|
AI Discussion
|
| DubbelDekker |
Posted on 09-10-2014 14:05
|
Small Tour Specialist
Posts: 2633
Joined: 20-04-2008
PCM$: 200.00
|
Throughout the season many race threads have been filled with AI frustration. The recent Berlin ProRace is a good example. I think we can agree that the PCM13 AI just doesn't do justice to the hours of work that organizers, reporters and managers put into this great game.
I think this is an important topic we need to discuss. But it is also something we must keep out of the race discussions. Hence this thread.
The main issue with the PCM13 AI seems to be that a lot of it is programmed around the actions of the player. If the player keeps his riders passive (like we do with the control team), the AI often thinks "fine; I'll just let the breakaway win". There are other problems, but this fact alone makes this AI unsuitable for the Man-Game.
I'm curious what your views are on this topic.
Let's keep it constructive and focus the discussion on which AI we want to use for next season. To this end, I'm wondering which PCM had the best Man-Game AI in your opinion.
Also, it'd be great if someone could test whether the "passive player" problem is still there in PCM14.
|
| |
|
|
| Selwink |
Posted on 09-10-2014 14:11
|
Grand Tour Champion
Posts: 8856
Joined: 17-05-2012
PCM$: 200.00
|
I believe this is purely PCM13. In EPIC we've got the exact same problem in the continental races. I think it has to do with the general programming of PCM13, which has made the breakaway way too strong and almost impossible to catch if everyone's in the same shape. I haven't heard anything of these issues with PCM14, so I hope that will be used next season. The AI has become very frustrating to me as well, considering my team is a real sprinter team and needs good results in these races.
|
| |
|
|
| roturn |
Posted on 09-10-2014 14:12
|
Team Manager
Posts: 22246
Joined: 24-11-2007
PCM$: 3900.00
|
First. I did some tests with PCM14 and while it looks better than PCM13, it would require far more tests when you see how many PCM13 races we have "tested" during this season. So I guess also PCM14 will have some AI weaknesses.
In classics, the control riders are abandoned before the start. So it`s not a passive control team but the bigger problem is that breakaways in especially on cobbled classics are far too strong as they are still going strong in the end.
For this season it probably doesn`t matter anyway as changing the game midseason is highly unlikely to ever happen. This would not only require huge DB work etc. as also would mix the results too much to be fair for all managers. |
| |
|
|
| TheManxMissile |
Posted on 09-10-2014 14:16
|
Tour de France Champion
Posts: 18187
Joined: 12-05-2012
PCM$: 0.00
|
The PCM13 AI discussion has been done to death in PCM13 Threads and numerous Man-Game threads going back to last season. The Breakaway AI sucks balls, especially in flat classics. Keeping to MG passive play there is nothing to be done about it.
PCM14 is a lot better in this aspect, but the MG stats matrix could mess it up, as PCM13 seems to struggle more than in standard play
|
| |
|
|
| Ollfardh |
Posted on 09-10-2014 14:22
|
World Champion
Posts: 14563
Joined: 08-08-2011
PCM$: 9100.00
|
As I can confirm from EPIC, the problem is the shorter classics in PCM13. The breakaway wins way too often by means of a passive control team. If you want, I can look up some races to find the breakaway succes percentage.
I follow Selwink's views on the problem. If there's no active chase from the player, the breakaway usualy wins it. From what I've seen, there's only a few teams controlling the race, and they can't catch up with the breakaway. It compensates in the longer races, as Stamina comes into play and the breakaway usualy doesn't stand a chance there.
Recently a few or my reporters have switched to PCM14. I'll let you now how that goes. The GT's already seem to give some wonky results, unfortunately 
Changed my sig, this was getting absurd.
|
| |
|
|
| Mhaley45 |
Posted on 09-10-2014 15:20
|
Protected Rider
Posts: 1036
Joined: 17-06-2009
PCM$: 200.00
|
I am not sure how I feel about the effect of the AI this season. I have had some wild success and some bizarre failures most likely both contributed to AI.
Before we condemn the AI we should probably undertake the in-depth analysis of a couple of race where the breakaway succeeded when many feel it should not.
I am wondering if the important (contenders for the stage/race) teams had the quality of support riders to up the pace to catch the breakaway? In one set of races I remember Dutch Food Valley Classic and Interlagos (both PCT) were in conflict and similar terrain, so the squads were often split with some quality reduction in both races.
It is easy to say the AI is wrong, bad, doesn't work, sucks whatever, when the results go against your team.
All that said, the AI, in my opinion, seems very suspect of allowing weaker riders who are in the breakaway a better opportunity of surviving and doing well.
As my example AMEXpc's Jan Dieteren has just a 72 Hill rating, a 77 Fighter rating and a 62 Sprint rating. Hardly the type of numbers who should win a race on the PCT. Yet he won out of a breakaway riding away at the finish.
Another of my young riders, American Robert Bush, got in the breakaway at the Avila Classic, peloton never caught them and he finished on the podium (3rd) ahead of a slew of better cobblers.
Maybe we should have a 3 or 4 man team, review a handful of questionable races and then post a report of their findings.
=========================================
Team Manager of AMEX - Navigon
Editor of the Cobbler
|
| |
|
|
| ggDonovan |
Posted on 09-10-2014 16:20
|
Breakaway Specialist
Posts: 897
Joined: 08-08-2011
PCM$: 200.00
|
It's true that the BA success more than usual this year but I agree with Mhaley that an in deep analysis of the races should be done before saying that the AI is messing up the game (that in some situations it is).
I played at four CMs (CM4, PCM09, PCM11 and PCM13). From all of them the one that I feel that have better gameplay is the PCM11 (for you?).
I know that it's exiting to play the newest game every MG season but picking an older game with better game play might be an option to look at.
This will lead to losing some of the uncertainty about how the new release of the game will treat certain situations but I've been around for two years now and the discussion about the AI is recurrent, last year with the unselective mountain and hill stages and this year with the flat stages and the BA.
So I put on the table two options:
1) Use the game with the best AI available even if it's old
2) Use the newest game available and accept the flaws as a variable inside the game and don't complain EVER about it
I feel it's pointless to keep discussing about the AI as we will end the season with this version of the game and next year, as it is right now, we will be using the PCM14.
I don't know if before this topic about bringing to live old games but for me it could be something interesting to discuss.
PS: I love the games from the 80s and early 90s so I'm not afraid of the awful graphics  Long live Monkey Island 
|
| |
|
|
| roturn |
Posted on 09-10-2014 16:25
|
Team Manager
Posts: 22246
Joined: 24-11-2007
PCM$: 3900.00
|
One thing to consider as well is obviously the reporter issue.
Eventually all managers/reporters should write somewhere what PCM versions they have.
I mean if only half of the reporters have PCM11, then this is a dead discussion I guess.
Same obviously counts for other versions. |
| |
|
|
| cio93 |
Posted on 09-10-2014 16:32
|
World Champion
Posts: 10845
Joined: 29-10-2007
PCM$: 500.00
|
roturn wrote:
I mean if only half of the reporters have PCM11, then this is a dead discussion I guess.
Same obviously counts for other versions.
PCM 11 is what, 5€ on Amazon? If that guarantees less mental breakdowns, I personally ship one to every reporter without it.
|
| |
|
|
| TheManxMissile |
Posted on 09-10-2014 16:40
|
Tour de France Champion
Posts: 18187
Joined: 12-05-2012
PCM$: 0.00
|
Also don't forget that PCM is not designed for passive play, it's built with the thought that the player will make inputs into the events of the stage. You cannot expect flawless AI in 3D mode.
I like the fact that the game switches over to the new version each season. It gives a nice identity to each year. It also rewards those managers who do know PCM slightly better and how it works in detail so they can pick slightly better riders/teams.
Obviously i'm really not fussed by the AI issues this year because my team is crap and has no interest in sprinting. I do feel sympathy for those managers who are more heavily affected by the breakaways, but this should not have come as a surprise. It was a known issue before transfers and setting your team up around those situations was more risky. At the same time the non-breakaway issues, like when leaders just don't perform, will happen in an version and is just a part of cycling.
There comes a point where PCM becomes redundant and we award everything based purely on who has the best stats, which we would all agree is not the point (or if it is can you please tell me as i misunderstood the game). Also, it's a f*cking game! Take it with a laugh and a smile and it will be a lot better
|
| |
|
|
| baseballlover312 |
Posted on 09-10-2014 16:58
|
Tour de France Champion
Posts: 16429
Joined: 27-07-2011
PCM$: 10438.70
|
I'm all for variation with breakaways. In fact a lot of my team is based on that. But when you build a team nearly completely around a certain discipline, like cobbles in my case, and then you see the exact same breakaway wins and favorite's attack patterns every single time, it gets to a point where it is frustrating beyond belief.
In response to Mhaley, I would say that my team is specifically made for the purpose of being able to catch the break. I made my cobbles team as strong as possible and yet they decide the let the break go instead of controlling the race every single time. That is the most frustrating part.
PCM 11 might be a good option. Most of the complaints I remember about it were based on user gameplay, but the AI ran pretty well aside from the non ability of a player attack in mountains to work, which is a non issue here.
However, I also see it as a last resort, jumping back 3 years is not ideal.
Edited by baseballlover312 on 09-10-2014 17:01
RIP Exxon Duke, David Veilleux, Double Feature, and Monster Energy |
| |
|
|
| Ad Bot |
Posted on 24-11-2024 04:12
|
Bot Agent
Posts: Countless
Joined: 23.11.09
|
|
| IP: None |
|
|
| SportingNonsense |
Posted on 09-10-2014 17:38
|
Team Manager
Posts: 33046
Joined: 08-03-2007
PCM$: 200.00
|
I'm no fan of jumping back PCMs. Whatever faults the AI may have, the organisational part of the MG has got progressively easier and less time consuming with each new PCM.
|
| |
|
|
| DubbelDekker |
Posted on 09-10-2014 17:39
|
Small Tour Specialist
Posts: 2633
Joined: 20-04-2008
PCM$: 200.00
|
Thanks for the interesting responses so far.
My primary reason for starting this thread was indeed to discuss the status quo of simply going with the new PCM every year.
We have many PCM race engines to choose from. I'd love to see us thoroughly test their ability to produce plausible Man-Game results and go with the best instead of the newest.
For me the charm of MG race reports is not in the pixel count of the screenshots.
|
| |
|
|
| DubbelDekker |
Posted on 09-10-2014 17:45
|
Small Tour Specialist
Posts: 2633
Joined: 20-04-2008
PCM$: 200.00
|
SportingNonsense wrote:
I'm no fan of jumping back PCMs. Whatever faults the AI may have, the organisational part of the MG has got progressively easier and less time consuming with each new PCM.
Ah, that's quite a big argument for going with the newer PCM's. Let's hope for a good PCM14 then.
How was PCM12 organization-wise?
Edited by DubbelDekker on 09-10-2014 17:46
|
| |
|
|
| SportingNonsense |
Posted on 09-10-2014 17:56
|
Team Manager
Posts: 33046
Joined: 08-03-2007
PCM$: 200.00
|
DubbelDekker wrote:
SportingNonsense wrote:
I'm no fan of jumping back PCMs. Whatever faults the AI may have, the organisational part of the MG has got progressively easier and less time consuming with each new PCM.
Ah, that's quite a big argument for going with the newer PCM's. Let's hope for a good PCM14 then.
How was PCM12 organization-wise?
The big advantage PCM13 has over PCM12 is the ability to change the rating of a stage in the Stage Editor - whether it is flat, hill or mountain, what the time gap setting should be, and what the H/M ratio is. PCM12 was a right pain to find hill stages that were hard, and had a ratio that actually suited the puncheurs. I got around it in some hill stage races by editing the actual Mountain and Hill stats of riders, to fix the ratio to what I wanted it to be.
It was possible to make these changes in earlier PCMs, but you had to go through every single stage in the database and choose ratings for all of them. Now it is done automatically, and we are able to fix any we want changed - which is ideal.
Compared to PCM11 or earlier, the advantage that PCM12 and PCM13 both share is that it is much easier to add races to the database.
|
| |
|
|
| sammyt93 |
Posted on 09-10-2014 18:32
|
Classics Specialist
Posts: 3634
Joined: 03-07-2012
PCM$: 300.00
|
You've also got to bear in mind the amount of reporters with each PCM version, as the older PCM you choose the more likely it is that people won't have it or won't have it currently installed if they do own it.
|
| |
|
|
| Kentaurus |
Posted on 09-10-2014 20:17
|
Classics Specialist
Posts: 3999
Joined: 26-07-2011
PCM$: 200.00
|
Just to put my two cents in, perhaps we could actually have the Control Team not be passive?
If you set all of the control riders on relay at 45-50 effort (pick a set number and keep it there), this would make breakaways much more challenging as the breakaway group would not be able to gain as big of a gap over the field. If other teams wanted to pick up the pace they still would do so, especially in the last half of the race.
The control team could maintain that tempo the whole race no problem, but would not impact the finish in any way.
Edited by Kentaurus on 09-10-2014 20:18
AZTECA - NBCSN  |
| |
|
|
| tsmoha |
Posted on 09-10-2014 20:38
|
Directeur Sportif
Posts: 11819
Joined: 19-07-2010
PCM$: 300.00
|
Kentaurus wrote:
Just to put my two cents in, perhaps we could actually have the Control Team not be passive?
If you set all of the control riders on relay at 45-50 effort (pick a set number and keep it there), this would make breakaways much more challenging as the breakaway group would not be able to gain as big of a gap over the field. If other teams wanted to pick up the pace they still would do so, especially in the last half of the race.
The control team could maintain that tempo the whole race no problem, but would not impact the finish in any way.
Guess this won't work. This won't prevent the peloton from stopping/sitting up, wherever it "decides" to do so. A steady chase by the control team wouldn't catch the break (which shouldn't happen anyway, cause of course breakaways should be able to win a race - even if it's flat classic). The annoying thing is the peloton just stopping (sometimes close before finish or when the gap is close enough to catch the break easily) and that won't be "repaired" by a "half-active-half-passive" human control. Just my guesses though.
Not talking about the other problems of this idea (no offense here) 
|
| |
|
|
| The Rider |
Posted on 09-10-2014 20:41
|
Grand Tour Specialist
Posts: 4543
Joined: 29-02-2012
PCM$: 200.00
|
For me with PCM2013 the breakaway AI has it so the break has too much energy and ability left in the final 20 kilometres, even if they have bad stats. So you will see riders with 65 flat and below going well over 50 km/h on the flat for many kilometres.....
That is the problem (the break AI being too strong), not so much the control team so there isn't much anyone can do about it. |
| |
|
|
| Kentaurus |
Posted on 09-10-2014 21:06
|
Classics Specialist
Posts: 3999
Joined: 26-07-2011
PCM$: 200.00
|
tsmoha wrote:
Kentaurus wrote:
Just to put my two cents in, perhaps we could actually have the Control Team not be passive?
If you set all of the control riders on relay at 45-50 effort (pick a set number and keep it there), this would make breakaways much more challenging as the breakaway group would not be able to gain as big of a gap over the field. If other teams wanted to pick up the pace they still would do so, especially in the last half of the race.
The control team could maintain that tempo the whole race no problem, but would not impact the finish in any way.
Guess this won't work. This won't prevent the peloton from stopping/sitting up, wherever it "decides" to do so. A steady chase by the control team wouldn't catch the break (which shouldn't happen anyway, cause of course breakaways should be able to win a race - even if it's flat classic). The annoying thing is the peloton just stopping (sometimes close before finish or when the gap is close enough to catch the break easily) and that won't be "repaired" by a "half-active-half-passive" human control. Just my guesses though.
Not talking about the other problems of this idea (no offense here)
Not sure what you mean, but it will stop the Peloton from the stopping/sitting up as they will always stay with that small (and relatively slow) relay. The steady chase at that speed isn't supposed to catch the break but rather keep them from getting the 10+ minutes they tend to be able to get. And will keep the main contenders in better condition as the teams won't completely burn out trying to make the catch (which is why they stop near the finish of a race).
Edited by Kentaurus on 09-10-2014 21:07
AZTECA - NBCSN |
| |
|
![imgur.com/KlJUqDx.jpg[/img[img]]https://imgur.com/yUygrQ.jpg](https://imgur.com/KlJUqDx.jpg[/img[img]]https://imgur.com/yUygrQ.jpg)
