I need a challenge!
Moderator: Oberlus
Forum rules
Always mention the exact version of FreeOrion you are testing.
When reporting an issue regarding the AI, if possible provide the relevant AI log file and a save game file that demonstrates the issue.
Always mention the exact version of FreeOrion you are testing.
When reporting an issue regarding the AI, if possible provide the relevant AI log file and a save game file that demonstrates the issue.
-
- Space Floater
- Posts: 37
- Joined: Thu Nov 19, 2015 10:39 am
I need a challenge!
After playing version 0.45 for a while I am a little bit disappointed by the overall difficulty level. I win virtually every game I start. (except egassem without natives/close AI) Manical AI, Low specials, Starlanes set to high, monsters set to low. (i could play with monsters off to make it easier for the AI, but that seems boring to me). I tried small universes (100systems) and big universes (300 systems). That does not seem to make a difference. According to the research/production/military-meters I dominate the game around turn 100. Thats when it gets boring, especially big universes.
I came up with possible solutions I want to share:
- Challenge mode. One button, like "Quick start": Creates a game with a new seed, random galaxy shape, size (100-300Systems? or [enemies+1]*25), all other settings random, race random, AI maniacal. The meters ingame show only values that are in your scan range (or are disabled completely). That would make the player react to the infomation of his scouts MUCH more. Example: If I know, the natives setting is high, i build troop ships turn 1 right away. If its random, that might be a mistake. That is more tactical planning and more interaction. The randomness of the settings could be tuned so that settings that make a strong AI are more probable.
- Native help for the AI: On the highest difficulty setting (or in challenge mode?) there is always one (unguarded, no planetary defense, just troops) native planet 1-2 starlanes away from each AI homeworld, compensating for their weaknesses:
Egassem: Scilor/happybirthday/fifty-seven/gris gruf
Humans: Hhhoh, or other good pilots
Gysache: Hhhoh, or other good pilots, Ugmors/abadonny/beige goo
Chato: Ugmors/abadonny/beige goo
george: scilor/happybirthday/fifty-seven
Cray: Ugmors/abadonny/beige goo
Etty: Ugmors/abadonny/beige goo/thaegirus
Lafena: Deatheran, gris gruf
Trith: Thaegirus, silexian, Hhhoh
Eaxaw: Ugmors/abadonni/beige goo
I hope I remember all the racial traits correctly. My rule of thumb was to give the species that are bad at science/instrustry the trait that they are bad at and if the race is balanced like humans, just give them pilots. Also, the natives should be on the other side of the wheel of habiatability (if able to build colonies). Of course there should be some randomness involved, the combination must not be fixed. That probably needs some testing if it is too strong. In this case one could mix "cannot colonize planets"-species into the native help.
I am absolutely clueless about the xenophobic races. Honestly I think that being xenophobic is to big of a downside for a race. ESPECIALLY if the race has a narrow envoiremental tolerance. A rough estimation in my games tells me that the xenophobia costs me about 20-30% of the total industry (including all bonuses). At least in the early stages of the game, wich is much more important. That is just to much. But to be sucsessfull in the galaxy you need other species .. Did you ever think about cutting the xenophobia distance to maybe 2 Starlanes? Maybe via a tech?
How does the AI handle xenophobia? Roleplaying mode: Kill all other species?
I came up with possible solutions I want to share:
- Challenge mode. One button, like "Quick start": Creates a game with a new seed, random galaxy shape, size (100-300Systems? or [enemies+1]*25), all other settings random, race random, AI maniacal. The meters ingame show only values that are in your scan range (or are disabled completely). That would make the player react to the infomation of his scouts MUCH more. Example: If I know, the natives setting is high, i build troop ships turn 1 right away. If its random, that might be a mistake. That is more tactical planning and more interaction. The randomness of the settings could be tuned so that settings that make a strong AI are more probable.
- Native help for the AI: On the highest difficulty setting (or in challenge mode?) there is always one (unguarded, no planetary defense, just troops) native planet 1-2 starlanes away from each AI homeworld, compensating for their weaknesses:
Egassem: Scilor/happybirthday/fifty-seven/gris gruf
Humans: Hhhoh, or other good pilots
Gysache: Hhhoh, or other good pilots, Ugmors/abadonny/beige goo
Chato: Ugmors/abadonny/beige goo
george: scilor/happybirthday/fifty-seven
Cray: Ugmors/abadonny/beige goo
Etty: Ugmors/abadonny/beige goo/thaegirus
Lafena: Deatheran, gris gruf
Trith: Thaegirus, silexian, Hhhoh
Eaxaw: Ugmors/abadonni/beige goo
I hope I remember all the racial traits correctly. My rule of thumb was to give the species that are bad at science/instrustry the trait that they are bad at and if the race is balanced like humans, just give them pilots. Also, the natives should be on the other side of the wheel of habiatability (if able to build colonies). Of course there should be some randomness involved, the combination must not be fixed. That probably needs some testing if it is too strong. In this case one could mix "cannot colonize planets"-species into the native help.
I am absolutely clueless about the xenophobic races. Honestly I think that being xenophobic is to big of a downside for a race. ESPECIALLY if the race has a narrow envoiremental tolerance. A rough estimation in my games tells me that the xenophobia costs me about 20-30% of the total industry (including all bonuses). At least in the early stages of the game, wich is much more important. That is just to much. But to be sucsessfull in the galaxy you need other species .. Did you ever think about cutting the xenophobia distance to maybe 2 Starlanes? Maybe via a tech?
How does the AI handle xenophobia? Roleplaying mode: Kill all other species?
Re: I need a challenge!
The AI handles Xenophobia very very badly.
And having played it through on the "kill them all" approach, I think they're still slightly overpowered (especially in the current Test builds where someone *cough*me*cough* has messed up the Self Sustaining maths and not had time to fix it yet), I do want to tweak the numbers, but it's definitely going to be 5 starlane jumps. Note, if you're playing on High Starlanes that's a far worse affect than if you're on Low.
If you want a different type of challenge, play Low Planets and Low Starlanes, set monsters to Low as well as the AI can get very stuck with this but it's a really different environment.
But yeah, the mid-to-late game challenge is currently not as good once you know how to play well, and it's something we need to work on, some of it is pure balance, research in particular is badly costed and we need to balance the costs of the techs you need against what you can get in RP far better than we currently do.
And having played it through on the "kill them all" approach, I think they're still slightly overpowered (especially in the current Test builds where someone *cough*me*cough* has messed up the Self Sustaining maths and not had time to fix it yet), I do want to tweak the numbers, but it's definitely going to be 5 starlane jumps. Note, if you're playing on High Starlanes that's a far worse affect than if you're on Low.
If you want a different type of challenge, play Low Planets and Low Starlanes, set monsters to Low as well as the AI can get very stuck with this but it's a really different environment.
But yeah, the mid-to-late game challenge is currently not as good once you know how to play well, and it's something we need to work on, some of it is pure balance, research in particular is badly costed and we need to balance the costs of the techs you need against what you can get in RP far better than we currently do.
Mat Bowles
Any code or patches in anything posted here is released under the CC and GPL licences in use for the FO project.
Any code or patches in anything posted here is released under the CC and GPL licences in use for the FO project.
-
- Space Floater
- Posts: 37
- Joined: Thu Nov 19, 2015 10:39 am
Re: I need a challenge!
Just so I get you correctly: You think, the xenophobic races are to STRONG??? Maybe its just because I don't like concentration camps .. thats what you get when going to school in germany.
Do you really think you can tune the numbers of the research-paths or anything else in a way, so that the AI is a challenge for a good human player??? Don't you think that no matter how good it is, the AI will always need an edge over the player to be a real challenge? I don't think that is a question of finetuning .. the AI will always be worse that a human player. So at some point in my opinion you will have to make the AI cheat, or give them some bonus.
What do you think of my suggestions?
Edit: ok, I just tried the trith and this self-sustaining thing is pretty good. Also, it seems I overestimated the xenophobic frenzy. Its -50% production/research, right? And like any other mali of such kind, boni are applied afterwards. Thats half as bad. Ok, so i would just ignore the frenzy, play it like any other species and take the malus as a little downside, thats compensated by the self-sustaining .. Not very game changing .. but thats mainly because I don't like concentration camps ..
Anyway, also that game was way to easy. Back to my original question ..
Do you really think you can tune the numbers of the research-paths or anything else in a way, so that the AI is a challenge for a good human player??? Don't you think that no matter how good it is, the AI will always need an edge over the player to be a real challenge? I don't think that is a question of finetuning .. the AI will always be worse that a human player. So at some point in my opinion you will have to make the AI cheat, or give them some bonus.
What do you think of my suggestions?
Edit: ok, I just tried the trith and this self-sustaining thing is pretty good. Also, it seems I overestimated the xenophobic frenzy. Its -50% production/research, right? And like any other mali of such kind, boni are applied afterwards. Thats half as bad. Ok, so i would just ignore the frenzy, play it like any other species and take the malus as a little downside, thats compensated by the self-sustaining .. Not very game changing .. but thats mainly because I don't like concentration camps ..
Anyway, also that game was way to easy. Back to my original question ..
-
- Space Floater
- Posts: 37
- Joined: Thu Nov 19, 2015 10:39 am
Re: I need a challenge!
Or maybe a help for the AI that is not soo strong and easier to code:
The negative racial traits do not apply to the AI on the hardest difficulty.
Egassem have more population an normal research, Trith are neither xenophobic nor intolerant, Chato do not have bad ground troops, and so on (Humans get a 25% population bonus). These negative traits apply as normal when the AI-colony is captured by the player.
If i knew how to do that i would code it myself and make my own testversion.
Talking about bad/good ground troops: Why does that trait not apply to offensive forces? The impact of -50% defensvie troops can be compensated by a technology that costs 10RP (calculated for colonies in the early game). That does not seem very balanced.
The negative racial traits do not apply to the AI on the hardest difficulty.
Egassem have more population an normal research, Trith are neither xenophobic nor intolerant, Chato do not have bad ground troops, and so on (Humans get a 25% population bonus). These negative traits apply as normal when the AI-colony is captured by the player.
If i knew how to do that i would code it myself and make my own testversion.
Talking about bad/good ground troops: Why does that trait not apply to offensive forces? The impact of -50% defensvie troops can be compensated by a technology that costs 10RP (calculated for colonies in the early game). That does not seem very balanced.
Re: I need a challenge!
I'm British, we invented Concentration Camps. We maybe ought to consider renaming them as forced labour camps or something to make them less historically problematic.Toastmartin wrote:Just so I get you correctly: You think, the xenophobic races are to STRONG??? Maybe its just because I don't like concentration camps .. thats what you get when going to school in germany.
But yeah, if you're not using th ecamps they're going to be underpowered, because one of their main strengths is the benefits the camps give them as a pre-unlocked starting tech not a mid game tech. Wipe out everything that isn't your starting species and you can blitz the galaxy easily.
I don't know how good we can get the AI, what I do know is that if you get ahead by the late mid game you race ahead after that as research gets really easy.Do you really think you can tune the numbers of the research-paths or anything else in a way, so that the AI is a challenge for a good human player??? Don't you think that no matter how good it is, the AI will always need an edge over the player to be a real challenge? I don't think that is a question of finetuning .. the AI will always be worse that a human player. So at some point in my opinion you will have to make the AI cheat, or give them some bonus.
I'm recovering from flu thus anything that requires major analysis is marked in the 'do this when you have a brain' pile. I'll get back to it.What do you think of my suggestions?
The Trith absolutely need to be wiping out other species as much as possible, their population malus can get quite nasty. You could use bombardment tech instead of camps, and part of me wants to give that to one of the xenophobe races instead of camps, but until that's done, camps are scary-good, way overpowered on their own.Edit: ok, I just tried the trith and this self-sustaining thing is pretty good. Also, it seems I overestimated the xenophobic frenzy. Its -50% production/research, right? And like any other mali of such kind, boni are applied afterwards. Thats half as bad. Ok, so i would just ignore the frenzy, play it like any other species and take the malus as a little downside, thats compensated by the self-sustaining .. Not very game changing .. but thats mainly because I don't like concentration camps ..
But yeah, the malus is approx 50% base production before other bonuses, at some point I want to get the numbers redone a bit to make it easier to explain and comprehend but it's really not that bad.
Geoff has started some work implementing some of your idea (Setup random), but it's going to need some work.Anyway, also that game was way to easy. Back to my original question ..
It does in current trunk, the trait got split up and we're testing the balance of it, I need to change some of the numbers and move some species into different settings but it is there now.Talking about bad/good ground troops: Why does that trait not apply to offensive forces? The impact of -50% defensvie troops can be compensated by a technology that costs 10RP (calculated for colonies in the early game). That does not seem very balanced.
Seriously, download the most recent test build once you've finished your next game, it's playable and you're at the stage now where feedback on the new features would be very much welcome.
Mat Bowles
Any code or patches in anything posted here is released under the CC and GPL licences in use for the FO project.
Any code or patches in anything posted here is released under the CC and GPL licences in use for the FO project.
- Geoff the Medio
- Programming, Design, Admin
- Posts: 13587
- Joined: Wed Oct 08, 2003 1:33 am
- Location: Munich
Re: I need a challenge!
I can't find the thread, but it was suggested somewhere that one could do nothing for the first ~50 turns of a game, then start playing, in order to increase the challenge.
- Cpeosphoros
- Space Kraken
- Posts: 124
- Joined: Sat Jan 30, 2016 11:29 am
Re: I need a challenge!
That would be a nice thing to include in the setup screen. Something like "turns handicap"Geoff the Medio wrote:I can't find the thread, but it was suggested somewhere that one could do nothing for the first ~50 turns of a game, then start playing, in order to increase the challenge.
All contributions are released under GPL or LGPL v2 or later, or under appropriate Creative Commons licence, consistent with project guidelines.
-
- Space Floater
- Posts: 37
- Joined: Thu Nov 19, 2015 10:39 am
Re: I need a challenge!
Is that turn handicap implementable without having to wait for several minutes until the computer has finished the computation of the 50 turns? Depends on the CPU, but how long would that take?
I still like my idea of fitting natives in the neighborhood better because it would be implemented behind the scenes. It does not look that artificial, it could just be coincidence, that the AI got a good native planet early . But I can see how the turn handicap is easier to code and and easy to understand for the player.
I still like my idea of fitting natives in the neighborhood better because it would be implemented behind the scenes. It does not look that artificial, it could just be coincidence, that the AI got a good native planet early . But I can see how the turn handicap is easier to code and and easy to understand for the player.
- Kassiopeija
- Dyson Forest
- Posts: 212
- Joined: Wed Jul 16, 2014 6:14 pm
- Location: Black Forest
Re: I need a challenge!
Giving the AI a headstart is a good & easy thing to do, although it doesn't really solve the issue. You just have to overcome this, and afterwards, can also overrun all AI without any of them being able to follow you.
What would be nice an option that gives the AI a set amount of extra turns, repeatedly. Such as, every 10 turns AI can make an add. turn, or, every 5 turns... with that number specified during setup by the player.
The problem with this is that it might be frustrating for some players when AI ships can move 2 times in a row.
Another way would be to use this system to give other stuff - not turns but instead, techs or products. So, every N amount of turns the AI will finish a project from the techqueue / prodqueue without paying for it. That will then advance their games, and more elegantly than with additional turns.
And just because a player can specifiy the frequency of this happening, he can finetune the game to a required strength.
Personally I like to play against a lot of monsters, and also, have very minimalistic setups via starlane setting etc so that it may happen I find myself in a bad starting position (which is good). But the AI can't handle this. So how about making the Monsters ignore the AI (not initiating battles, blocking supply, attacking planets). Monsters would essentially become a gainst-the-player feature.
And give the AI simply some extra range so they don't happen to be bottlenecked.
What would be nice an option that gives the AI a set amount of extra turns, repeatedly. Such as, every 10 turns AI can make an add. turn, or, every 5 turns... with that number specified during setup by the player.
The problem with this is that it might be frustrating for some players when AI ships can move 2 times in a row.
Another way would be to use this system to give other stuff - not turns but instead, techs or products. So, every N amount of turns the AI will finish a project from the techqueue / prodqueue without paying for it. That will then advance their games, and more elegantly than with additional turns.
And just because a player can specifiy the frequency of this happening, he can finetune the game to a required strength.
Personally I like to play against a lot of monsters, and also, have very minimalistic setups via starlane setting etc so that it may happen I find myself in a bad starting position (which is good). But the AI can't handle this. So how about making the Monsters ignore the AI (not initiating battles, blocking supply, attacking planets). Monsters would essentially become a gainst-the-player feature.
And give the AI simply some extra range so they don't happen to be bottlenecked.
Re: I need a challenge!
Maybe rather then a specific item make it a bonus, ie: +50 research that turn. Maybe an Advanced AI option that allows various bonus's to be given to the AI, but it should be entirely optional.Kassiopeija wrote:Another way would be to use this system to give other stuff - not turns but instead, techs or products. So, every N amount of turns the AI will finish a project from the techqueue / prodqueue without paying for it. That will then advance their games, and more elegantly than with additional turns.
And just because a player can specifiy the frequency of this happening, he can finetune the game to a required strength.
-
- Programmer
- Posts: 389
- Joined: Sun Feb 14, 2016 12:08 am
Re: I need a challenge!
Throwing flat numbers or extra turns sounds like it would be difficult to keep up with.
Not just for the player but even more for those doing the balancing.
If a handicap is applied, a percentage of the human or top players output could be easier, even as a stop gap.
'AI value' += 'Human value' * 'AI Handicap'
Or decrease research cost for techs based on diplo status.
Unknown = 0, Enemy = -5%, Ally = -15%
Not just for the player but even more for those doing the balancing.
If a handicap is applied, a percentage of the human or top players output could be easier, even as a stop gap.
'AI value' += 'Human value' * 'AI Handicap'
Or decrease research cost for techs based on diplo status.
Unknown = 0, Enemy = -5%, Ally = -15%
Any content posted should be considered licensed GNU GPL 2.0 and/or CC-BY-SA 3.0 as appropriate.
Re: I need a challenge!
Right now the game is still evolving, and is not ideally balanced.
Maybe chose the path less taken, for the added difficulty?
(And provide feedback on the experience, for when decisions are made to better balance the game.)
For example, I think the robotic ship hulls (robotic / self-gravitating / titanic) are so much better than the other choices, at the moment. As a challenge, I am currently trying to only use organic ships, despite the turn delays and increased RP costs.
Maybe chose the path less taken, for the added difficulty?
(And provide feedback on the experience, for when decisions are made to better balance the game.)
For example, I think the robotic ship hulls (robotic / self-gravitating / titanic) are so much better than the other choices, at the moment. As a challenge, I am currently trying to only use organic ships, despite the turn delays and increased RP costs.
Re: I need a challenge!
Really? I think they're marginally better and marginally easier to use, didn't think the difference was that pronounced.mem359 wrote: For example, I think the robotic ship hulls (robotic / self-gravitating / titanic) are so much better than the other choices, at the moment.
This migt be because when I started playing heavily they were ridiculously underpowered and organics then asteroids were OTT so I learnt to use them well before switching and getting the balance right.
If people generally think the Robotics are that much better overall then I might need to go in and tweak some of the research and build costs a bit more.
(worth observing that stealth doesn't really work brilliantly as a strategy currently and that's a strength of organics so they are a bit more of a challenge but not much)
Mat Bowles
Any code or patches in anything posted here is released under the CC and GPL licences in use for the FO project.
Any code or patches in anything posted here is released under the CC and GPL licences in use for the FO project.
-
- Juggernaut
- Posts: 854
- Joined: Wed Aug 26, 2015 6:15 pm
Re: I need a challenge!
Robotic hulls are terrible and weak and no one uses them!!!MatGB wrote:If people generally think the Robotics are that much better overall then I might need to go in and tweak some of the research and build costs a bit more.mem359 wrote: For example, I think the robotic ship hulls (robotic / self-gravitating / titanic) are so much better than the other choices, at the moment.
-
- Space Floater
- Posts: 37
- Joined: Thu Nov 19, 2015 10:39 am
Re: I need a challenge!
tried the latest test version (2016-2-16, dont have the version number at the moment). Only had one game so far. The AI seems harder than before. The automatic weapon upgrades while in supply range definitely help the AI very much because they are not coordinating their research and military production. So the research helps them even if the tecnology was completed one turn after the ship production begun.
I still won, but it was not as easy as before .. The Egassem to my left had an early scylior colony .. the trith in the center had an early muu ursh .. did you implement my idea of giving the AI some native help? Nah, probably just coincidence .. (see, thats the beaty of the solution.)
Is it a new feature that xenophobic frenzy limits the polulation of a planet? Like it. I figured out, that concentration camps are only a good idea, if you are xenophobic. A constant source of production/research with the option to settle more planets almost always more valuable than the short industry spike in the log run. Even a 'bad' species like furthest is more valuable than anybody else on their home planet because of the home planet population bonus. So giving the xenophobics even more incencitive to do their thing is good. That means a substentially different playstyle. Is that so hard to code for the AI? Just kill every other species (autobuild concentration camps after conquest) and use the prodution for more Trith/eaxaw/exobot colonies?!?
I don't like the seperation of offensive and defenseive ground troops. Unnessesary complicated, not motivated by lore/species background, hard to understand/recognize for beginners. Why not just apply the ground troop bonus/malus to both offensive and defensive forces and tune the numbers accordingliy. (Muu Ursh would be too strong with ++pilots and ++troops) If you want a native planet to be hard to conquer (the only viable motivation for defensive troop stat seperation) just give that planet the native defense-special.
I like the starsystem-cloud that hinders the supply range of the system. Makes +supply more meaningful.
I still won, but it was not as easy as before .. The Egassem to my left had an early scylior colony .. the trith in the center had an early muu ursh .. did you implement my idea of giving the AI some native help? Nah, probably just coincidence .. (see, thats the beaty of the solution.)
Is it a new feature that xenophobic frenzy limits the polulation of a planet? Like it. I figured out, that concentration camps are only a good idea, if you are xenophobic. A constant source of production/research with the option to settle more planets almost always more valuable than the short industry spike in the log run. Even a 'bad' species like furthest is more valuable than anybody else on their home planet because of the home planet population bonus. So giving the xenophobics even more incencitive to do their thing is good. That means a substentially different playstyle. Is that so hard to code for the AI? Just kill every other species (autobuild concentration camps after conquest) and use the prodution for more Trith/eaxaw/exobot colonies?!?
I don't like the seperation of offensive and defenseive ground troops. Unnessesary complicated, not motivated by lore/species background, hard to understand/recognize for beginners. Why not just apply the ground troop bonus/malus to both offensive and defensive forces and tune the numbers accordingliy. (Muu Ursh would be too strong with ++pilots and ++troops) If you want a native planet to be hard to conquer (the only viable motivation for defensive troop stat seperation) just give that planet the native defense-special.
I like the starsystem-cloud that hinders the supply range of the system. Makes +supply more meaningful.