Many players have complained that the monsters treat AI differently than they treat the player. Brad says that on challenging, the monsters treat the AI exactly the same way they treat the player.
Many players claim that the AI dogpile on the player without rhyme or reason. Brad says that this only happens on higher difficulties.
Brad may be right. I would not know, I seldom play Expert, and never below that.
This is why I am going to play ironman on challenging with what is considered the worst sovereign, Queen Procipinee. When the game is over (or has become too boring) and I will go back to my saves, and take a look at what the AI is doing (someone will have to tell me how, but I know it can be done)
By the way, do you have to do something before you start the game to be able to look at the AI moves from saves? If so, I would really appreciate it if you tell me before I start. (Which will be about 30 minutes from the timestamp of this post)
The experiment has been concluded, and the results can be found here.
Is this a patch for the difficulty rounding error? What will it affect?
So Tuidjy has done all this work and research and it is all invalid because you think the AI does not cheat?
Here is a reminder on the definition of the word cheat.
Anyways back to the subject.
Tuidjy I sympathize with the crashes. They come fairly quick when you reload from a save too many times. Obviously with your research into alternate turns this becomes common
With your research on the pioneer I have somehting to add. I had a pioneer do one of those stupid auto plot moves to a square I did not want to in my last game. There was a dragon that actually ignored that pioneer. I have seen this a few times beofre it is like it takes a turn or two for the monster to "wake up" and then once it does it will chase anything nearby. I would imagine that same circumstance could apply to some AI units as well.
Actually, I have NOT started alternative turns yet. But I am at fault. I used my XP laptop because my Win7 desktop was crunching numbers, and got tempted to use firefox while the game was running. I was more pissed off at myself losing the fight with Verga, and then for sort of save scumming, even if unintentionally. (Because on top of everything, Dodge I is a damn good pick)
It should affect most of the "challenging" related bugs you reported.
Can you be a bit more specific? If it covers everything, then I should just drop this experiment, the job is done.
But does it cover any questions about roaming monsters giving AI pioneers a free pass? Does it affect how long it takes Big Beast to wake and whom they target?
This rounding caused the AI to cheat because it was treating the AI like an expert AI instead of a challenging AI. So, the aggressiveness of the monster should treat each pioneer the same. I think Tuidjy that you should grab his private build and see if it has been fixed.
This would be a good test as you can still use your save games here and see if it did fix it. You know re-load and check what's going on with this build, then switch exe and see if the same results are there.
You know, testing like this is a lot of tedious tedious work. In fact I've worked with some professional QA testers before, and even they are seldom as thorough as Tuidjy in testing features. And he has been doing it for free.
I played some game with the 1.0 build, and I noted that the monster seems to give a free pass to pioner in general, not only to the AI (I walked freely with a pioneer in a moster's infested area, and get attacked by the same monsters when entered with an army).
Vice-versa, in my experience the monster target the AI in the same way of human player.
Noted because, in my last game, Resoln stopped to send troops against me, and when I sent my sov army to investigate (and crush Resoln), I saw a monster army (the black golem monster + a couple of savages Juggernauts) happily engaged in destroying army after army of Resoln troops.
Yes, I know that this is a lot of tedious work. But he has the save files and he can definitely finish this off to see if this has been fixed or not.
The point was I didn't want him to feel all this work was wasted, when he didn't even get to the point of going back and testing what he wanted to test. It's all up to Tuidjy whether he wants to do this or not. I for one am not going to force him to test these things out.
I applaud Tuidjy on his work and he makes the game better when the devs listen and try to solve the problems he is pointing out.
It's not wasted, we found the rounding error. And I'm doing the saves now. (I wiped out Karavox in a few turns, everyone but Altar is allied to me, Altar has fallen slightly behind, and I do not feel like leading Procipinee in wiping out a human faction, so I'm calling it a won game)
I'm doing the save on 1.00, then I will do the most blatant discrimination tests with the next version.
I've found after a crash or two it's ususally benefical to do a full shutdown and restart of my computer before replaying.
I'm running a fairly high end gaming Laptop (2 yr old Asus G73Hj with the 1.6GH overclocked to 2.8 Quad i7 and the ATI 5870 card ) I shouldn't be running into memory issue but my guess is thats whats happening.
It may be the graphics card conflicts... the latest ASUS BIo and the ATI graphic drivers seem to have created an issue that pops up occationally (can't play online streaming video without the screen breaking up.. havn't had time to work out the solution yet) so far it's annoying but not much worse than most other games I've come across as far as stability goes
Here we go. The results of the experiment.
Ways in which the AI is playing by different rules,
or is being treated by monsters differently than the player.
(on Challenging difficulty, in version 1.0)
* The AI can see the map without scouting.
Confidence: High
Reasoning: constant flow of pioneers towards valuable tiles or resources that the AI cannot possibly have scouted, ceasing as soon as the player claims the tile/resources. Observed three times.
* The AIs can settle where the player cannot.
Confidence: Beyond Doubt
Reasoning: Observed.
* The AIs receive bonus essence when they settle a city
Reasoning: Observed
* Monsters are less likely to attack an AI army than a player army
Reasoning: On two separate turns (83 and 104) there were at least one player army and one AI army next to monsters. In one case, it was a stronger player army, and a weaker AI army. In the second case, it was an average player army, one weaker AI army and a stronger AI army.
In the first case, the player army got attacked 6 times out of 6 (7/7 if you count the canon turn)
In the second case, the player army got attacked 5 times out of 7. (8/10 if you count the canon and the two crashes on that turn) The weaker AI army got attacked three times. The stronger AI army never got attacked.
* AI rushes production without paying for it
* Monsters target player cities more often that AI cities
Reasoning: Every single time that a monster lair containing an army stronger that a city garrison was disturbed by the player dominion, after 4,5,5,3 turns, the monster headed towards the nearest player city. In the replays, when the city did not get reinforced, the monster attacked in 5 out of 8 cases as soon as it reached the city. In two more cases it attacked a few turns afterwards.
In the replays, I saw at least 7 instances of monsters staying in their lairs 10 or more (up to 30) turns after their lair was covered by the city's influence. In only two cases did monsters destroy AI cities. In one case, the city was built in the path the monster was already traveling.
Note: This was by checking saved turns from the canon play through. As soon as I removed the fog of war, the monsters seemed to become active much earlier. Monsters would also become active if a player unit approached an already disturbed lair (the player unit would be two tiles away, but the lair would have been in AI dominion for a while.) If I were working on this game, I would check into FoG mattering where it should not.
* AIs gang up on the player in war.
Confidence: Low
Reasoning: I was consistently near the top of the pack. No one attacked me. The AIs ganged up on Kraxis as I was crushing it. If the AIs gang up, it's on weaker nations, not on the player specifically. Of course, with all the ways the AI gets an advantage, it's easy for a player to end on the bottom.
* AIs get (at least) research bonuses
Confidence: Average
Reasoning: Altar did too well against me. I had more cities, I was developing them pretty well, prioritizing research, and I had conquered extremely nice conclaves from Resoln, Yithril and later Kraxis. According to the debriefing graph, I had passed Altar's research around turn 100, and never looked back.
Still, if I started a trade, and traded all tech with Altar (both theirs and mine) a few turns later Altar would have twice the tech to trade, even if I would first get all the tech from Guilden and Tarth (paying only resources for it) Altar and I were both researching civilization.
I may be missing something here, which is why my confidence is only average.
* Monsters seem to be seeded around the player but not around the AIs
Reasoning: When I reveal the map, I do not see any really dangerous monsters anywhere but around my position. The only dragons and slag are either part of wildlands, or within 12 tiles of my starting position. It may be completely random, though.
Astounding QA Tuidjy. I've been suspicious of FOW issues for a while too.
Tuidjy merits free Stardock games for life.
Seriously.
AI ganging up on weaker players seems like the right conclusion from my experience as well. People are just too focused on themselves to realize it.
But the last 2 doesn't seem quite right. I will often out research the AI without much trouble on challenging (without inspiration - and I usually only build 1 conclave if any at all)... so I don't think they are getting any bonuses there. I wonder if it's possible that they are demanding techs from the other factions to keep peace...
Also I've seen tough monsters wandering quite a distance from my start. Last game I saw a dragon wandering around quite some distance from my territory (between two other factions). It wasn't destroying AI cities (was next to one for quite some time), but it never went after mine either. Also an undisturbed Lord of Elemental lair as well. Maybe it's not wise to make conclusions about monster distribution/behaviors from just one game.
Tuidjy in my book you deserve a medal for that inspection, most of thoseFOW issues, Monster Issue, AI Sight and AI settlingI already "kind of knew", but since I could not prove it my theories were turned away, so I thank you for putting up this elaborate testing, and I hope some of it is atleast considered.
I will say to Frogboy, Thanks! for trying to fix some of these issues halfway through, that said could you please post more detailed information next time, I was looking at your post and it gave precious little information to Tuidjy about what was changed, so he would have no idea about which of these items would be different. All I wonder is, do normal AI's have all the same flaws (since there were some rounding error, which I hope was checked for ALL the difficulties).
Sincerely~ Kongdej
I have a question about fog of war, and i can't think of a way to test it. it's like shroedinger's cat, it needs dev help to look at it.
Let's say the monsters have limited field of view based on who is around them. My understaanding is that it originally was based on what the player could see, so the ai had a certain stealth bug. This was fixed in the manner that the ai field of view was added to monster's vision so the monsters could see ai equally well as players.
If you use the fog of war cheat to look at everything, do the monsters gain the field of view to see eveything or do they still see what the players and ai present around the monsters see. How can you possibly test that?
Further comment about the fog of war: Is it possible that the monsters are only activating in the areas revealed? Thus, when you deactivated Fog of War, all monsters started moving??? Obviously you can't check this because you have to disable fog of war to check it...which causes them to move...heisenberg principle...
Thats hard, mostly because monsters don't always act as I think they should.
If you have unlimited time you could turn on cheats, then go and play a game for 100 turns and save each turn, load each turn up and view the whole map through the save, compare what happens on a 5-10 turn basis what the difference is between loading the turn by turn saves, and just ending the turn with the whole map revealed.
Just a stray thought.
That's why the 'Confidence' was only average. But the point is, I _was_ out-researching Altar, both by the graph and by adding up his cities' research. Our research values were roughly equal, with mine maybe 10% or so higher. Technology treaty bonuses are already taken into account in the graph and in the resource bar.
Still, when I would go to trade tech, he would have almost twice the tech than I would, despite the fact that I had already traded with the other players, and gotten theirs without giving out any of mine. And he really had almost, but not quite twice my research to trade. So it looks as if his tech is somehow doubled, which should not happen on challenging.
Note that I am not saying the VALUE of his tech that is doubled. Yes, that is doubled too, but its due to his influence and is a totally separate matter. The AMOUNT of tech he had to trade was the double of what I had.
As for the last point, I did not even include it in the post the first time, and only did it after I started a new game and revealed the map. It definitely SEEMS that that BBs are rarer around AIs locations. It may be random.
When I say that confidence is "Low" or "Average" that means that I consider something unlikely, or at best possible. The "high", "very high", "beyond doubt" are the ones I stand behind.
Just wanted to post and say thank you very much Tuidjy. Excellent work. I've been down on Stardock for releasing FE too early for a while now and a large part of this is because there are so many small to medium sized bugs, including those you have identified. Each one may not matter on its own but together they led me to not trust the game, even in the last few betas.
I really wish Stardock had fixed more of these first, any piece of software has a 'feel' about it based on how polished and reliable it seems to play and FE's feel just isn't as nice as it could be, which will be having a negative affect on review scores.
Oh well, it could be in a worse condition. I think it is pretty decent, and now that it is released it can't be unreleased, so let's just make the best of it we can!
Thanks Tuidjy for helping to improve the game with your very careful analysis!
I'd love to know how many of Tuidjy's findings are accurate versus bugs versus inaccurate. Quite a few of those have a pretty big implication on game play.
The part that becomes frustrating is when the AI does behaves like it's playing by a completely different set of rules. I'm sure it has to a certain degree to be competitive, but it's hard to know as a player how to anticipate how my enemies will act if they can do things I can't (like rush things for free or see things I can't).
I'm on the same page as well. I think it's still a fun game but all the issues and crashes and UI annoyances have put me off the game for awhile. I was quite surprised when some label this game "polished", when in my opinion things like the oddly different choice of scrolling mechanics for the tactical battle and strategic map screens just highlights the lack of it in fact.
That being said, I'm also confident Stardock will properly address these issues whether in a patch or expansion pack sometime in the future.
Tuidjy---pretty amazing
There are many great features available to you once you register, including:
Sign in or Create Account
