Who is planning to remove 1UPT with the new Mod?
- Thread starter BjoernLars
- Start date
rschissler
King
While I basically agree with this, what if the AI doesn't get fixed? Should Firaxis scrap 1UPT, because they can't fix it? Is there anything in the February patch notes about making the AI work better with 1UPT? They've been working on the Civ V and 1UPT development for two years and didn't come up with anything, so what makes you think they can fix it now?
It's easy to say "I hate stacks of doom", but the AI could handle stacks much better than it can handle 1UPT, so how can 1UPT be considered such an improvement?
What are the core faults with the 1UPT AI? I have read about the AI's inability to protect it's weaker ranged units. What other major faults are there?
To me Civ 5 combat seems to be geared more for meeting your enemy in the field and less about City conquest. City conquest is the ultimate goal, but defeating eneny armies should open the door for easier city conquest. Cities in Civ 5 are a bit too powerful and the 2 hex attack range makes it very difficult for melee units to attack unless there are a large number of them attacking at once.
In looking ahead, I think the game could move in one or two directions. The first would involve addressing specific 1UPT AI deficiencies such as doing a better job of protecting weaker ranged units and siege units. The AI could also be taught to use air and sea power more effectively.
The other direction the game could go would be to allow for some sort of reduced stacking. Units could be allowed to stack but would attack and defend separately. Perhaps a total of 2 or 3 units could reside in a given hex. To clear the hex, one would have to defeat all the units separately. One could also remove the ability for a city to attack directly and allow for more than 1 unit to be fortified in a city. The amount could be equal to the population size of the city up to a max number.
I think that given a proper amount of time that Firaxis will be able to shore up the Civ 5 combat AI to the point where it will offer up some form of challenge again.
-Hrnac
To me Civ 5 combat seems to be geared more for meeting your enemy in the field and less about City conquest. City conquest is the ultimate goal, but defeating eneny armies should open the door for easier city conquest. Cities in Civ 5 are a bit too powerful and the 2 hex attack range makes it very difficult for melee units to attack unless there are a large number of them attacking at once.
In looking ahead, I think the game could move in one or two directions. The first would involve addressing specific 1UPT AI deficiencies such as doing a better job of protecting weaker ranged units and siege units. The AI could also be taught to use air and sea power more effectively.
The other direction the game could go would be to allow for some sort of reduced stacking. Units could be allowed to stack but would attack and defend separately. Perhaps a total of 2 or 3 units could reside in a given hex. To clear the hex, one would have to defeat all the units separately. One could also remove the ability for a city to attack directly and allow for more than 1 unit to be fortified in a city. The amount could be equal to the population size of the city up to a max number.
I think that given a proper amount of time that Firaxis will be able to shore up the Civ 5 combat AI to the point where it will offer up some form of challenge again.
-Hrnac
I guess I'm one of the few people who liked the combat in Civ4. Constructing a good stack with the right mix of units and promotions was half the fun. Then judging the right moment when it was strong enough to take on a heavily defended enemy city (or you could cunningly launch a surprise attack on a weakly-defended one), how many casualties you'd be likely to take and whether you'd be able to regenerate in time to take the rest of your enemy's cities before a counter-attack. The feeling of dread when Catherine declared war and turned up with a huge stack of Cossacks... could you get enough defenders into your city in time? Your stack was a character in itself, and the decisions were more strategic and interesting than the minutiae of manoevering every last unit around a grid.
True, in the late game it could become a bit much and rather tedious when you had so many units and cities, but early / middle game was fantastic.
True, in the late game it could become a bit much and rather tedious when you had so many units and cities, but early / middle game was fantastic.
Not only the AI is not able to properly protect its units, it hasn't any proper understanding of areas/regions/battle theatres (however you would like to call it) either.
In short, stacking meant vertical grouping. As many units as allowed by game rules on ONE tile (hex). That is something which the AI can grasp.
1upt means horizontal grouping. ONE unit per tile (hex) on as many tiles (hexes) as given per map. This is much harder (and needs much more computing power) as now even the deployment of units counts, not to mention the influence of enemy units somewhere running around.
That is exactly what was promised pre-release and what has never been delivered.
It would require the understanding or regional concepts (once again), giving such a region a certain "weight" and based on that determine whether to attack/defend that region or not.
And given the three elements:
* current processing power
* available memory (especially for a 32bit program)
* tolerable turn-change times
you won't see such a system in the near future.
It might be coded (theoretically) but it would require the best processors, more RAM than available for most players and finally would result in turn times somewhere in the range of minutes, if not even more.
Would, could, should, might ...
I refer to the above mentioned restrictions.
Very, very, very unlikely.
That is exactly the point.
Actually, due to higher number of units (for both, you and your opponents) there were much more strategic decisions to be made.
But even the tactical choices were much more, as you would be thinking about how to get that town with as few losses as possible (after all, you would have to place some garrison there and to move on with the rest, if possible).
They probably can't fix 1UPT AI, but the AI they'd come up with for stacking would be just as bad. That's because there are fundamental problems with how they believe the AI should behave, and these same problems would arise under either 1UPT or stacking.
The problem isn't the AI's ability or inability to arrange units across a 1UPT map. It's more deeply flawed than that.
It's not a terrible idea, but not really playing to the AI's strengths, either. Probably the opposite direction they should first head in when trying to improve the tactical and strategic AIs.
All of which was completely lost on the AI.
CIV AI worked on brute force. BTS just tailored the AI's brute force approach to be more challenging, which is where CiV is at now, except people have this expectation that 1UPT means the AI can't/shouldn't brute force; it should, because that's the AI's greatest strength (research/production/gold/maintenance bonuses).
Which is why I say Firaxis won't improve the AI, because they're not playing to its strengths. They tried to program it to play like a human might play.
I agree insofar as Firaxis has lost all reliability, as far as meaningful development is concerned.
I agree with the AI having been weak in Civ4.
I don't agree with the assumption that the AI could brute force under an 1upt environment ("carpet of doom"). Even less now, since cities have become stronger.
Completely the wrong move to strenghten the AI, if you're asking me.
Where in a stacking system the AI could make up for improper composition of the stacks by having bigger ones/more of them, there is only that much place available on any given map.
Which once again was an improper move.
Not only does it requires quite some skills (not to mention man-time and size of team) to produce an AI which really could emulate human behaviour, the attempt also means that they tried to get rid of the single player game.
They tried to make Civ5 a multiplayer game without having to have human opponents.
But in a game humans behave differently from what they would do in real life. In a game, it is fun to backstab your neighbour/friend, just to hear him shout in surprise, anger and despair. In real life very often you don't do so because of the consequences.
Therefore it is a valid approach to have the AI "roleplay" in single player mode. At the end of the day, this behaviour is much more "human-like" than trying to emulate your giggling friend - for the abovementioned reasons.
Jolly Rogerer
Prince
- Joined
- May 16, 2008
- Messages
- 402
The AI problems for 1upt are several.
1) Pathfinding is not simple even on a clear map. Cluttering the map with moving obstacles makes it orders of magnitude more difficult. Even simple deployments of forces in friendly terrain during peace time have to be done as a whole rather than piecemeal, lest the movement of one unit made in the wrong order block the planned movement of another.
2) While there are fewer units than in Civ4, there are more "stacks" to maneuver, and more options (ranged combat etc.), further complicating #1 above.
3) A lot of people expect that since chess programs have decent "AI" it should be fairly simple to use brute force to solve for optimal play. But everything is much more complicated in Civ. There are many more tiles with variable terrain types, all units can move every turn, new units can appear, you have limited map intelligence etc. In addition combat is significantly more complicated, simply moving into an enemy unit does not provide a single simple result every time, but a range of possible results. All this has to be factored into #1 above, ie where should my units go and in what order, where you can't know with certainty what state some moves will leave the map in until you make them. As more units are added the complexity of the solution goes up exponentially, and has to be re-figured periodically as events proceed. Hence your blazing machine ends up grinding to a halt on AI turns and still produces poor results.
1) Pathfinding is not simple even on a clear map. Cluttering the map with moving obstacles makes it orders of magnitude more difficult. Even simple deployments of forces in friendly terrain during peace time have to be done as a whole rather than piecemeal, lest the movement of one unit made in the wrong order block the planned movement of another.
2) While there are fewer units than in Civ4, there are more "stacks" to maneuver, and more options (ranged combat etc.), further complicating #1 above.
3) A lot of people expect that since chess programs have decent "AI" it should be fairly simple to use brute force to solve for optimal play. But everything is much more complicated in Civ. There are many more tiles with variable terrain types, all units can move every turn, new units can appear, you have limited map intelligence etc. In addition combat is significantly more complicated, simply moving into an enemy unit does not provide a single simple result every time, but a range of possible results. All this has to be factored into #1 above, ie where should my units go and in what order, where you can't know with certainty what state some moves will leave the map in until you make them. As more units are added the complexity of the solution goes up exponentially, and has to be re-figured periodically as events proceed. Hence your blazing machine ends up grinding to a halt on AI turns and still produces poor results.
ohioastronomy
King
- Joined
- Dec 14, 2005
- Messages
- 714
Movement. It's just an incredibly hard traveling salesman problem to get armies from A to B; basically, the map is the wrong scale relative (too few hexes, too few movement points) to what they're trying to do with it. This is a design failure, not an implementation failure.
Txurce
Deity
There's not much point in arguing the pros and cons of 1upt at this point, except as an exercise. Everyone has played with it enough to arrive at a conclusion that's not likely to change. And I'm sure 2K has arrived at its own conclusion, given the results of every single poll on the subject: 1 upt is here to stay. To think otherwise is to think wishfully.
rschissler
King
That's true. Even if 2K was debating whether to change it, they probably wouldn't in the end. It would almost be humiliating for them to do so. Maybe in Civ 6, if there is one, they will be smart and come up with a variation of it, which could appeal to both camps.
Considering the poll results show that 80% of the responses on this forum wouldn't remove 1UPT and replace it with SoD, the camp that wants it gone isn't very large. Just vocal. As usual, those who like the game are out playing it, while those who don't are on the forums posting about it. So the ones who don't want SoD back stop by, hit 'NO', and leave, and the other 20% (poll at this time) fill the thread with negative posts for the most part.
(At time of posting, poll was 200 'NO' to 50 all other responses)
need my speed
Rex Omnium Imperarium
Perhaps, but most people who would have voted yes have, by now, given up on Civilization V, I think. That's why I haven't voted to, I don't plan to play the game anymore, so...
rschissler
King
I'd say the poll doesn't provide the true scope of the issue. I would like to choose 1UPT, but it really doesn't work with the scale of the game, so my only other choice is Stacks of Doom? That's no solution either. In any event, a change like this should come trom 2K/Firaxis, not a mod, which is what this poll is really about anyway.
P.S. I didn't vote for anything
Polls sucks, customers don't know what they want, they know what they have. Show them a solid stacks solution in Civ6 and they quickly change their mind.
And really, most of us who wants stacking back doesn't want Civ4-combat. We have different solutions. And even a lot of people who say they like 1UPT asks for armies, superunits and so on which can be seen as just another form of stacks. Polls are simply not that interesting and often simply misleading. Real discussions on combat models are interesting.
When it comes to this poll it's about modding and not 1upt vs stacking. Of course most people are not going to mod the game at all.
Of those capable most are probably realizing that simply removing the 1upt restriction doesn't make much sense by itself. What you need is an alternative combat model and this certainly requires an almost complete rewrite of the AI, and extensive modifications of units and maybe tile yields. UI also needs tweaks.
If you ask in a poll "Do you think the AI is good?" You'll get a lot of negative answers. Few will say it's good, almost nobody will say it's great. But if you ask "Who is going to write a better AI?" almost nobody is going to say yes.
I havnt played CIV V for months cause 1UPT is the worst idea for a civ game possible (i cant even understand how one can like it; there ABSOLUTELY nothing good about it)
But hey easy wins over a monkey brain AI seem to suit a lots of people who like to make themself feel smart!
Thanks to this mod... there's now some hope to save this game... and this franchise
But hey easy wins over a monkey brain AI seem to suit a lots of people who like to make themself feel smart!
Thanks to this mod... there's now some hope to save this game... and this franchise
- Joined
- Mar 31, 2008
- Messages
- 14,926
Part of the 20%.
sketch162000
Warlord
- Joined
- Oct 12, 2010
- Messages
- 221
You're right, diplo/space/culture victories are too easy. The whole game is on a level where a monkey could play and win.
That's diplomacy; I'm talking about the tactical and strategic AIs.
I used the term might because the AI seems to be modeled on a poor human approach. Or perhaps it would be better described as a more specialized human approach. Either way, it's not an optimal approach given the AI's bonuses.
Carpets of doom are the result of an inefficient or otherwise negligible unit limiting system. The player shouldn't be able to create them, much like the player shouldn't be able to pure ICS. The AI, aided by its bonuses, could reach that threshold, but shouldn't because it has been programmed to go to war beforehand. That's how you implement tactical combat in a strategic map. You can't simply slap the unit maintenance mechanic from CIV onto 1UPT and expect it to work.
You do raise a good point on the issue of cities having been buffed, but that's irrelevant to the larger issue of can the AI brute force.
I really suggest you take the time to read through the tactical moves XML if you're going to comment on 1UPT. I've stated repeatedly how amazingly poor they set some of those thresholds, and how those settings reflect a misguided philosophy which presumably extends to more core aspects of the strategic and tactical AIs, contained in files that go beyond my limited programming knowledge, probably including core files that cannot be accessed at this time.
I'm now going to move past asking people to take my word for it and actively request they put the effort in before making claims about how 1UPT was implemented.
And I'll once again ask for some kind soul to supply me with the tactical moves XML, because I no longer have CiV installed and have no intention of ever installing it again. One block of quoted text from that can demonstrate just how shortsighted the AI development process was.
Similar threads
- Locked
- Poll
