Papy on 5/7/2010 at 06:57
Quote Posted by Chade
Playing against an AI which has no hope of winning is just masturbation.
Even if it's not its goal, isn't the effect of adaptive difficulty to make a game where the AI has, in the end, no hope of winning?
Quote Posted by Chade
This is very much a "worst case" scenario for dynamic difficulty, however ... 99% of games out there are not so deterministic.
We are all pretty good at noticing trends which are very subtle. In fact, we are so skewed to notice trends that we even see trends that do not exist at all. Because of that, I believe an adaptive AI in a less "deterministic" game will still be easily noticed after a bit of time and the player will adapt its way of playing because of this.
Of course, a casual player might not have the time to notice adaptive difficulty in a single game. But if adaptive difficulty becomes something "normal" with video games, then even the casual player will expect it and he will adapt his way of playing because of this. If he knows that if he tries hard to succeed then the only result will be an even more difficult game, I'm guessing he will simply and voluntarily do less. A game should reward effort, it should not punish it.
Quote Posted by Chade
Finally, moving away from the rubberband effect, there is no fundamental reason why dynamic difficulty adjustment has to take effect within the confines of a single race (or a single level or region in an fps or rpg).
Yes, but in this case, don't you think it becomes pointless to have adaptive difficulty?
Quote Posted by Chade
Human psychology is not that complex, at least within the confines of a game.
That paper was not about human psychology, but about trying to use data mining techniques on a very limited set of data (and in this case, with obviously no useful results). Correlation about game completion time, number of deaths, cause of deaths and the number of times the player ask for help to solve a puzzle certainly don't even begin to answer if the player liked the game, if he thought it was too hard or too easy, if it was fun, frustrating or boring.
Having said that, I will admit that maybe the reason I think this paper is worthless (also maybe why I'm against adaptive difficulty) is because my goal as a player seems to be different than the goal of the developers who make the game. My goal is to play the game the way I like it, not to play the game as intended by the game designer.
Quote Posted by Chade
Keep in mind that any algorithm doesn't have to perfect, it just has to perform better then the player choosing their own difficulty ... a very low bar to clear!
What? Are you completely out of your mind? The fact is I will be always better at selecting my own difficulty level than any algorithm you can think of. In case your fear is that people are dumb and forget they can adjust the difficulty level while in the middle of a game, you can always remind them with a loading screen or something.
Anyway, now I know why the soda machine gave lemon-lime to Gunther instead of orange like he wanted. It was not because of some kind of plot or because of the maintenance man wanted to piss him off, but simply because you wrote the software of the soda machine and the algorithm you wrote came to the conclusion that lemon-lime was what Gunther really wanted.
Jason Moyer on 5/7/2010 at 09:17
Quote Posted by Papy
Even if it's not its goal, isn't the effect of adaptive difficulty to make a game where the AI has, in the end, no hope of winning?
It could also mean the player(s) has no hope of winning. Or that the chances of winning/losing are artificially modified one way or the other. In any case it fixes the outcome to a certain degree.
Racing games have traditionally had a slider that lets you adjust the overall AI speed to suit your tastes, although as with all things useful that's been basically nonexistent on consoles or console-derived games.
Chade on 5/7/2010 at 22:28
Quote Posted by Papy
Even if it's not its goal, isn't the effect of adaptive difficulty to make a game where the AI has, in the end, no hope of winning?
...
Yes, but in this case, don't you think it becomes pointless to have adaptive difficulty?
No and no. I'm not sure where you're coming from here ... I think we are imaging very different things.
Quote Posted by Papy
If he knows that if he tries hard to succeed then the only result will be an even more difficult game, I'm guessing he will simply and voluntarily do less. A game should reward effort, it should not punish it.
You keep assuming that the dynamic adjustment will be so powerful it will undo all the reward given for being better at the game. I would consider that a sign of a poor algorithm.
Any game will alternate between throwing up challenges and offering rewards. There is no reasons that dynamic difficulty should result in less rewards being offered. For instance, rpgs commonly ensure that bigger challenges result in bigger rewards. (Of course, open ended rpgs typically offer the player his choice of several challenges of varying difficulty, thus allowing the player to continuously adjust the difficulty right throughout the game.)
Quote Posted by Papy
That paper was not about human psychology, but about trying to use data mining techniques on a very limited set of data (and in this case, with obviously no useful results). Correlation about game completion time, number of deaths, cause of deaths and the number of times the player ask for help to solve a puzzle certainly don't even begin to answer if the player liked the game, if he thought it was too hard or too easy, if it was fun, frustrating or boring.
It is possible to tell (roughly) how much the player is enjoying himself by observing his play style. If the researcher can access the players, he/she can ask players how much they enjoyed the game, and compare that data to the predictions coming from their models. The results show (
http://cswww.essex.ac.uk/cig/2007/papers/2024.pdf) decent predictive power.
Quote Posted by Papy
What? Are you completely out of your mind? The fact is I will be always better at selecting my own difficulty level than any algorithm you can think of. In case your fear is that people are dumb and forget they can adjust the difficulty level while in the middle of a game, you can always remind them with a loading screen or something.
No, Papy, the
facts are that people aren't good at selecting their own difficulty when presented with a difficulty screen before the game starts.
Well, I call it a fact, but I've long since forgotten the source, so I really shouldn't be so cocky ... some of the papers I looked at yesterday also repeated this "fact", and also didn't provide a source, which I found quite frustrating. That being said, I do distinctly remember coming across a study that found that most people select "normal" difficulty level regardless of skill.
And this has nothing to do with intelligence. I would not expect the player to be able to map between the arbitrary difficulty level names and his consequent game experience when he hasn't even experienced the game yet!
Papy on 6/7/2010 at 03:54
Quote Posted by Chade
No and no. I'm not sure where you're coming from here ... I think we are imaging very different things.
For me, the idea of adaptive difficulty is about the game adjusting difficulty to make sure the player has the right "challenge". If the player dispatches enemies "easily", if he has "lots" of resources, then the game will become harder to make sure the player is not bored with the game. On the other hand, if the player dies "often", if he's "low on resources", then the game will become easier to make sure the player can go through the game and does not become frustrated.
That's a great idea in theory, except in practice it just doesn't work. First, the idea of "easily", "lots", "often" and "low" is a very personal feeling. With BioShock, I think I died about 3 times (medium difficulty). For me that's reasonable. Dying 10 times would have been a sign that I was playing at a difficulty to high for my skill. For other people, dying 50 times would just be normal and if they don't die every 5 minutes they'd just raise the difficulty level. It's all about personal taste and no adaptive difficulty algorithm can know that.
Second, if the player know or realize the game will change its difficulty because of his result, then he will change his way of playing. I'd even say this will kill his motivation to do well. That's exactly what happened with Final Lap. We all used the rubberband effect to our advantage and to make the game easy to win. Instead of always trying our best, we voluntarily reduce our speed to fool the adaptive difficulty.
As a side effect, this will kill most feelings of immersion as he will stay aware of the mechanics of the game.
Third, there is a big difference when you say to your friend "I beat the game at hard" (while your friend only beat the game on medium), and saying "I beat the game", knowing that adaptive difficulty can adjust the game for even the 90 years old woman who never played video games before (I'm exaggerating, but you get the point).
Fourth, ... Well, I could continue for long time, but I'll stop there for now. My point is there are many things wrong with adaptive difficulty.
Ok... Here's an example of a real life adaptive difficulty. When I play badminton against someone who has no chance of winning, I will play easy with him. I won't smash. My clears won't be as far and my drop shots not as near the net. The idea is to make sure my opponent can touch the shuttlecock once in a while. But you know what? this is not playing, this is training. It is still somewhat fun, but not near as fun as if it was a real game.
So... What is adaptive difficulty for you?
Quote Posted by Chade
You keep assuming that the dynamic adjustment will be so powerful it will undo all the reward given for being better at the game. I would consider that a sign of a poor algorithm.
What I'm saying is that any step into dynamic adjustment will destroy part of challenges and rewards. You cannot have your cake and eat it too. There is no good algorithm.
Quote Posted by Chade
Any game will alternate between throwing up challenges and offering rewards. There is no reasons that dynamic difficulty should result in less rewards being offered. For instance, rpgs commonly ensure that bigger challenges result in bigger rewards.
The discussion about what is a reward in the context of a video game is very complex. For example, with Oblivion, I stop caring about any "reward" quite fast. In fact, a lot of people did. Do you guess why? Can you tell me why each augmentation canister was (at least for me) a reward with Deus Ex, while biomods with Invisible War were mostly uninteresting?
This whole discussion is going nowhere because we are talking about a lot of different things at the same time. An arcade racing game is very different than an RPG. Also, when you talk about adaptive difficulty, you never define what it is. So how about we limit the scope of things and you describe what you really mean with adaptive difficulty and how you think it will be implemented?
Quote Posted by Chade
It is possible to tell (roughly) how much the player is enjoying himself by observing his play style.
No, it's not. Not at all. I explained a few days ago how I played Deus Ex. For me, it was playing very slowly and trying to plan everything in advance. That's how I enjoyed it. The reply I had was this way of playing must have been very boring. I'm sure that person had a lot more fun going fast, shooting everything, quicksaving regularly and reloading after dying. Personally, when I do this, when I try to go as fast as possible, it is not because I'm eager to advance and to see what's next, it is because I find the game utterly boring and I want to end it as fast as possible.
So if you know who I am, then by looking at my style of play you can (roughly) tell if I'm enjoying myself. But if you don't know who I am, then you simply can't as two completely different feelings can lead to the same play style.
Quote Posted by Chade
No, Papy, the
facts are that people aren't good at selecting their own difficulty when presented with a difficulty screen before the game starts.
The Fact is I'm quite good at selecting my own difficulty level. Generally, I try the most difficult when there are three difficulty level, or the third when there are four. I then play for about 30 minutes to have a feeling of the game. If I'm satisfied, I continue, if not I restart with a different difficulty level. The only time when I'm not satisfied is when the highest difficulty is still not enough (like with Invisible War).
Chade on 7/7/2010 at 22:19
Quote Posted by Papy
For me, the idea of adaptive difficulty is about the game adjusting difficulty to make sure the player has the right "challenge".
That's the basic idea, but ...
Quote Posted by Papy
Second, if the player know or realize the game will change its difficulty because of his result, then he will change his way of playing. I'd even say this will kill his motivation to do well ... Third, there is a big difference when you say to your friend "I beat the game at hard" (while your friend only beat the game on medium), and saying "I beat the game"
... none of this necessarily follows.
People strive to maximize their rewards, usually defined by whatever reward system the game offers. With Final Lap and it's poor dynamic difficulty algorithm, the player was rewarded for winning a race, regardless of how difficult the race was. That's not inevitable. Secondly, the dynamic difficulty algorithm over-compensated for your performance. That's not inevitable either.
There's also no reason that a dynamic algorithm can't give you feedback about what level you are "beating the game" at. That can be an in-game reward, or something that exists outside the game (just like selecting a difficulty level).
Finally, while dynamic difficulty adjustment is usually associated with the game observing the player and selecting an appropriate difficulty for him, there's no fundamental reason why you can't use the same observation techniques to observe the player and then offer a range of difficulty levels.
Quote Posted by Papy
It's all about personal taste and no adaptive difficulty algorithm can know that.
As the link I posted before shows, people's tastes are not quite as unpredictable as you make them out to be. That being said, ...
Quote Posted by Papy
I ... I ... I ... I ... I ...
... you do have fairly unique tastes. Deus Ex is a game that encourages a slow methodical approach, and many people play it like that (if you analysed people's DX playstyle, I'm sure you would find an "explorer" category), but you take it to the extremes.
You're an outlier, however, so frankly I don't think your personal experience adds much to the discussion.
Although I wonder if we could predict your preference for iron-man challenges by looking at the saved game you select when you die ... ;)
Quote Posted by Papy
Also, when you talk about adaptive difficulty, you never define what it is. So how about we limit the scope of things and you describe what you really mean with adaptive difficulty and how you think it will be implemented?
While I appreciate your distaste for generic arguments, I am trying to convince you that dynamic difficulty can come in many different forms. So
not narrowing the scope of things is really the entire point!
And in fact, this argument has actually become narrower then I am comfortable with, because originally I wasn't just advocating dynamic difficulty adjustment, bur rather any alternative to the poor method of asking the player to select a difficulty before he has played the game (which includes dynamic difficulty adjustment, as well as an array of methods through which the player can make things easier or harder for himself during play).
Papy on 12/7/2010 at 09:03
Quote Posted by Chade
People strive to maximize their rewards, usually defined by whatever reward system the game offers.
The problem I have with this is the reward systems of games is now somehow returning to what it was 30 years ago. It is now more and more outside of the game world. Achievements are about the same thing as the high score of pac-man. For me, it is a step backward and a sign the gameplay provides no reward feeling in itself.
Quote Posted by Chade
With Final Lap and it's poor dynamic difficulty algorithm, the player was rewarded for winning a race, regardless of how difficult the race was. That's not inevitable. Secondly, the dynamic difficulty algorithm over-compensated for your performance. That's not inevitable either.
I wouldn't have qualified Final Lap's dynamic difficulty algorithm as poor. If the player always tried his best, then winning a race would still have been a challenge. The problem is the minute the player realized he got a boost when he was behind, he then began to abuse this dynamic difficulty system in order to "cheat".
I guess you could try to make an algorithm that is trying to detect when the player tries to "cheat". But this will only lead to an arm race... and I believe that in the end the player will always win.
Quote Posted by Chade
There's also no reason that a dynamic algorithm can't give you feedback about what level you are "beating the game" at.
The level the player "beating the game at" will probably go up and down depending on how he is doing. The result will be exactly the same as with Final Lap. The player will choose to play at a lower level than what he is capable of in order to get more resources, and then, in the end, he will suddenly raise his level of play in order to achieve a better ending "beat level".
If you choose a cumulative or an average score, then you will penalize the player who could still beat the level without any help from the dynamic difficulty. For example, if the player has no more ammo, the game might decide to lower the difficulty level so he can find more. But what if the player could still manage to survive with his wrench? The fact is some people perform best when they are under stress, while others perform best when they feel secure. Will your algorithm also take that into consideration? Don't you think the algorithm is becoming an endless problem?
Quote Posted by Chade
That can be an in-game reward, or something that exists outside the game (just like selecting a difficulty level).
Selecting a difficulty level is not about getting an external reward, it is about defining our own challenge. It is a completely different thing.
Quote Posted by Chade
Finally, while dynamic difficulty adjustment is usually associated with the game observing the player and selecting an appropriate difficulty for him, there's no fundamental reason why you can't use the same observation techniques to observe the player and then offer a range of difficulty levels.
Either you believe people can't select their own difficulty level, or you believe they can. So what would be the point of this compromise?
Quote Posted by Chade
As the link I posted before shows, people's tastes are not quite as unpredictable as you make them out to be.
Sorry if I'm a bit harsh, but, as I said, I think the link you posted just proves that it is possible to use data mining techniques in order to show what you want to show. More specifically, if you use a very limited set of data as the source of information, you will get a simplistic (if not very misleading) result at the end.
Again, to me what is important is not how people play the game, but if they like what they are doing of not. And even that piece of information may be misleading. If they choose to systematically skip an optional gameplay, you cannot infer they wouldn't like it after an initial learning period.
Quote Posted by Chade
... you do have fairly unique tastes.
I don't think so. I know myself and I a play a wide range of games, so I have a fairly good idea of how to get something I like out of a game, but I think my taste for challenging games is far from being unique. I think Gothic was a great game and Oblivion was a piece of shit. Considering what I read on Bethesda and Jowood forums, I know I'm not the only one to think that way.
Quote Posted by Chade
Deus Ex is a game that encourages a slow methodical approach, and many people play it like that (if you analysed people's DX playstyle, I'm sure you would find an "explorer" category), but you take it to the extremes.
Extreme? I have another explanation... I don't really like FPS. To me, shooting is boring. When I find a way to avoid shooting, it is a reward for me. Seriously, compared to the people who can play Bejeweled for hundred of hours, I think spending 15 minutes watching guards in order to take find the best moment to take them out with the baton is less extreme.
Quote Posted by Chade
You're an outlier, however, so frankly I don't think your personal experience adds much to the discussion.
So you think I should stop playing games right now because I'm not a part of the majority who view video games only as an easy pastime to replace watching TV?
Quote Posted by Chade
Although I wonder if we could predict your preference for iron-man challenges by looking at the saved game you select when you die ... ;)
Wrong conclusion from your part. The same way I have no interest for limiting myself when I play a game and generally choose the easiest solution, I have no interest for iron-man challenges. When I played Medal of Honor, I was saving at least every two minutes. I didn't save with Deus Ex because I loved the feeling of relief after the tension. As I was not dying often, it never became frustrating. I was not seeking an iron-man challenge, I was only doing what I liked.
By the way, I said that I don't like to limit myself, but Deus Ex was a bit of an exception. When I realized my side was not the "good" one, I decided not to kill NSF anymore. But that kind of exceptions are rare.
Quote Posted by Chade
While I appreciate your distaste for generic arguments, I am trying to convince you that dynamic difficulty can come in many different forms. So
not narrowing the scope of things is really the entire point!
If you can't define precisely what you are talking about, then you are just talking about a fuzzy intuition... and the discussion is pretty much pointless.
Quote Posted by Chade
And in fact, this argument has actually become narrower then I am comfortable with, because originally I wasn't just advocating dynamic difficulty adjustment, bur rather any alternative to the poor method of asking the player to select a difficulty before he has played the game (which includes dynamic difficulty adjustment, as well as an array of methods through which the player can make things easier or harder for himself during play).
Do you think an alternative method is necessary for all kind of games, or do you think this is important only for a limited number of games? Do you think that Grand Prix would have been a better game with an alternate method?
Chade on 15/7/2010 at 22:04
In case it's not already obvious, I'm finding it increasingly hard to find the time to hammer out another one of these mammoth posts ... so I'm pretty much going to ignore most of what we've been arguing about so far.
Quote Posted by Papy
The problem I have with this is the reward systems of games is now somehow returning to what it was 30 years ago. It is now more and more outside of the game world. Achievements are about the same thing as the high score of pac-man. For me, it is a step backward and a sign the gameplay provides no reward feeling in itself.
Totally branching out here, but I'll certainly agree that the industry hasn't gone in the direction I thought it would a decade or so ago. I don't know if this qualifies as a step backwards, though ... I think we were all a bit naive in glorifying immersion so much. In hindsight, it seems to me that most people aren't all that interested in being deeply immersed in a fictional world for extended periods of time.
Quote Posted by Papy
The problem is the minute the player realized he got a boost when he was behind, he then began to abuse this dynamic difficulty system in order to "cheat" ... I guess you could try to make an algorithm that is trying to detect when the player tries to "cheat".
Now, I keep saying this, and if there's one thing I'm going to achieve in this argument, I am going to convince you that there is
absolutely no reason that the player can gain an advantage by cheating!
Think about it, how does the player cheat? He intentionally does poorly! Dynamic difficulty allows him to make up some of this lost ground later by cheating, but it is by no means necessary that the game must allow the player to make up more ground due to dynamic difficulty then he lost in order to cheat in the first place.
Let's say that we have a very simple algorithm where the player has max speed 150 km/hr, and the AI races with speed 75 + (average player speed over last 10 seconds)/2 km/hr ... the player can match the AI by playing consistently at his best, but he cannot improve his results by reducing his speed. At best, the AI will lose only half the distance the player loses. At worst, if the player slows down right near the end, the AI loses almost none of the distance the player lost.
There's a whole bunch of other stuff that I could reply too, but it's just taking up too much time. Hopefully you don't interpret this as me trying to weasel my way out of the argument. :erg:
Papy on 16/7/2010 at 03:20
Quote Posted by Chade
In hindsight, it seems to me that most people aren't all that interested in being deeply immersed in a fictional world for extended periods of time.
I agree with that, but I think the main reason is video games fictional worlds are extremely superficial and feel ridiculously artificial. There are so many flaws that very few people are able to let go enough (and imagine all the things that are not there) in order to have a feeling of immersion.
With video games having a fictional world, I can think of small moments where I felt some emotions. But that's all they are : small moments. That's not good enough.
Quote Posted by Chade
Now, I keep saying this, and if there's one thing I'm going to achieve in this argument, I am going to convince you that there is
absolutely no reason that the player can gain an advantage by cheating!
With your example, you basically start with a fixed difficulty level, and then adjust that difficulty to make the player who is better than that fixed level win with a less impressive margin, and you make the player who is worse than that fixed level lose in a less shameful way. But in the end, it is a fixed difficulty level.
(Actually, your example makes the AI unbeatable, the "75" should be lower, but I get your point)
Ok. I will assert two things... Tell me if you agree with them or not.
First, any dynamic difficulty level MUST have a delay before changing the difficultly level, either up or down. Also, those changes must be progressive and the delay must be large enough to make sure the player can't see it.
Second, all dynamic difficulty system start with a fixed difficulty which is independent of the player. That means the adaptive difficulty will HAVE TO become worse or better than the player in order to compensate for previous misjudgment from the dynamic system. Adjustment toward the player is not sufficient.
So now... what if the player try to hide his own skill in order to fool the dynamic difficulty system and suddenly raise his level of play near the end? Because of the necessary delay of reaction from the adaptive system, don't you think it will be an easy win for the player?
Quote Posted by Chade
There's a whole bunch of other stuff that I could reply too, but it's just taking up too much time. Hopefully you don't interpret this as me trying to weasel my way out of the argument. :erg:
Don't worry, I don't.
Thirith on 16/7/2010 at 05:52
Quote Posted by Papy
So now... what if the player try to hide his own skill in order to fool the dynamic difficulty system and suddenly raise his level of play near the end? Because of the necessary delay of reaction from the adaptive system, don't you think it will be an easy win for the player?
I know that you're mainly having this conversation with Chade, but this bit made me wonder: why would this be a problem? If a player wants to 'cheat' in some way or another, they can. Most if not all games have systems that can be abused. Is it worth putting in the effort to make sure the player can't do that, as long as he'd have to go out of his way to do so? Most players won't play a game trying to fool it into thinking they're less good than they are, because that takes an extended effort and, frankly, isn't likely to be very enjoyable.
Papy on 18/7/2010 at 15:59
Quote Posted by Thirith
Most players won't play a game trying to fool it into thinking they're less good than they are, because that takes an extended effort and, frankly, isn't likely to be very enjoyable.
If the player needed to actively find ways to fool the system, then I would agree. But that's not the case. It's the system which will adapt to the player. The only thing the player will have to do is simply take it easy (and maybe just do a final sprint at the end). That's pretty much a natural state and does not require more effort than playing his best. As for being enjoyable, I'd say it will be a lot less rewarding to play carelessly, but most of us are naturally lazy. We all like going out and having fun, but that requires some effort and many times we end up saying we're tired after our work day and stay home... bored and regretting not going out.
For me, the question is : how do you make a game with adaptive difficulty rewarding? How to you push the player to do his best so he can end up satisfied with the game rather than bored?
