Screw Attack E3 videos

Nihilism on 30/6/2010 at 17:51

Quote Posted by negativeliberty

Sorry I see you only got round to DX about 3 months ago. But if you bought both on Steam then why would the metascore put you off playing IW, or are you still replaying DX currently?

The metascore hasn't put me off; it actually isn't a bad score. It's the word of mouth disdain I've come to witness. On top of that, there are many other games I want to check out, so I have to prioritize. I'm already caught up in Dragon Age at the moment.

It's up to them to think outside the box in this regard. I just thought of an example of a choice with unforseen consequences: In Mass Effect 2, you're supposed to rescue your crew on whatever planet -- SPOILERS (if you care) -- and if you don't go immediately, they die. The game doesn't tell you this will happen, and with game events not unfolding in real time, you wouldn't expect it to. Fallout 3 could have done things like that too, but they just didn't think to.

I think we can agree that the intrigue of a story is not knowing what happens next. With choices the way they are, we can always make a fair guess as to what the outcomes are. So why not introduce an element of ambiguity to the recipe.

You guys are right. At least Fallout 3 has branching narrative depending on your limited choices. But my point was, for a game touted for its choice system, it really doesn't offer any compelling choices. So the game is an example of the primitive status of choice the industry has yet to evolve on with any significance.

Chade on 30/6/2010 at 23:36

Quote Posted by Muzman

But anyway, I'm only on this track because it was suggested modern gamers wouldn't want a multi slider difficulty system like System Shock. Whether that was the point or not, I do think the industry would broadly agree and I reckon they'd be wrong.

I guess the reason I'm saying we should discuss "the issues separately", is that I don't believe that a difficulty system like that actually works in the first place. There's no point discussing usability concerns, if the act of selecting game difficulty prior to starting the game doesn't help gamers choose an appropriate difficulty setting in the first place.

If the player can change difficulty throughout the game, having some idea of whether the game is currently too hard or too easy, then we can start talking about how to implement the feature ... but, if this is something that the player is going to do throughout the game, then there's no reason to have an external difficulty menu any more! Players can choose an appropriate difficulty in-game by trading achievements for safety, re-ordering tasks, spending more/less time farming for resources, etc ...

Quote Posted by Jason Moyer

Among whom? Rubberband AI is the first thing most reviewers, not to mention genre fans, bitch about if a racing game implements it

Is that rubberband AI, or unsubtle rubberband AI? When reading the literature some time ago (I was trying to decide what to do my honours in, although I didn't end up doing anything to do with dynamic difficulty), I got the impression that rubberband AI was extremely common in racing games.

Papy on 1/7/2010 at 00:57

Quote Posted by Chade

The effect of a rubberband AI that is too subtle for the player to notice is also too weak to serve its purpose. Anyway, the rubberband effect is only useful for arcade machines when players compete against each other, not for AI. To me, an AI adaptive difficulty system for games that you replay over and over is pointless. I guess it could be used to add a bit of suspense to the game, but I don't think this is worth it.

The problem with adaptive diffilculty is that the game has to guess how the player feels (frustrated or bored), but also what he expects. When I play an RPG, I don't expect to die. Never. That's one of the reasons I don't save. For me, dying once is enormous failure. I played BioShock on medium difficulty because of that. I tried "hard", but I died a few times, I didn't like it so I restarted the game at medium and I found that level fun. I think I died two or three times during the whole game and that was exactly what I expected. But for other people, they expect to die frequently. If they don't, then they will try a more difficult level. So how the adaptive difficulty is supposed to know what the player wants? If you make the adaptive difficulty for the player who think dying a lot is normal, then I will feel frustrated with the game. If on the other hand you tailor the adaptive difficulty for someone like me, than the guy whose style of gameplay include dying a lot will find the game too easy and end up being bored.

You can think of a lot of statistics to base an adaptive difficulty system (like the shooting hit/miss ratio), but in the end, nothing really works because each person has different expectations. The only one who knows what I expect in a game is me, so let me adjust my own difficulty.

As for completely removing an external difficulty menu for everything, that's also a bad idea. Setting a external difficulty is my way to create my own challenge and then doing everything I can to win at that level. Look at it as a high jump competition. It certainly would be possible to have several bars (let's say one every 10 cm) so the goal would be to jump as high as you can and those bars would be only the measure the height of our jump, but don't you think it's more fun to have only one bar and to try to pass over it?

Jason Moyer on 1/7/2010 at 01:00

It's common and it sucks. What's the point in a game that models a competitive exercise if the AI is designed to artificially manipulate the results? Most people who are serious about racing games do it against humans, although I guess you could say the same thing about shooters or flight sims or strategy games or whatever.

Chade on 1/7/2010 at 01:30

Quote Posted by Papy

The effect of a rubberband AI that is too subtle for the player to notice is also too weak to serve its purpose.

Says who? How will you back up such a statement?

Quote Posted by Papy

Look at it as a high jump competition. It certainly would be possible to have several bars (let's say one every 10 cm) so the goal would be to jump as high as you can and those bars would be only the measure the height of our jump, but don't you think it's more fun to have only one bar and to try to pass over it?

No, the appropriate metaphor is a long jumper approaching a high jump for the first time in his life, being asked to permanently set the one and only bar that he needs to pass. What you describe as ideal is one of many more reasonable alternatives: an experienced high jumper setting a challenging goal for himself, which he will raise once he meets it.

Quote Posted by Jason Moyer

It's common and it sucks. ... Most people who are serious about racing games do it against humans, although I guess you could say the same thing about shooters or flight sims or strategy games or whatever.

Why is it so common? In a genre as old and well understood as racing, I would not expect designers to continue making such an elementary mistake. It's reasonable to believe that play testing show the alternative sucks more (e.g., the player spending 90% of his time alone on an empty race track). Do you have any evidence that a better alternative exists? (Lets limit the discussion to one player racing against a dozen AI on a reasonably large track.)

As for "serious gamers" choosing multi-player, that's no argument ... the same holds true for any genre. How perfect do you expect an AI to be?!

Papy on 1/7/2010 at 02:48

Quote Posted by Chade

Says who? How will you back up such a statement?

A few years ago, I played a lot with Grand Prix 2. We were a bunch of friends doing full races and our laps were extremely regular. We knew exactly how cars were supposed to handle and when one of us was doing a mistake. Still, most of the time, those mistakes were not enough to make a difference.

A rubberband effect must be strong enough to make a difference. So if we were able to spot small changes which were still not enough to really change anything, I kind of guess it's obvious we would be able to see a change strong enough to effectively do something, either in our car or in the cars of others.

Quote Posted by Chade

Why "permanently set"? Where did I say that the player had to replay the game always at the same difficulty level? Anyway, setting a challenging goal for myself, which I will raise once I meet it, is exactly what I do with a difficulty level. So, I'd like to know, how can I set my challenging goal if I can't select a difficulty level?

I agree that some games do not need a difficulty level in a menu. Trackmania is one of them. Unfortunately, a shooter, or any game which are "do or die", needs one.

Quote Posted by Chade

Why is it so common? In a genre as old and well understood as racing, I would not expect designers to continue making such an elementary mistake.

If difficulty levels were a mistake, I would not expect designers to continue making such an elementary mistake...

You know, I'm probably not as stubborn as I look and I could change my mind quite easily. The problem is you are talking about adaptive difficulty as a broad idea, without any details on the implementation. Personally, I think if a game could adapt its difficulty so the player could have exactly what he wants, it would be marvelous. Unfortunately, I also think my computer has absolutely no idea about how I feel and what I want at a particular moment when I play a game. So how could it give me what I want if it has no way of knowing what I want?

Personally, in the case of an RPG, I think what would be an acceptable solution (at least for me) would be to have the player do a training mission which would allow him to compare the different difficulty level before playing the real game. Do you think this could be acceptable to you?

Jason Moyer on 1/7/2010 at 16:04

I'm fairly certain the Grand Prix games don't use a rubberband effect. That's also the only racing series whose AI I'd consider to be competent - not necessarily fast unless you adjust the driver skills with a third party editor, but they tend to exhibit human behavior that I've never seen in another racing title, sim or otherwise. Grand Prix 4 is fairly primitive even for its time in many ways, but the singleplayer experience can be very rewarding.

Chade on 2/7/2010 at 05:53

Quote Posted by Papy

So if we were able to spot small changes which were still not enough to really change anything, I kind of guess it's obvious we would be able to see a change strong enough to effectively do something

Quote Posted by Jason Moyer

I'm fairly certain the Grand Prix games don't use a rubberband effect.

I don't want to get bogged down on the issue of whether one particular game uses rubber band AI. From the arguments you are making, however, it appears that neither of you see anything wrong with saying "err ... well, I never saw it" ... a statement I just can't imagine taking seriously! (For instance, changes could occur on the other side of the track outside your pov.)

Quote Posted by Jason Moyer

That's [Grand Prix] also the only racing series whose AI I'd consider to be competent

While I am suitably impressed with your refined tastes, I find it hard to believe that the Grand Prix series is the only racing series whose single player component appealed to the broader market.

Quote Posted by Papy

Why "permanently set"? Where did I say that the player had to replay the game always at the same difficulty level?

This is a limitation with the metaphor. A high jump competition has no mechanic that mirrors the hours of progress you lose when you change difficulty in many games.

Quote Posted by Papy

If difficulty levels were a mistake, I would not expect designers to continue making such an elementary mistake...

Well, there is a movement away from permanently fixing the difficulty when starting a new game!

But sure, people usually won't completely screw up their experience if they choose a middling difficulty setting rather then an extreme one. And they may refine their selection by starting the game again. And the alternatives are not always easy to do well - and are still an area of active research. Just because better alternatives exist doesn't mean that they are easy to implement or that it is super urgent to switch to them.

You know, I'm probably not as stubborn as I look and I could change my mind quite easily. The problem is you are talking about adaptive difficulty as a broad idea, without any details on the implementation. Personally, I think if a game could adapt its difficulty so the player could have exactly what he wants, it would be marvelous.

Quote Posted by Papy

Unfortunately, I also think my computer has absolutely no idea about how I feel and what I want at a particular moment when I play a game. So how could it give me what I want if it has no way of knowing what I want? ... Personally, in the case of an RPG, I think what would be an acceptable solution (at least for me) would be to have the player do a training mission which would allow him to compare the different difficulty level before playing the real game. Do you think this could be acceptable to you?

Player modelling is an active area of research, btw.

As far as "acceptable solutions" go, I would hate to limit myself into rating difficulty systems on a binary scale. The training mission you suggest would certainly be an improvement in general. The best solution will no doubt vary on a case by case basis.

Papy on 2/7/2010 at 09:56

Quote Posted by Chade

IFrom the arguments you are making, however, it appears that neither of you see anything wrong with saying "err ... well, I never saw it" ... a statement I just can't imagine taking seriously! (For instance, changes could occur on the other side of the track outside your pov.)

I don't understand what you meant. As for Grand Prix, it didn't have rubberband effect.

I think there is one thing you miss in what I said. Once we play a bit with a racing game, all our laps are about the same. They become extremely regular. Most of times our laps were within one ot two tenth of a second from each other (no, I'm not exaggerating)... and we expect the same kind of behavior from the AI, no matter if it's in front or behind us. If we easily pass an AI, then we don't expect this same AI to pass us again. Of course it's a lot more complicated than that, but I'm sure you get the point. So again, a rubberband effect or any kind of adaptive difficulty within one race is obvious for anyone who is not a novice. You cannot hide it.

Quote Posted by Chade

And the alternatives are not always easy to do well - and are still an area of active research. Just because better alternatives exist doesn't mean that they are easy to implement or that it is super urgent to switch to them.

Until those "better" alternatives are implemented, they do not exist yet. Personally, I don't think it can be done. Human psychology is too complex.

Chade on 5/7/2010 at 01:06

Quote Posted by Papy

I think there is one thing you miss in what I said. Once we play a bit with a racing game, all our laps are about the same. They become extremely regular. Most of times our laps were within one ot two tenth of a second from each other (no, I'm not exaggerating)... and we expect the same kind of behavior from the AI, no matter if it's in front or behind us. If we easily pass an AI, then we don't expect this same AI to pass us again. Of course it's a lot more complicated than that, but I'm sure you get the point. So again, a rubberband effect or any kind of adaptive difficulty within one race is obvious for anyone who is not a novice. You cannot hide it.

If you easily pass an AI in a game that is as deterministic as you suggest, then of course you would not expect the same AI to pass you again. However, in such a game I would expect you to play against AI that moved at roughly the same speed as you ... otherwise what's the point? Playing against an AI which has no hope of winning is just masturbation.

(Unless, of course, the AI are just there for atmosphere, in which case the problem of providing an appropriate difficulty does not even come up.)

This is very much a "worst case" scenario for dynamic difficulty, however ... 99% of games out there are not so deterministic.

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).

Quote Posted by Papy

Personally, I don't think it can be done. Human psychology is too complex.

Human psychology is (http://www.itu.dk/people/yannakakis/CIG09_IOI.pdf) not that complex, at least within the confines of a game.

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!