[BERT] Goofy Play by AI Lena Ebner

[vorlon_mi]

Noticed this in my recent game as Slavic Federation.

First, really bad city placement. The access to the ocean is blocked by the ice!

Second, I'm not sure if this is a bug, or amazingly foolish play.
INTEGR seems to have completed all the key research for a Transcendence victory, but look at her Affinity score. How is it possible to complete all that research while only accumulating 7 affinity score? And if she really achieved 14 Purity points, why is she not launching a satellite and pursuing Promised Land?

 

[Tiberiu]

Yes, really bad city placement. Not a bug, just bad AI. AI is really bad in this game, even worse than the game before it which is civ 5. Civ be has more complexity and AI is unable to handle it. You would be tempted to think that something as simple as settling a city would have been easy to code. Just make sure surrounding tiles have certain values, that there is no ice blocking if there is a coastal city, etc.

The question about key research for Trancendence and 7 affinity score is explained by the fact that these key techs don't really give that much affinity. Assuming no Might virtue or Ansible, only Swarm Intelligence gives 20 affinity, the rest give 7. So it's 34 affinity points. The discovery of the techs can be explained by them trying to gain advantages that are unrelated to Transcendence victory.

A way to try to make sense of what's happening in Lena's mind is that she's not interested in winning by Affinity victories and has decided for some reason for Contact or Domination.

We could make hundreds of examples. In a game I am playing now an AI explorer just walked over a progenitor ruin one turn earlier than I could do it, but then it moved away next turn instead of excavating the ruin. It's certain that they had a module, it was very early turns and I saw the explorer almost from the start. There might be a conflict in that AI is also trying to explore searching for pods. But nothing can justify simply ignoring a Progenitor ruin. Or how AI likes to attack stronger units with their Air units even if they will die in the process.

 

[vorlon_mi]

I finished this game last night. Learned a few things that I will need to take forward.

First, I had a challenge to finish the game, as noted in this thread https://forums.civfanatics.com/threads/late-game-crash-turn-337.676528/ My victory was still ~18 turns away on turn 336, when the game crashed. Lately I have been not hurrying to victory, since the scoring system gives no bonus for finishing quickly. I wanted this game (as Kozlov) to score higher than my last game with him.

Recovery Plans: I remembered making a conscious decision to pursue some Supremacy techs on the way to finishing the 3 required techs for a Harmony victory. Clearly, to avoid the crash at turn 336, I would need to finish my Mind Flower sooner so that it could connect before turn 336. I tried going back to a save from turn ~280, but the tech decisions were still baked in at that point. I went all the way back to a save from turn 215 or so, where I could beeline to the required Harmony techs.

Accelerating City Improvements: Elsewhere in the strategy forum, I had read that the improvements need to be finished *before* the Mind Flower is completed. I can confirm that. I can also confirm that a Mind Stem that finishes *on the same turn* as the Mind Flower doesn't help. By playing to beeline the Harmony techs, I started the Mind Flower earlier and it finished building just before turn 300. It required 36 turns to connect, so I was sweating whether the crash would still occur.

AI Changes: In the previous attempt (that crashed on turn 336), the objectors had been PAC (my immediate neighbor to the north, on the same land mass) and INTEGR (on the other big land mass). This time, as I started the Mind Flower, the objectors were PAC and ARC. That was good news, in a way, because ARC was somewhat weaker at that point. Fielding was still pursuing a Contact victory, making significant progress on Decode Signal at that point. She would go on to start building the Beacon.
ARC had a fast start, wiping out Polystralia with her navy. ARC was located in the southern hemisphere, with a clear ocean path to Polystralia's cities. She overextended herself attacking NSA (neighbor to the west, accessible through a southern ocean passage. Where before PAU had been poaching a few INTEGR cities and keeping Lena in check, this time through INTEGR built up an army. Took over ARC's conquests in the former Polystralia (far to the east of INTEGR) and later swept down to take out NSA in the south.

AI Victories -- Human Saved by Stupidity: As I worked to build my Mind Flower as quickly as I could, two AI nearly beat me to the finish line.

PAC Supremacy: Daoming had been steadfastly building her affinity level and her army. Lots of skirmishes with Africa and INTEGR in the northern ocean passages, but also settling a bunch of cities. She gets to affinity 15, builds the gate, and ... puts it in the water. I made this mistake in a previous attempt with Supremacy using NSA; all land units are embarked when they go through, so their strength is reduced. When she declared on me (as my Mind Flower waited for connection), I took a while to get troops and planes into position to attack the gate. She pushed through ~300 points worth of troops before I could mount a serious threat. Lesson 1 in complacency by the human player.

INTEGR Harmony: As I noted in the first post, Lena had teched all the way to level 14 in Purity, while seeming to have a change of heart to pursue Harmony. She had amassed quite an empire with multiple cities and lots of land. She made it to level 15 level in Harmony and built her Mind Flower in a size 23 city that was NOT her capital. A bit too late; she finished her Mind Flower while mine still had 10 turns to go before connecting. According to the victory log, her MF would have needed the whole 40 turns to connect. Did not appear to have built any accelerating buildings in time.
I like looking at the "replays" in the game main menu, just to see how I did compared with the other AI. I was out-teching and out-producing PAC, but INTEGR crushed me in score, tech, productivity, even military might. Lesson 2 in complacency by the human player... I should have intervened to slow her progress. When the AI gets a good start, on another land mass, they have the potential to run away. Her empire was strong enough to support two victory conditions at nearly the same time -- like mine have been in previous games.

 

[legalizefreedom]

After the initial crash, did it crash in the same spot again upon loading the latest autosave? I don't remember the last time BERT crashed on me, but it was certainly isolated. I would venture to guess your computer ran out of memory or something along those lines. Playing for hours straight at a time possibly. I wouldn't bet it would happen again in the same place after a restart, but anything is possible I suppose.

I've generally settled on playing on Soyez difficulty to keep the game from being a beeline race.

 

[vorlon_mi]

After the initial crash, did it crash in the same spot again upon loading the latest autosave? I don't remember the last time BERT crashed on me, but it was certainly isolated. I would venture to guess your computer ran out of memory or something along those lines. Playing for hours straight at a time possibly. I wouldn't bet it would happen again in the same place after a restart, but anything is possible I suppose.

I've generally settled on playing on Soyez difficulty to keep the game from being a beeline race.

For this save, yes, it crashed on the same turn. All of them -- autosave, quicksave, and even a conventional save made just before clicking "Next Turn" all crashed with the same exception dialog box. I agree, that it doesn't happen often.

Soyuz is providing the right amount of competition for me, to make it fun. If I have an achievement to go for, I might consider stepping back to Gemini, just to make sure that I get the achievement.

 

[legalizefreedom]

For this save, yes, it crashed on the same turn. All of them -- autosave, quicksave, and even a conventional save made just before clicking "Next Turn" all crashed with the same exception dialog box. I agree, that it doesn't happen often.

Soyuz is providing the right amount of competition for me, to make it fun. If I have an achievement to go for, I might consider stepping back to Gemini, just to make sure that I get the achievement.

That has happened to me before, but maybe only once or twice. I figured it was some magic order of operations that "does not compute!" Seems like I was able to get out of it by going back a couple of turns and just doing different things than I did the first time.

Yeah, I'm all about taking in the experience. I don't like AIs too focused on the end because I don't want to be. I'm a pretty efficient player after all these years even when I'm not trying to be, so Soyuz is usually a balance between not rushing and not dragging on for me.

 

[Sclb]

Yes the most recent crash I had was around turn 290+ and after finishing most of the tech web. Crash as soon as I open the tech screen to select next tech and always the same crash if I reload the same auto save. Only way I know to fix is to reload several turns earlier before the crash and to select all the techs needed till the end of game in one go.

Yes the crash is rare for me since I rarely drag the game for so long but I needed some late wonders for steam achievements

 

[legalizefreedom]

The one thing that crashed games for me (before I got my latest computer upgrade) is launching a ton of satellites. I can be a bit of a settling and satellite whore. Having a bunch of satellites has always caused some graphic flashing glitches which get worse the more you have. I've had to kill some satellites towards the end of the game to finish it up.

That and the black tiles loading in every time you pan (I always play on the largest map) are the two things that annoy me about the game. Seems like there was a fix for black tile loading. It helped.

 

[vorlon_mi]

The one thing that crashed games for me (before I got my latest computer upgrade) is launching a ton of satellites. I can be a bit of a settling and satellite whore. Having a bunch of satellites has always caused some graphic flashing glitches which get worse the more you have. I've had to kill some satellites towards the end of the game to finish it up.

Please tell us more about having a bunch of satellites. I tend to launch relatively few, so I'm interested.

• I always launch the free ones that come in a resource pod / expedition. Those are especially sweet since they usually come before I have researched the tech to unlock them.
• I usually always put a Station Sentinel over my "pet" stations. But sometimes that's only one. In about half my games, the stations either withdraw because no one linked them to a trade route, or the AI clobber them rather than trade with them.
• In the late game (after turn 200), I put up 2 or 3 Spy Satellites so that I can see the invaders from my end-game-wonder objectors before they declare war. Allows me to get the air cover in place, and often declare on them first.

Should I be building more Solar Collectors? Do you build one for each city? I will sometimes build a Tacnet Hub to park near the city where the AI is invading. I rarely build Weather Controllers or Orbital Fabricators, since I think there are better things to build. Do you build those?

 

[Darsnan]

Please tell us more about having a bunch of satellites. I tend to launch relatively few, so I'm interested.

It'll be interesting to see who all responds to this question (because I am curious how others perceive and utilize the orbital domain) and how different people handle the orbital layer of the game: it is relatively unique versus other God Games (at least that I am aware of).

Personally I rarely throw up Spy Satellites. Instead I throw up Orbital Lasers: I find the coverage delta (6 versus 3 tile radius) isn't that significant in my games. I still get the surveillance capability I would with Spy Satellites, and I also can start softening up the AIs forces before we get up close and personal (either offensively or defensively). I do use Station Sentinels quite extensively, especially for the bonuses they give, and before I get Orbital Lasers I'll throw up some Tachnet Hubs, both for the healing effect, as well as to claim the orbital domain.

It is interesting how my thoughts on the AIs use of satellites has changed over time: originally I thought the AIs were just spamming orbitals willy nilly, however I'm pretty sure the intent on the AIs orbital spam is to claim the orbital layers around their territory and prevent a human from domineering the AIs via the Orbital domain (either via Spy Sats, Orbital Lasers, or Planet Carvers). It takes time for me to clear all of the orbitals from around AI cities, and by the time I do clear the AIs satellites from an area, the ground/ air/ sea operations are usually in cleanup mode, so why bother throwing up an orbital that will then be useless because the front has moved on from there? It is a little thing, but the little things definitely do count!

D

 

[Tiberiu]

It'll be interesting to see who all responds to this question (because I am curious how others perceive and utilize the orbital domain) and how different people handle the orbital layer of the game

I think the final state of the game makes it so that Orbital units are not really needed in order to obtain victories (of course, with the exception of the ones that are required in order to fulfill the tasks for the victories).

For example Solar Collectors are cool to find and launch and use during the early-mid game, but it is hardly important to rush their tech so that you can spam them.
Weather controllers and Orbital Fabricators are also quite nice, but again it is hardly necessary to use them, and they are also hard to get to, reseach wise, unless you luck out on "Science" pods.
Holomatrix? The same, they are cool but they aren't really needed and it feels as if there is always something better to do than discover the tech for them.
The miasma-related satellites are mostly irrelevant I think. At least I never saw them as relevant, maybe others have a different experience.

In my opinion the most relevant orbital unit is the Station Sentinel. It is cheap, can come relatively early in the game, and also it can have a huge impact with the right Station and Virtues/Traits.

edit : totally forgot about Tacnet Hubs which I use quite frequently and of course the Comm Relay which is sometimes needed.

 

[vorlon_mi]

What a great insight, @Darsnan . A tradeoff that has several aspects.

Orbital Lasers (adv): longer duration, able to engage (disadv): require more techs, a little more production to build

Spy Satellites (adv): available sooner, less production required to complete (disadv) only look, no action

I usually include Tactical Robotics in my mid-game (turns 120-150) tech plan, but may not acrtually build any Spy Satellites until closer to turn 200 because I don't have far frontiers to keep watching.

I do see more AI-produced satellites when I wage war in the end-game-wonder phase; I see fewer AI satellites if I'm fighting before turn 300.

 

[vorlon_mi]

Miasma-related satellites seem very niche / situational.

I've had some maps where a neighboring region -- about 15 hexes, where I want to put cities -- is just covered with miasma like a blanket. It's more efficient (measured in turns) to clear miasma with a "Miasma Replulsor" than with a squad of workers. Situational.

If I were role-playing a Harmony strategy, where I took all the bonuses for Immortals and Xeno Swarm for "attacking in miasma", I might consider deploying some "Miasma Condenser" satellites to prep the battle field. Most of the time, I take the other bonuses for those units since a lot of the miasma was cleared in the early game.

 

[Darsnan]

A tradeoff that has several aspects.

Which I believe is inline with the Sid Meier paradigm of "a game should be an interesting series of choices".

I do see more AI-produced satellites when I wage war in the end-game-wonder phase; I see fewer AI satellites if I'm fighting before turn 300.

That's kinda my experience as well, but the 300 number is variable so I do usually start throwing up some Tacnet Sats in areas of interest so as to claim the orbital domain before someone else does.

If I were role-playing a Harmony strategy, where I took all the bonuses for Immortals and Xeno Swarm for "attacking in miasma", I might consider deploying some "Miasma Condenser" satellites to prep the battle field.

I like that idea! I always play Supremacy so have never really delved into the other affinities and their nuances that deeply. Usually when I'm expanding and run up against a large miasma field I use that as a buffer that the AI then has to trudge through to get to me. Usually works, although I have been surprised a few times when the AIs come thru with healing units amongst their Carpet of Doom and their ready to fight.

D

 

[legalizefreedom]

My bread and butter are Weather Controllers and Orbital Fabricators for their ability to create two resources each. I use them where I have two low yield tiles within one satellite range. You have to leave the tiles unimproved for them to spawn, with the exception of roads. Within 30 turns both are likely filled in with resources. If you want to micromanage them, you can note when they've spawned both and kill them if you need room for another, but I usually just enjoy the passive bonuses until they die. You do have to specifically go to them on the tech web, but I usually use my Institute free tech for one of the branches and grab the techs when the time is right.

The other one that is a must is the transporter. It's mid to late game when you can get to them, but they really allow you to stake out all the sweet uncolonized areas as well as respond to any surprise attacks with extreme prejudice.

I usually find one or two Solar Collectors (using the Ruins and Pods mod, which comes highly recommended for fixing quests and distributing rewards properly) which are usually enough to get me over the early game energy shortage. I wait until I have a group of 7 tiles being worked as to not waste the benefit. Same with culture sats. If I find one, I'll use it, but almost never build them unless for a specific strategy.

Miasma sats are very situational and I almost never need to build them.

The weapon sats have a lot of potential, but as implemented are very limited. By the time you place them, maybe you get a couple of shots off and you've advanced and they are useless. If they were able to move, even one hex per turn, their usability would go up exponentially. As is, they are rarely worth building. It's a shame too, since they are very cool.

If there's an epic struggle happening or a surprise dow, I will occasionally launch a defense/healing sat, esp around water cities being attacked, but that is situational as well. As with the weapon sats, you'll want to kill them once they have served their purpose as to not pay the upkeep.

Station sats are decent, but how often are Stations within your borders...?

 

[vorlon_mi]

I always get an early quest (say, in the first 20-25 turns) to approve a station close by. I tend to expand around that station, settling near resources and wrapping the station in my orbital coverage. Repair Facilities, Launch Complexes, and even Observatories can expand my orbital coverage. So, I usually get one.

Getting another is very hit-and-miss. I can "inherit" a station that is near one of the AI's cities when I conquer those cities. Other games, the stations all leave if no one establishes a trade route so I'm left with only 2 or 3 on the whole planet.