And back to the original topic (rushing vs. Agg or regular AI). A case could be made that's the opposite of what you seem to be thinking. If you enjoy rushing and you want a challenging game, you should play the version of the game in which rushing is more frequently optimal (i.e. regular AI), and turn up the difficulty to compensate. If you enjoy trying to out-tech the AIs while periodically defending yourself, use Agg AI and turn up the difficulty to compensate. This way, you're playing the game that you enjoy at a level that is challenging.

All great points. In fact you perfectly demonstrated an example of two players not agreeing on what is considered fair, because I have never thought twice about "chain-transports" as you called them. The reason I don't believe it's an exploit is because you don't actually increase the "bandwidth" across an ocean, you simply increase the speed of the transporting when there is less to transport. But since the AIs don't use it I can definitely see a case for me to reconsider the way I have used this. You probably changed my mind. Anyway, I digress a little...

Your suggestion that a case could be made for Agg AI being for the non-rushers and reg AI being for the rushers is an interesting. It's certainly true for someone who wants to enjoy their strategy most.

Your mention of possibly not being allowed to declare at Friendly is the true eye opener for me at least. I also would never have thought of this, but then again it would be rare that I would be declaring war on a friendly player. If I'm going for conquest I've probably declared war on enough people that at least some of them were his friends, etc.. But since declaring at Friendly is something the AI is programmed not to do, it makes me feel I should not do it.

In fact there are probably many many things which human players do, which AIs can't, and which could therefore be considered unfair. For example, AIs probably adon't know how to prechop forests, or build a Globe Theatre Draft city, or create a supermedic etc. etc. - all things a human player will often do.

In any case, obviously what is fair and what is not fair can differ from person to person, as you said, but I still find it really interesting to gather opinions on such subjective things. At the very least I can form new ideas on how I will play the game in the future.

When it comes to people comparing things at a competitive level (HoF play etc.) I think the need for fairness is more important than the need to get a higher score, but it does not mean I will confront those players and tell them so. HoF players who take advantage of certain features of the game that most would argue are abuse of mechanics, may be very skilled in the way they play the game and really enjoy the way they play the game, but I would be inclined to say "Oh those poor AIs!".

I think my problem is that I'm putting way too much character (for want of a better word) into the AIs. Other players will not feel this way and so if I talk about things that I believe are unfair to AIs, they may object to my comments because they feel I am saying they play in a less fair way than I do. I don't wish to take any sort of "fairness" high ground like that - what I intended to do (though not as diplomatically as I hoped) was to discuss the ways in which we each like to be fair to the AI.

When I ask whether one considers it fair to rush on regular AI, I am genuinely asking that question - I'm not implicitly charging a player who says yes with being unfair in any absolute sense. Just as I may wish to say a player who builds the Great Wall on raging barbs is not really being fair, and I might politely tell him that, I will 100% not object with them doing that if that is how they derive enjoyment from the game. If it is likely that describing what they do is unfair will offend them, I'm much less likely to tell them it is unfair.

So I guess with rushing regular AIs, it boils down to the fact that probably the majority of players consider it fairgame, if you will, and most players would be offended if I told them it were unfair. On the other hand, building the GW on raging barbs is more commonly considered fair, and so I'm only likely to offend a very select minority.

So to sum up, it is my opinion that rushing on regular AI (before 1000BC) or whatever is not fair on the AIs, but unless you actually care about AIs there is no reason you should think twice about doing it. Because it is only an opinion, I would hope there is no need to be offended, especially because it could only possibly relate to the way you're playing a single player game in the comfort of your own home.

 

[Olodune]

... because I have never thought twice about "chain-transports" as you called them. The reason I don't believe it's an exploit is because you don't actually increase the "bandwidth" across an ocean, you simply increase the speed of the transporting when there is less to transport. But since the AIs don't use it I can definitely see a case for me to reconsider the way I have used this.

I'm being a little nit picky here but you are incorrect in an important example: national units (missionaries+executives). Since you are limited in the number of active units on the map, the speed you can cross the ocean is proportional to the rate at which you can spread (infest?) the AI with your religions/corporations (before air travel, anyways). Using chaining can clearly be "advantageous" in this situation.

 

[PieceOfMind]

I'm being a little nit picky here but you are incorrect in an important example: national units (missionaries+executives). Since you are limited in the number of active units on the map, the speed you can cross the ocean is proportional to the rate at which you can spread (infest?) the AI with your religions/corporations (before air travel, anyways). Using chaining can clearly be "advantageous" in this situation.

Good point... I never would have thought of that if you hadn't brought my attention to it!

It's also clearly advantageous in situations where getting some troops there faster is more critical than getting more troops there later. In fact you get just as much troops later but some arrived earlier than they otherwise would have. But to you I know this would have been more obvious than the reason you gave.

I have once or twice used this to beat AIs to various islands for settling cities. I'd chain a few galleons for the settlers and garrison and possibly chain a few caravals for the missionary to spread the religion. I feel a bit bad for doing it now.

 

[lilnev]

My personal objection to chain-transporting is not that the AI doesn't use it -- as you point out, there are lots of things that the AI doesn't use, like Medic III and GlobeDraft, and I have no qualms there. It's that chain-transporting feels (to me) outside of the intended functionality of ocean transportation, which is "supposed to" be somewhat slow. The circumnavigation bonus is "supposed to" make your ocean transport faster, rather than making your teleportation machine require fewer components. Again, just how I feel about it.

More generally, my test for fairness is, "Is it within the intended scope of the game for the human player?", not "Is it omitted by, or even forbidden to, the AI?". Declaring on Friendlies, for example, comes up reasonably often in my games. I often don't war until the Renaissance, and if my 2 or 3 immediate neighbors share a religion, I'll adopt it too. When it comes time to expand, I make the choice based on proximity, expected resistance, and land quality. Ruthless, yes, but I think human ruthlessness is within the intended scope of the game.

As always, just my opinions about what feels right to me.

 

[Airefuego]

The reason I wish the AI could rush occasionally isn't so much due to fairness, as the fact that it would make the game more interesting.

Better AI is always better for the game, and ultimately it would be nice if Deity meant EQUAL starting units and production bonuses, but against very smart AIs that make very good calculations for best tile yields, very cunning strategy, etc etc.

Lesser levels could have player bonuses plus less clever AIs - eg, AIs that don't rush...

I'll join the queue for Civ VI now...

 

[MkLh]

The reason I wish the AI could rush occasionally isn't so much due to fairness, as the fact that it would make the game more interesting.

The AI does "rush" (pre 1000BC DoW) sometimes and I find it very annoying when it happens. There is no way to build all the early infra and enough defense at the same time, unless you're on a level below your skills. Even if you can keep their first SoDs away, they don't give a peace unless you give them a city (no Alphabet) and a game is basically ruined.

 

[aelf]

The AI does "rush" (pre 1000BC DoW) sometimes and I find it very annoying when it happens. There is no way to build all the early infra and enough defense at the same time, unless you're on a level below your skills. Even if you can keep their first SoDs away, they don't give a peace unless you give them a city (no Alphabet) and a game is basically ruined.

Yes, this happens. And even though you kill wave after wave of their units, they still refuse peace and you're forced to mount a costly counter-offensive to take a few of their cities that might be very far away. The 2 of you get left behind eventually and the game is unplayable.

Does this make the game more fun? Count me out.

 

[PieceOfMind]

Yes, this happens. And even though you kill wave after wave of their units, they still refuse peace and you're forced to mount a costly counter-offensive to take a few of their cities that might be very far away. The 2 of you get left behind eventually and the game is unplayable.

Does this make the game more fun? Count me out.

Exactly. This is why I won't rush the AI usually - if they did it to me, it'd ruin the game, so if I did it to them, it'd ruin the game for them...

 

[aelf]

Exactly. This is why I won't rush the AI usually - if they did it to me, it'd ruin the game, so if I did it to them, it'd ruin the game for them...

The issue is not so much with rushing per se, but the fact that the AI rushing the player tends to end the game, which is not exactly the point. If they rush successfully, you're dead, of course (which is quite alright). If they're unsuccessful or not so successful, they don't seem to know when or how to stop, which may make the game unplayable (which is bad). This way, AI rushing is often just wasting your time. The occasional loss is fine, but the protracted ending is just frustrating. And this is because the AI's rushing strategy seems to be rather tacked on.

I always see it this way: Rush or be aggressive if you think it benefits you, but play at a high enough difficulty such that you pay the price for rushing. You may conquer one AI neighbour, but that might not mean you can win the game. The price may just prove to be too high. You don't need the AI rushing you or to abstain from rushing to make the game challenging and fun.

 

[Airefuego]

\If they're unsuccessful or not so successful, they don't seem to know when or how to stop, which may make the game unplayable (which is bad). This way, AI rushing is often just wasting your time. The occasional loss is fine, but the protracted ending is just frustrating.

I agree. I would add that a 1000BC attack is not really a 'rush', and a constant trickle of attacking units after an unsuccessful first wave is definitely not a rush.

I maintain that if the AI was GOOD at rushing, and did it occasionally, it would be good for varied gameplay - and would be a better way of making higher levels challenging than, say, just giving big production bonuses...

 

[MkLh]

I agree. I would add that a 1000BC attack is not really a 'rush', and a constant trickle of attacking units after an unsuccessful first wave is definitely not a rush.

I maintain that if the AI was GOOD at rushing, and did it occasionally, it would be good for varied gameplay - and would be a better way of making higher levels challenging than, say, just giving big production bonuses...

Actually, if the AI would rush more often and more efficiently, it would make an early rush the only viable starting strategy every time you have a neighbour other than Gandhi. There's no point to start a builder game if there's a significant risk that the AI will (c)rush you.