@SorenJohnson: a couple of bugs. Firstly, the bug reporter in the in-game menu isn’t working (unable to submit).
Secondly, this event fired as I was conquering the city, not as it was being attacked. There were no nearby enemies. The generated troops were in Marad, on the opposite side of the map.
Sadly it looks like the AI doesn’t know what to do with island maps, even with all the excellent AI work in other areas.
Good choice for a peaceful/builder type game, I suppose.
I suspect this is the same situation as the AI not knowing how to use siege units. I haven’t seen the AI do anything with naval units, and I’ve been almost exclusively playing the Seaside map, which seems to put Tribes out on islands off the coast (I don’t know if this is how the Seaside script works or if it’s just been a coincidence). As near as I can tell, these are pretty much mine for the taking because of the AI’s inability to play the naval game. Is anybody seeing anything differently?
I’m also worried about the AI and religion. I’m setting up games now to specifically avoid World Religions so I can see how the AI handles them. Does it build religious buildings? More importantly, does it slot theologies to make its religions more powerful? I hope so, but I haven’t seen it yet. Have any of you guys seen the AI leverage World Religions?
My concern is that there are whole gameplay systems the AI simply can’t handle. If that’s the case, I can feel gravity starting to pull me down the far side of the Chick Parabola. :(
-Tom
I played an Archipelago map. I saw precisely one successful AI naval invasion, against a tribal village. I did see it anchor some biremes about to generate bridges to enemy cities, but other than that once it never actually sent any units. Perfect for a safe builder game, but definitely not what’s really needed from the AI.
The thing I’m most concerned about right now though is the AI’s ability to improve its lands. I did a late game attack (on Magnificent) and my target’s cities didn’t even have all their special resources improved, let alone many other tiles. I’m not sure how they’re supposed to remain competitive, really.
Yeah, that was a double victory on turn 89 on Contemplating Competition (the Strong) difficulty, with like 13 friggin’ cities at the end. The AI just never went anywhere. I wonder if it needs to be all-but-forced into taking Exploration instead of Epics on watery maps, and it definitely needs to understand bridges better.
I saw the AI doing a lot of pointless naval unit dancing in this game (I had vision of Persia’s capital after they declared on me from an event, not really sure why but anyway). It definitely built the boats, just seemed not to know what they were for.
Boy, this would be a big one if the AI isn’t approaching religion effectively. Religion is (or can be, anyway) key to discontent management and science, and can generate a ton of culture as well.
One thing I’ve now noticed a couple times is a kind of weird interaction between what I assume is either a soft or hard cap on characters in your empire and the way new characters are spawned. I have seen a couple times a very middle-aged court (including family characters who are serving as governors/generals/heads) finally start dying off, at which point they die off in freaking droves, and then 3-7 turns later a big wave of youth finally show up after getting no youngsters in the system for years and years.
Dunno if it’s a difficulty thing where death chances are turned down and you end up with these waves of characters dying and spawning or what, but it at least seems like it’s a thing that’s happening. Could very well be the ol’ human pattern-recognition algorithm misfiring too.
I really hope Old World isn’t relying on the model I mentioned earlier, similar to a racing game where you start in last place and have a certain number of laps to catch up. Seems to me the worst case scenario for the AI in Old World* is that it’s counting on starting out with the huge advantage of developed cities and then stumbling along while you spend your 200 turns trying to catch up and eventually pass it.
I would love to know more about how population spawning and death work in the game. I asked about character deaths in gameplay questions thread, but didn’t get any responses.
-Tom
* which is still pretty good compared to the average braindead 4X
I think that asterisk, or the remark attached to it, belongs at the beginning of your commentary. And “pretty good” is an understatement of British dimensions. It seems Old World is suffering from an excess of love here.
Just to clarify, I’m talking about its AI, which I’m less and less impressed with the more I play. Even if my concerns bear out, I think the AI is “pretty good compared to the average braindead 4X”. But I’m more reserved than some others in these threads for how I’m not ready to sing it any hosannas yet.
-Tom
The AI will definitely build religious improvements, and I have seen it establish theologies but it’s possible that doesn’t happen often. (The issue is probably making sure they have enough civics on hand.) The AI used to anchor their navy a bunch… in fact, the problem we had a few months ago was the AI overbuilding their navy, so it could be there are some issues there. At any rate, the AI should/will be able to handle all of these unit actions, and we are still digging out of launching issues, so it’ll all be handled in time. Our approach is that the AI is supposed to play the unit side of the game as well as the player, and we will keep updating weekly. This won’t be one of those things where you have to hang your hat on some patch that might or might not happen next year. Going to send that save to Alex as I am curious why those three catapults went to the eastern city - although I was afraid from scanning the thread that the AI was literally not using their onagers, which at least is not the case.
For characters and families, by the way, the game adds characters periodically if there are less than a certain threshold which goes up per city that family has. Based on pure chaos, they should be of all different ages by the mid-game but would be happy to take a look if they do have some sort of weird generational resonance.
Yeah, I lost a unit to onagers due to foolishly advancing without getting scouting info first. There were four of them unlimbered on hills on the near side of a mountain range.
Now that did mean that I was able to swarm them and wipe them out the next turn. If the AI had been playing like a player here, it would have realised my army was too large for it to handle and retreated all its units to the far side of the 2-hex wide mountain pass that protected its core cities. I suspect a determined defense there with all those onagers could have caused me some difficulty. Didn’t help that it lacked any roads through the pass so couldn’t manoeuvre particularly quickly.
Indeed. I wasn’t clear. I was taking issue with your judgement on how the computer opponent plays. The computer gives me a tough fight every time I play it (typically on Noble difficulty). If we’re judging this coding in absolute terms then I would say it’s very good indeed. If we’re judging in comparitive terms then that judgement shifts to wildly excellent.
I think the recent focus of this discussion must be helpful for Mohawk since instances of AI weaknesses are being flagged. The ships thing and the onagers thing in particular. But “pretty good” is damning with faint praise. It’s far better than that.
The question is how much of the difficulty is due to the AI’s starting advantages, and how much is due to its skill.
There’s no doubt that the AI is pretty good at a lot of things. It expands its territory really well in the early game, and is very good at picking off injured units. Going down in competence a bit, tactically it’s capable of using overwhelming force to win battles and conquer cities.
All of this is far ahead of the AIs for other relatively recent 4X games, so I don’t mean to be overly down with recent posts. It’s more that it’s interesting (and a bit disappointing) to be picking holes in the behaviour of a good AI. Far surpasses staring in bewlderment at an AI that doesn’t pose the slightest threat.
Diplomacy is a much more interesting design than in previous games and the AI seems fine at it, if occasionally hamstrung by random events.
Where I do feel the AI is losing out at the moment is in strategic concerns (plus ça change). It seems to build a reasonable military but doesn’t do a good job of building enough improvements (should it be making more workers?). This leaves it behind the player in the long term in both resources and science.
All of that means that an early war with an AI nation is a severe threat to the human, generally forcing some painful tribute payments to avoid it. Once the player has risen out of the starting hole, the game feels pretty much won though. That does require capturing territory though, so if you don’t have space to expand without going through an AI nation, that’s probably pretty difficult.
Starting position remains therefore a very big modifier to difficulty. My most recent game was on a random map, and I rolled an arid map full of mountain chains (though with a very large central sea, so I’m not sure it was arid plateau). As it happened, I had one neighbour who made things difficult early on, but importantly I had a huge backline area full of barbarians and tribes. I built a reasonable military then sent it all off conquering the minors, paying heavy tributes to my neighbours to avoid war and buttering them up wherever possible. By the time I’d conquered everything in the backlines, I had more cities than anyone else and a far better economy and was finally able to secure my other border and feel certain the game was mine.
But yeah, if I’d had another AI nearby taking two thirds of those backline tribal spots I would have been in a lot of difficulty.
Entirely fair. And I would add it’s not just the AI who gets “occasionally hamstrung by random events” : )
You’re not wrong about that. :)
I was meaning that specifically the ability of the AI to have a consistent diplomatic strategy towards the player must surely be affected grossly by events that can start or end a war suddenly. This could mean that the diplomatic AI is actually a bit predictable under the hood, and the random events serve to spice it up.
From a roleplaying perspective, the biggest problem is that the event will say something like “the Persians have been preparing for war and only tribute can prevent it”, while in game the Persians have their military miles away and aren’t ready to invade you at all. Given that I think that event only happens at Much Weaker anyway, it probably just gives the player a stay of execution.
I agree. I play on Glorious level and win about 1 in 4 attempts. There are so many interlocking, well-designed systems in the game that, even after a year of on-and-off play, I am still learning new things and finding the gameplay totally absorbing. In EA, Mohawk made many changes based on player feedback and I’m sure that will continue. I can’t imagine there could be a more committed and responsive dev team.
You can change the development setting so they have one city like a player. They do pretty well by 4X standards since as you say, they expand really well.
Any other 4X game, the AI would be playing with pretty substantial resource advantages throughout the game. Old World AI operates at quite a bit of a handicap compared to that. They don’t get massive resource, tech, and happiness boosts like they do in the Civilization series.
By the way, we are going to increase the map size a bit for each Development level so that the larger AI empires don’t crowd the player as much.
I think that’ll be a good change.
While we’re talking about positive changes, any possibility of porting over the Center Bias option to the MP lobby?
What would you like to use it for? (Didn’t know it wasn’t there…)
In SP I can toggle whether or not center bias is enabled for the player. In MP, it’s always enabled. My typical setup is a 6-player map (2v2v2 or something when playing with a friend) and the player is always placed in the middle. I like some variety in starting locations where sometimes I’m in the center, other times I get a corner. I like exploring and figuring out where I am in relation to other players, rather than starting out every game knowing the world is revolving around me. :)