AI Players' way of playing is very suspicious!!!

[Lord of Civ]

Thats it people, i will leave this thread for you all to discuss and i shall ignore this troll. I just don't wanted to ruin or derail the thread any further.

Another thing to remember: YOU DON'T HAVE TO CREATE A CONSTITUTIONAL LAW ON THIS THREAD TO TREAT AI THE SAME AS HUMANS BECAUSE AIs ARE ALGORITHM CODES. WE HAVE TO FOCUS MORE WITHIN THE GAME LIKE WHAT BREAKS THE STANDARD GAME RULES RATHER THAN SAYING A HUMAN CAN PRESS A BUTTON AND AI CANNOT DO THAT. PRESSING THE OPTION BUTTONS IS NOT CHEATING BECAUSE IT DOES NOT BREAK THE STANDARD GAME RULES WITHIN THE GAME. EVERTIME YOU START A GAME YOU MUST GENERATE THE MAP ONTO MONITOR SCREEN THEREFORE RESTARTING IS REGENERATING. AND.... ITS NOT CHEATING... ALSO AIs WERE EXECUTED WHEN PRESS A BUTTON. ITS A BUNCH OF EXECUTED PROGRAMMING CODES AND NOT A CHARACTER. WHY WAISTING TIME ON THIS NONE SENSE?

Please feel free to post about AI being suspicious and the infamous AI Magic Eyes. Please stop blaming me since it would not make this thread discussion very constructive but you can disagree on my opinion.

hehehee!! This thread is like the OReilly Factor Show on Fox News. I was being Bill OReilly.. hehee!!
Actually, this thread might turn into flame war discussion like on that Show. hehee!!
At least, i am a good thread creator, today i got 1500s viewers!! heheehe!!! This is funny... This show would be on for months while im gone, for sure!!! JUST KIDDING....heehehee!!!

Thanks, and see yah!

PS: SORRY FOR THE DOUBLE POSTING, CAUSED BY TECHNICAL PROBLEMS.

 

[bitplayer]

You know I dont really understand some of you guys. The level I play on at atm is hard enough- for me- to know that if I get a bad start Im doomed. As in game ova. Done for. Not gonna happen. Nu uh. No way. No how.

Either I have a chance at winning or I dont.

Crud lemme put it another way. Last night I played a mp with two of my buds. Small map with us and ai. My freind got copper in his fat cross. I, on the other hand, got a coast to my n s and e after scouting out everything.

Unfotunately he was vicky and my closest neighbor. He settled his 3rd city right next to me with his fast settlers. So I dropped the warriors bomb on him and raised two of his cites after whipping my 2nd city.

Unfortunatly I couldnt get past his defense on the copper- after a long azz archer/warrior fight- and he eventually started to churn out axes. Which I couldnt counter.

So I lost a desperation gambit that was doomed to begin with.

Nobody wanted to play after that as I was out of the mix. We like to play these things over the course of days in our freetime btw so a weeks worth of goofing went as soon as I did

Which is my fault as I should of called a halt to it as I know better.

I sort of see it as the same in sp. If I cant win why not restart? Its not like Im sandbagging on the level Im playing at or anything. Saves time and trouble as well.

Anyways as far as the op goes. Meh. If that is how the ai is then that is how it is. Ill compensate eventually.

 

[Aussie_Lurker]

You know. In an MP game I once lost my beautiful Capital/Holy City to the barbs at a time when I still only had two cities. Did I give up? No, of course I didn't, because for me the challenge is what makes a game good. If I am in a position where I am steamrolling all my opponents, then it is usually there that I quit. The same goes for bad starts. Even if I know I will probably lose, I struggle on regardless to see how well my empire goes in spite of a bad start. Heck, I didn't even know there WAS a regenerate map option until a few months ago .
Anyway, I have otherwise grown tired of this incredibly boring thread. I didn't think 'Magic Eyes' were real, but I have now seen enough evidence to say that it does exist-albiet not as a 'cheat'. I think it needs to be changed to make it-at the very least-less obvious (and less exploitable by human players). That was the point of the OP, I have said my peace on that point, and now I am going back to some intelligent conversation in the Creation and Customization threads.

Aussie_Lurker.

 

[DaviddesJ]

I sort of see it as the same in sp. If I cant win why not restart? Its not like Im sandbagging on the level Im playing at or anything. Saves time and trouble as well.

That makes perfect sense to me. Just because it's something the AI can't do, that doesn't make you a bad person. Editing the map to give yourself 1000 Modern Armor doesn't make you a bad person, either. Everyone should decide for themselves what is fun.

 

[DaviddesJ]

I was being Bill OReilly.. hehee!!

Finally, something we can agree on.

 

[Mango]

Lord of Civ should be banned for his English.

 

[cabert]

5 pages for this seems a bit too much...

All this because people think of the various AIs as persons...
Their isn't really any AI in the game! It's only a bunch of algorithms.
The different "characters" are coded, use (and don't abuse like a human does) the various algorithms to do what they are coded for. That's it.
There is no "logic" involved in the "characters" we call AI.
If there was, you couldn't win a diplo win.

The AI can't cheat, because it doesn't care about winning.

 

[Roland Johansen]

What the heck is this Roland saying??

I wanted to return to the original discussion about foreknowledge of the AI and I doubted that you were interested in the discussion as you didn't respond to any of my counterexamples or explanations of AI behaviour. Note that I gave these examples and explanations before the discussion about cheating started and repeated them later, to no avail.

For the record: I never called anyone a cheater in this thread. I even tried to diffuse that discussion by comparing restarting to 'making the game easier' which is a far less sensitive term than cheating.

Prove me wrong by discussing the foreknowledge of the AI and lets stop this discussion about cheating. Everyone has a different definition of cheating. Let's just accept that and move on.


On the original topic about foreknowledge of the AI.

I do think that the path finding algorithm uses the information of units that are not in visible range. This can lead to movements of the AI that seem to indicate that it knows that units are present in areas it cannot see (in choke points for instance). But it works the same for the human player. So I would call this a game mechanic and not foreknowledge of the AI.

The AI seems to know if you defend your resources with units and will react to it accordingly (to pillage or not to pillage). This is foreknowledge of the AI that is not available for the human player. I think this has probably been a conscious design decision by the developers to compensate for the lack of intelligence in the AI. It would have been better if it had been solved in a better/smarter way, but it is very difficult to make an AI seem to behave intelligent.

I do not think that the AI has foreknowledge about your units in your cities. I gave a counterexample to this in my first post of the thread and explained how an apparent knowledge about your city defense could be explained. I'll just copy paste it here, no need to type it again:

The AI also does not have magic eyes about your city defences. It does spread its religion to your cities and if some of your cities have a religion of which they have the holy city, then they can see the city, its defences and the immediate surroundings. It will also use spies to check out your cities. And last but certainly not least, it will actively scout your lands; ships moving along your borders, units of nations with whom you have an open borders treaty will enter your lands to find out where you have the most units.
This scouting information will of course not be up to date when they attack. I've noticed this once when a nation opened a surprise naval invasion at the tile next to the staging area for my own invasion. 10 turns earlier, when the nation had scouted the area, it had been a moderately defended area and the 15 unit naval invasion could have caused me serious problems. Now, it only succeeded in postponing my own naval invasion plans a few turns as my 25 unit stack needed time to recover from the damage it had taken. If the AI had really known about my city defences, then it would not have attacked my most heavily defended city.

The AI is also programmed to combine a frontal assault land attack, with a backdoor naval invasion in the hope that your backdoor cities are weakly defended. Of course that is often the case, so that is just good programming.

Last but not least I would like to comment on some (fore)knowledge that the human player has and the AI doesn't have. I posted this earlier so I'll just repeat it here.

The human player can see how the AI feels about him (all the pluses and minuses in diplomacy and the attitude of the AI player) while the AI has nothing similar towards the human player. Now, the AI actions are not completely determined by the pluses and minuses in diplomacy, but they are extremely important in its actions. The human player can plot and scheme to destroy the AI player while keeping a friendly face ingame towards the AI player.

I personally think that the AI was programmed quite good. We have some extra knowledge and the AI has some. Obviously, the AI is a lot less intelligent, but still it somehow works and can give you a challenging game (although some people need to give the AI the difficulty level bonuses).

Does anyone agree or disagree about the amount of foreknowledge the AI and the human player has?

 

[The Great Apple]

As far as I can tell from the code the AI does have foreknowledge of a cities defences, but only if the unit/stack it is currently looking at is within a certain distance of that city. This distance can vary depending on the algorithm that is running.

The largest radius is a 5 * (unit moves + 1), although it's much more likely to be 1, 2, or 3 * (unit moves + 1). It's a bit weird as not all of the units have the information, only those which are close - and they can have it with varying range. The same is true for general attacking.

 

[bitplayer]

That makes perfect sense to me. Just because it's something the AI can't do, that doesn't make you a bad person. Editing the map to give yourself 1000 Modern Armor doesn't make you a bad person, either. Everyone should decide for themselves what is fun.

Im sorry did I say anything about editing the map? And yes, fyi, I play this game for fun but I also play it to win. Otherwise I wouldnt be playing on levels where I know what an instant loss is when I see my starting pos.

Talk about a back handed compliment. No wonder the op got a bit peeved.

 

[Roland Johansen]

As far as I can tell from the code the AI does have foreknowledge of a cities defences, but only if the unit/stack it is currently looking at is within a certain distance of that city. This distance can vary depending on the algorithm that is running.

The largest radius is a 5 * (unit moves + 1), although it's much more likely to be 1, 2, or 3 * (unit moves + 1). It's a bit weird as not all of the units have the information, only those which are close - and they can have it with varying range. The same is true for general attacking.

Thank you for your good contributions to this thread. Your information from the SDK (now and in post 38) is the biggest contribution this thread has received. Good work!

This explains why an AI can move its troops to another attack position just before it is in viewing range as the OP reported occurred to him. I haven't noticed it myself, but I'm usually attacking so that is not that strange.
I personally have no problems with this foreknowledge as the human player can 'guess' far better than the AI and the range of foreknowledge is pretty small. Still, it would have been better if they could have programmed a smart AI not using this foreknowledge. Probably too hard.
Is the strength of the attacking stack and the defending stack and city (cultural) defence and nearby defending troops all a factor in the decision to find another softer spot in the defence?

Note that this does not explain why an AI tends to sent an expeditionary force to your backdoor cities as the unit viewing range is too small to detect lesser defences in that area. As I have read from a Firaxian, it was purposefully coded in the AI to try the back door. If it is wide open, then it will profit from that.
You can even see the AI gather its forces for a naval backdoor invasion if you happen to have line of sight in the AI city where the forces are being gathered.

I would like to know how my example fits in all this. The Great Apple, can you explain why the AI sent a force to my strongest city? Does it have to do with some specifics about the naval operations? I'll quote myself again, so that you won't need to look it up.

This scouting information will of course not be up to date when they attack. I've noticed this once when a nation opened a surprise naval invasion at the tile next to the staging area for my own invasion. 10 turns earlier, when the nation had scouted the area, it had been a moderately defended area and the 15 unit naval invasion could have caused me serious problems. Now, it only succeeded in postponing my own naval invasion plans a few turns as my 25 unit stack needed time to recover from the damage it had taken. If the AI had really known about my city defences, then it would not have attacked my most heavily defended city.

So why did the AI not divert its forces to another city of mine? Did it make the decision 10 turns earlier and does it not reconsider the decision? That would be very different from Lord of Civ's experience where the troops diverted to another city every few turns. The difference is that in Lord of Civ's case it was a land invasion and in my case it was a naval operation. Maybe the landing point of a naval operation is not reconsidered after the moment the ships leave the harbor. Could that be it?

 

[cabert]

So why did the AI not divert its forces to another city of mine? Did it make the decision 10 turns earlier and does it not reconsider the decision? That would be very different from Lord of Civ's experience where the troops diverted to another city every few turns. The difference is that in Lord of Civ's case it was a land invasion and in my case it was a naval operation. Maybe the landing point of a naval operation is not reconsidered after the moment the ships leave the harbor. Could that be it?

I think it's a "goto" issue. The naval units don't face resistance, so landing is safe = no turning back.

 

[The Great Apple]

It's quite complicated.

Every turn the AI updates it's targets on each continent with a 1/8 chance. It only has one target per area (continent/sea... but not seas in this case). On updating there is a 1/3 chance there will be no target. This target is selected using quite a few factors (population, coastal, culture, wonders, resources, closeness... lots of stuff). This is all at the time the decision is made - it's perfectly possible this could stay the same for many turns.

This target city is the AIs target, and units which don't have anything better to do (which could include attacking local units, local cities, pilaging, healing...) will move to attack this city.

The backdoor invasions are handled differently. That doesn't take targets into account as far as I can tell...

 

[Roland Johansen]

It's quite complicated.

Every turn the AI updates it's targets on each continent with a 1/8 chance. It only has one target per area (continent/sea... but not seas in this case). On updating there is a 1/3 chance there will be no target. This target is selected using quite a few factors (population, coastal, culture, wonders, resources, closeness... lots of stuff). This is all at the time the decision is made - it's perfectly possible this could stay the same for many turns.

The backdoor invasions are handled differently. That doesn't take targets into account as far as I can tell...

It sounds complicated indeed. It's probably difficult to explain in words what you have read in the code.

If I understand it correctly, there is a great chance that the AI will not turn back when it sees a large force in and around a city (whether it sees it like the human player in vision range or in the AI way just beyond normal vision range). But it might turn back if the original target has become unattractive. It is good to govern this decision with chances so that the AI doesn't become too predictable.

If the AI cannot see a target because it is outside its AI-extended vision range, will it then use its last documented information about the target (from scouting)?

I'm very curious how the AI handles the backdoor invasions. If you can tell something about it, I would be gratefull. Apparently, the information that the backdoor invasions are coded in specifically is not a lie from Firaxis.

This target city is the AIs target, and units which don't have anything better to do (which could include attacking local units, local cities, pilaging, healing...) will move to attack this city.

I've seen the AI sent out an attack force while I was conquering him (not necessarily a bad move in all cases, but in this case the units could have been used better). The attack force was on a different side of the front as my attack force, so I guess that it didn't consider my troops to be 'local'.

I must say, it is very interesting to hear about how the AI 'thinks'. Many people don't realize that it is totally different than the human way (one can hardly call it thinking). Thanks again for the info.

 

[wioneo]

I personally have no tolerance what so ever of whining people like Mr. "Lord" of Civ here, so, excuse me if I seem rude. I am also not a "nice" guy, however, this one chick called me crazy. Though I don't know if I would constitute that as being a "mad man".

Stop crying! The AI's DO cheat in quite a few ways that I know of. For example, they get free units, they research and build faster, they have more diplomatic options, they capitulate to each other more often. Al of these things(other than the diplomacy and capitulation) are to even out the obvious, or at least expected superiority of the human player. If you can't use your ability to think to at least sometimes defeat the AI, then you are pathetic.

The one thing that you said correctly was that retiring and then playing a new game is not cheating. it is just quitting like a pathetic loser, or at least that is my opinion. Regenerating the map, however, IS cheating. The AI's can't do that, can they? Using the World Builder is REALLY cheating(unless you are testing or actually building a world). By the way, the next time that your capitol starts in the middle of a desert, maybe you should take a closer look at the tiles and notice that every single one of them is a flood plain. I have yet to experience a start with a single desert(non-flood plain), ice, tundra, or mountain tile. If you do, kindly post a screenshot.

I have noticed that the AI seems to know about my cities defenses/worker positions a bit more than they should, and that is an annoyance which I happily exploit. Try leading an enemy stack around in circles as they chase after a worker. It is quite amusing. Another thing, did you notice how the AI's have well defended cities? Maybe you should try that. If I see little ol' Mansa sitting on some far away island with a weak looking coastal city, it will burn(or not, it really depends on the city).

Well, I'm done with my rant, and now you can go back to complaining. Have a nice day

EDIT: On a lighter note, your foolishness has inspired the exchange of some very interesting information. Give yourself a pat on the back

 

[Ceritoglu]

I play Texas Hold Em quite frequently - you learn a lot about people and human nature from that game. One thing I've noticed, is that players who lose rarely blame their own gameplay. They often cite unlucky cards, sometimes claim foul-play and at other times will explain that the opposition was merely lucky. Yet it's only the best players who sit down and analyse their own gameplay, see how they could have avoided those defeats. This is not only applicable to poker, but to life in general. Rather than focusing on the AI's marginal advantage and letting that stop you from developing as a player, find out ways to minimalise this advantage.

The AI may have obtained some unfair intelligence on you, but it's the realisation that if you had a universally well-protected nation that the AI would have directed its attention elsewhere - if not it would have had a tough campaign in your territory. It's foolish to complain when the AI uses the advantages it has been given, in the same way that it's foolish to complain that human players use the advantages they've been given. If the programmers disapprove, they'll patch; if they don't, focus on getting around the problem rather than complaining about it and refusing to adjust your gameplay. It's the ability to adapt which sets apart the great from the mediocre.

 

[JBConquests]

wioneo, I respectfully disagree.

I don't usually post but couldn't pass this up. Regenerate map is a very good and handy tool and is certainly not cheating. Here is how I use it, if I am playing on a skill level that I at least have a chance of winning (Prince for me) then I don't regenerate because it is possible for me to win on Prince with a bad start. (I may lose, but that can be fun too...)

However, If I go up to the next skill level, I admit I have a very difficult time winning at that skill level. (I rarely do - I need more practice or more good strategy articles) So, I regenerate my map until I get a very good starting position. I need to do this to even have a chance to win at this skill level. Eventually, once I figure out how to play at this skill level so I have a decent chance of winning, I will quit regnerating my map again because bad starts are as much fun as good starts - even more fun actually. But, there is no point in playing a map that you know you don't have a chance of winning!

 

[wioneo]

But, there is no point in playing a map that you know you don't have a chance of winning!

I too disagree, if you have to regenerate the map to win, then why not just stay at prince? It's not like anyone will care what level you're at. Or at least I don't. I personally play on noble and almost always win but refuse to increase the difficulty level because noble is the "medium" or "normal" level. I always try to play at normal levels... Anyways, I really learn more by losing, or starting to lose, than just dominating the entire game.

 

[JBConquests]

I win most of the time at prince so to me, I'm getting bored. I like to play at skill levels where I can lose 50% of the time. (on average) But, unfortunately, I move up a skill level, and I almost always lose - if I stay at prince, I win 90% of the time. I hate starting a game knowing I have a 90+ % chance of winning.

I guess to me, playing on a higher skill level (where maybe I don't quite belong yet) is part of the fun.

That is also how I advance to a new skill level is by playing it and getting my clock cleaned.

 

[DaviddesJ]

If you are "between levels", another thing to try is playing with a "bad" leader, or more difficult map type, at the lower difficulty level. Or playing with a "good" leader, or more favorable map type, at the higher difficulty level.