Wednesday, December 14, 2005
Fair Game or Drama?
As has been pointed out in the comments of some of the Game Moments blogs, human beings crave drama. A one-sided sports event is BORING. A story where the hero is always successful and makes the best decisions is TEDIOUS. While it pains us to see Indiana Jones take it on the chin, and fail in his attempts to get the Ark of the Covenant back, that just makes it all the more thrilling when Jones, beaten and battered, keeps coming back for one more try.
The same is true in videogames. The common thread in several of my "Game Moments" articles has been the same thing - the drama of coming back from a setback or going up against seemingly impossible odds - and SUCCEEDING. While I have a few fond memories of having my butt handed to me in amusing ways, it's the "amusing ways" part that makes them memorable. Whether I succeeded in spite of being confined to a tiny corner of the galaxy in Master of Orion, or slogging my way out of a deadly dungeon after encountering a particularly nasty random encounter, or going up against a foe that has nukes and stealth fighters when I'm still stuck with horse-drawn cannons and muskets, what made me enjoy myself was the dramatic struggle.
So here's the question as a game designer. Do you give the player a fair game, or a dramatic one? Obviously, players are far more thrilled by your game when it's dramatic. You want player's to love your game, right?
Some games "cheat" to provide this struggle. Several people from SingleTrac enjoyed playing Daytona USA in the arcade. Daytona let you play against another player, which we did often. We noted with amusement that the player in second place - if he was distant enough - could catch up to the leader even when his speedometer read a slower speed. The game "cheated" to keep the race dramatic. Once we figured it out, though, the game became less exciting to play for us. Kinda like watching professional wrestling (which I had to do at ANOTHER game company a few years later, when I was working on ECW2) - while they do their best to make it dramatic and the outcome is unknown to the audience, when you know it's all staged you have a tougher time staying interested.
Part of what made me less interested in Civilization II was this strong suspicion that the AI was cheating - their cities were staying at a parity with mine regardless of the circumstances. When I'd capture them, there'd be hardly any improvements remaining. Granted, maybe I could have destroyed almost everything but a granary and a church on my way in, but it still didn't seem quite right.
So the answer isn't so easy. You want to provide dramatic action, a close fight no matter what. But you MUST maintain the illusion of a fair fight as well, or you run the risk of your players throwing down their controllers in disgust of your cheap tricks. We are providing both entertainment AND a challenge to the player, and so we must succeed at both levels.
The Solution
The trick - like so much else in game development - is being able to strike a delicate balance. While I grew annoyed at Civilization's "cheating", the success of the franchise means it really wasn't all that bad, or it was just too well hidden. Since I STILL love the game, I am guessing it's more of the former.
At one point, during the development of the first Twisted Metal for the Playstation, we had a really bad problem. Twisted Metal SUCKED! While it was possible for some of us to beat it through very clever fighting (which seemed like good, challenging gameplay for a while), the problem was that all of the opposing cars would converge on the player and just CLOBBER him. Again - you could play carefully to avoid it, but once you were stuck and surrounded, the game ceased to be fun.
We hit on a solution that, to the best of my knowlege, was still being used on Twisted Metal Black for the PS2 (which was made by many of the same guys). I don't remember if Dave Jaffe suggested it, or if we came up with the solution ourselves, but we decided that at any time only TWO cars would be "engaged" with the player at a time. Any car not "engaged" would just putter around in kind of a stand-by pattern on some other part of the map. Who was considered "engaged" could change - if the player was closer to (or especially if he attacked) another car, it would quickly become one of the "engaged" vehicles. Non-engaged cars would still take an attack of opportunity if they got the chance. But really, at any point in the game, the player was really only fighting two cars at a time.
As far as I know, nobody ever noticed. And even if you did know this, the knowledge didn't sufficiently cheapen the experience for most players. That was how we managed to strike a balance between Drama and the Fair Game. Considering the success of the first game and the franchise, I'd say we succeeded.
In Void War, I had a problem in that the AI was too accurate. Because of the Newtonian physics of space, the player couldn't alter his flight path quickly enough to "dodge bullets." The solution was to make the AI "miss believably". It became readily apparent close-up that the AI was deliberately aiming away from the player sometimes, and was sometimes forgetting to pull the trigger when the player was right there in front of it. That was my attempt to balance the game and provide a combination of Drama and the Fair Game.
The first two Wing Commander games had another less balance-affecting trick. When the Kilrathi ships were within the player's view, they'd perform what they called "Flight Choreography." They'd zip around perfoming all kinds of aerobatic maneuvers that were cinematically exciting, but actually not very effective. When they were out of the player's view frustrum, however, they pretty much flew straight and boring to wherever they wanted to go with maximum efficiency. But as far as the player was concerned, they were corkscrewing and cartwheeling all over the sky all the time. It was a minor "cheat," but it worked for hundreds of thousands of fans.
Don't slave yourself to "simulating" the game environment, particularly in a single-player environment. But be careful that you don't forget that above all else, a game must be fun and should entertain and thrill your player. Stray too far in either direction, and you risk losing your audience.
Labels: Game Design
- posted by The Rampant Coyote @ 10:36 PM (Permalink)
Comments:
Links to this post:
<< Home
Re: Twisted Metal.. In addition to what you said, I also recall (maybe heard from Jaffe?) that the A.I. cars weren't even aware of each other.. that if they "engaged", it was just random chance, mostly due to them occasionally being on the same path. We found that pretty interesting and amazing since we were at a point of figuring out how capable our own A.I. should be.
- Jim Buck
- Jim Buck
# posted by 
: December 15, 2005 1:21 AM
: December 15, 2005 1:21 AM
Interesting info, Jay, thanks for sharing. I've told you before that I think Twisted Metal was one of the best games made. This balancing idea is one of the pieces that worked so well. I never noticed the two-car engagement, nor heard or read about anyone else noticing it.
Twisted Metal's PVP didn't seem nearly as balanced. If your opponent took Specter on any map that wasn't wide open, you were doomed. Were any on-the-fly changes made to the car stats in PVP to better balance the matchup?
Twisted Metal's PVP didn't seem nearly as balanced. If your opponent took Specter on any map that wasn't wide open, you were doomed. Were any on-the-fly changes made to the car stats in PVP to better balance the matchup?
Yep, that's true. The AI generally ignored each other. I vaguely recall putting in some code (I wrote the weapons and weapon-firing logic) that would have them take pot-shots at each other IF they happened to be in each other's sites, but I don't remember if I left it in or not. But they didn't seek each other out.
We might have even made them immune to each other's attacks (or take reduced damage) - we experimented with a lot of things, but the bottom line was that the game was not fun when the AI attacked each other. If you just hid while they killed each other - what was the fun in that?
Another level of cheating that DID bug me a little bit was on the weaponry. For a while, I kept track of what weapons were being used by the AI cars, and I was working with Steve Poulson (the guy handling the driving physics & driving AI) to have them gather weapons the same as the player. Then I was told to not bother with that, to just give the AI cars a random chance of using particular weapons, customized per car, so that some cars favored certain weapons. For example, we gave Darkside a higher chance of using a Freeze Missile so he could have a better chance of using his ramming attack.
I think this was the right decision, as the cost of having an unfair fight against the AI with infinite weapons was less than having the AI gather all the weapons themselves so you couldn't ever acquire any yourself. Or having the AI turn tail and run because its out of missiles.
Maybe there could have been a happy medium in there somewhere that didn't offend my "simulation" sensibilities quite so much, but it was the best decision at the time for the game we were making. I still had my head in the "Car Wars" mentality (from the old Steve Jackson game), whereas the designers (Dave Jaffe, Michael Jackson, Kellan Hatch, Scott Campbell, and Mike Giam) were thinking more of a "Street Fighter With Cars."
It's hard to argue with success. They were right.
We might have even made them immune to each other's attacks (or take reduced damage) - we experimented with a lot of things, but the bottom line was that the game was not fun when the AI attacked each other. If you just hid while they killed each other - what was the fun in that?
Another level of cheating that DID bug me a little bit was on the weaponry. For a while, I kept track of what weapons were being used by the AI cars, and I was working with Steve Poulson (the guy handling the driving physics & driving AI) to have them gather weapons the same as the player. Then I was told to not bother with that, to just give the AI cars a random chance of using particular weapons, customized per car, so that some cars favored certain weapons. For example, we gave Darkside a higher chance of using a Freeze Missile so he could have a better chance of using his ramming attack.
I think this was the right decision, as the cost of having an unfair fight against the AI with infinite weapons was less than having the AI gather all the weapons themselves so you couldn't ever acquire any yourself. Or having the AI turn tail and run because its out of missiles.
Maybe there could have been a happy medium in there somewhere that didn't offend my "simulation" sensibilities quite so much, but it was the best decision at the time for the game we were making. I still had my head in the "Car Wars" mentality (from the old Steve Jackson game), whereas the designers (Dave Jaffe, Michael Jackson, Kellan Hatch, Scott Campbell, and Mike Giam) were thinking more of a "Street Fighter With Cars."
It's hard to argue with success. They were right.
Tom:
Yeah, we worked pretty hard trying to balance the game for both PVP and single-player. We had some in-game tweakers throughout all of development that could change values of each car and weapon system on-the-fly. This allowed everyone to play with the values, and submit suggestions for how to tighten things up balance-wise.
Balancing things better in PvP typicall changed the AI's lethality in single-player, but we always decided that PvP came first, and that we would deal with adjusting the AI according to the new balance levels later.
One of the problems we had was that unlike a pure fighting game, the different levels made a huge difference in how the game played. Specter had an advantage in the city levels, since he could shoot through corners with his ghost missile and had the speed to get away quickly (and his missile shot so wildly that he could shoot at someone behind him with a non-trivial chance of hitting).
But in the arena, Specter was probably the weakest vehicle in the game. Darkside would COMPLETELY own him in the hands of another player anywhere near the same skill level. (Darkside pretty much owned the arena, period).
On the freeway, the cars with the most speed had a definite advantage - and once again, Specter was a weak choice, but a lot better off than the arena where speed was of little value.
The problem you run into is that if you try and balance all vehicles for all arenas, you risk losing the distinctiveness of the vehicles. So it's a difficult balancing act. On top of that, you also have to balance the vehicles against each other. The distinctiveness of the vehicles again caused a problem, because certain cars played to the weaknesses of other vehicles. So one car might be too powered compared to one car, and too weak compared to another.
I preferred Specter, personally, as did a couple of testers. In our balancing, people came up with good strategies to counter him. The one that I thought was potentially too powerful in the final release was Thumper, but it took a really good player to make Thumper reach his potential. But in the right situation, he could kill vehicles faster than any other.
If you are asking about changes to stats of cars that occured dynamically (Like decreasing the effectiveness of Darkside's ramming attack in the Arena to make him more balanced in THAT level) - no. I don't think it even occured to us. That's a clever idea though!
Yeah, we worked pretty hard trying to balance the game for both PVP and single-player. We had some in-game tweakers throughout all of development that could change values of each car and weapon system on-the-fly. This allowed everyone to play with the values, and submit suggestions for how to tighten things up balance-wise.
Balancing things better in PvP typicall changed the AI's lethality in single-player, but we always decided that PvP came first, and that we would deal with adjusting the AI according to the new balance levels later.
One of the problems we had was that unlike a pure fighting game, the different levels made a huge difference in how the game played. Specter had an advantage in the city levels, since he could shoot through corners with his ghost missile and had the speed to get away quickly (and his missile shot so wildly that he could shoot at someone behind him with a non-trivial chance of hitting).
But in the arena, Specter was probably the weakest vehicle in the game. Darkside would COMPLETELY own him in the hands of another player anywhere near the same skill level. (Darkside pretty much owned the arena, period).
On the freeway, the cars with the most speed had a definite advantage - and once again, Specter was a weak choice, but a lot better off than the arena where speed was of little value.
The problem you run into is that if you try and balance all vehicles for all arenas, you risk losing the distinctiveness of the vehicles. So it's a difficult balancing act. On top of that, you also have to balance the vehicles against each other. The distinctiveness of the vehicles again caused a problem, because certain cars played to the weaknesses of other vehicles. So one car might be too powered compared to one car, and too weak compared to another.
I preferred Specter, personally, as did a couple of testers. In our balancing, people came up with good strategies to counter him. The one that I thought was potentially too powerful in the final release was Thumper, but it took a really good player to make Thumper reach his potential. But in the right situation, he could kill vehicles faster than any other.
If you are asking about changes to stats of cars that occured dynamically (Like decreasing the effectiveness of Darkside's ramming attack in the Arena to make him more balanced in THAT level) - no. I don't think it even occured to us. That's a clever idea though!
Thanks for the extra info. I did feel that PvP was well balanced across all arenas -- I hope it didn't sound like I was complaining!
Yes, that's what I meant, tweaking weapon damage and/or ac based on what cars and arena were chosen.
On your original subject, I know Twisted Metal's PvP provided the elements for drama and let the players write their own:
Hiding Warthog behind a rooftop ramp pointed at a weapon pickup, then turbo-ramming Mr. Grimm off the roof when he went to delicately pick them from the edge. He'd hit the ground plane before you did and you'd win.
Erasing a parked white car on a warehouse district street, parking Specter in its place, and waiting for your friend to just drive by into your sights.
And the most dramatic -- when down to a speck of health with your friend coming up the ramp behind you, knowing which corner to approach the rooftop glass pyramid from so that when you blow it you drop right into the health cage.
The people I played with still tell and laugh at these stories 10 years later.
Yes, that's what I meant, tweaking weapon damage and/or ac based on what cars and arena were chosen.
On your original subject, I know Twisted Metal's PvP provided the elements for drama and let the players write their own:
Hiding Warthog behind a rooftop ramp pointed at a weapon pickup, then turbo-ramming Mr. Grimm off the roof when he went to delicately pick them from the edge. He'd hit the ground plane before you did and you'd win.
Erasing a parked white car on a warehouse district street, parking Specter in its place, and waiting for your friend to just drive by into your sights.
And the most dramatic -- when down to a speck of health with your friend coming up the ramp behind you, knowing which corner to approach the rooftop glass pyramid from so that when you blow it you drop right into the health cage.
The people I played with still tell and laugh at these stories 10 years later.
# posted by : December 15, 2005 1:55 PM
: December 15, 2005 1:55 PM
Heh - so do I.
That was why we suspected Twisted Metal would be a hit (and Jet Moto, too, though that was more of a sleeper hit). We were still playing it and having a blast playing it in the final stages of development. When you are really SICK of the game in general, but you are still having a blast playing it.
Of course, multiplayer helps a lot. We had a lot of great Twisted Metal battles the last couple of weeks of development.
Post a Comment
That was why we suspected Twisted Metal would be a hit (and Jet Moto, too, though that was more of a sleeper hit). We were still playing it and having a blast playing it in the final stages of development. When you are really SICK of the game in general, but you are still having a blast playing it.
Of course, multiplayer helps a lot. We had a lot of great Twisted Metal battles the last couple of weeks of development.
Links to this post:
<< Home
