Are patterns happy accidents?

[tcv]

Ever since a child, I always liked games that didn't have many goals to them. I'd have more fun, for instance, playing the first level of Bubble Bobble over and over again than I would trying to get to Level Infinity. Why? Well, whenever I came to a challenging part, I'd become extremely angry and frustrated that I couldn't get through it.

An example: earlier this weekend, I tried the first Rainbow Islands (NES) and I became very frustrated after having encountered the first boss. I thought, "Why the hell can't they just have a bunch of levels?! Why do there have to be these big spots where you have to stop and figure stuff out?"

Well, it occurred to me that maybe I'm approaching games all wrong. Maybe the trick isn't the score anymore, maybe it's in trying to figure out what the designer did and how the game plays. Maybe that's the whole point of games.

But is it?

Did Namco purposely place patterns in Pac-Man so that people could figure out how to get so good at the game they could crash it?

Does every boss have a special pattern that needs to be figured out?

Or are all these things just happy accidents? Something the designer never really intended but didn't realize he was creating at the time.

[Sweater Fish Deluxe]

The patterns in Pac-man were definitely not programmed intentionally. They're just a bi-product of the particular way the individual ghosts behave.

However, by the time you get to, say, the shooters of the later '80s boss patterns are definitely intentional.

I'm not sure about Rainbow Islands, I never played it enough to have burned the boss battles into my brain. (New Rainbow Islands on the DS isn't very good, by the way.)

I think you can tell the difference between patterns that were intentionally orchestrated by the programmers and those that aren't though. With intentional ones, you generally don't have much choice but to follow the pattern if you want to actually beat the boss and all the hints are pushing you into the pattern. With Pac-man, on the other hand, there's countless other ways to complete the levels, they just happen to be less certain (mostly because they haven't been established as working patterns...in reality any path that leads you to all the pellets without getting killed, is a pattern).


...word is bondage...

[Emuaust]

well with my limited knowledge of programming I would have to say
that at this point in time it is really impossible to create software
that acts totaly random, even if random features are implemented
into a program they are implemented on a mathematical probability
of randomness, thus resulting in a pattern of some sorts.

Therefore even if unknowingly game makers implement patterns,
its the nature of the actual program.

[Sweater Fish Deluxe]

well with my limited knowledge of programming I would have to say
that at this point in time it is really impossible to create software
that acts totaly random, even if random features are implemented
into a program they are implemented on a mathematical probability
of randomness, thus resulting in a pattern of some sorts.

Therefore even if unknowingly game makers implement patterns,
its the nature of the actual program.

It's not hard to create an algorithm that comes close enough to total randomness to be good enough for a video game, though. No one would ever accuse a game like Tetris of having patterns. The randomness of the pieces may not be truly mathematically random, but it comes close enough.


...word is bondage...

[Emuaust]

alas close enough isnt exact and the fact is if there was a person
witht the time and mental capacity they would be able to
work a basic pattern out even for the likes of tetris, yes it seems
far fetched but is in deed doable.

[cyberfluxor]

The reason these games work the way they do is from algorithms, not randoms. They program each ghost with an AI that's set instead of ever changing. Newer games have the capabilities for additional processing and memory to change the AI due to player movements. What you're trying to discuss is pseudo-randomness, which is a bit different. Of course with pseudo code there is a statistic built behind it which may allow a collapse onto a few different predictable states.

If you'd like some cool information check out:
http://en.wikipedia.org/wiki/Game_theory

[rbudrick]

No one would ever accuse a game like Tetris of having patterns.

I would, because it does. It depends what version, but lemme give some examples. On the NWC cart, there's 4 distinct falling orders. Thor Aackerlund figured this out early on and used it to a hefty advantage.

In the original GB version I have had instances where I lost quickly and got the same order of blocks on the very next game. This leads me to believe there's a whole lot more than 4 patterns on that version, but there are patterns nonetheless. I'm not saying your actions don't change that pattern, but they are there and only vaguely appear random.

-Rob

[Sweater Fish Deluxe]

No one would ever accuse a game like Tetris of having patterns.

I would, because it does. It depends what version, but lemme give some examples. On the NWC cart, there's 4 distinct falling orders. Thor Aackerlund figured this out early on and used it to a hefty advantage.

In the original GB version I have had instances where I lost quickly and got the same order of blocks on the very next game. This leads me to believe there's a whole lot more than 4 patterns on that version, but there are patterns nonetheless. I'm not saying your actions don't change that pattern, but they are there and only vaguely appear random.

-Rob

I've never seen such a thing and I play Tetris religiously. My favorite version is Tengen's version, though, so maybe that has a better randomness generator, I don't know. I've played the Gameboy version quite a bit, but have never noticed the kind of patterns you mention. It's surprising to me that a professionally developed game would have such a poor random generator that someone just playing the game without even looking for patterns could spot them. Many homebrew games, including homebrew versions of Tetris, have defects like that. Homebrews often base their random generators on the vblank, so it's all based on how much time the player takes, for instance, on the title screen. The better homebrews have more complex ways of generating randomness, though, and I would have assumed most any commercial game would, too, especially in a game like Tetris where randomness is so important. I guess not, though.

In any case, what I said before is still true. It's possible and quite easy for the programmers of video games to include algorithms that come so close to randomthat any human being would be fooled. Whether this or that game doesn't have such high a level of randomness or whether it's not true mathematical randomness is beside the point. It comes close enough for a video game.


...word is bondage...

[ProgrammingAce]

Tetris may or may not be random, depending on how you look at it. When you create a 'random' number, you need what's called a seed number to generate it. The seed number doesn't have to be random, it just has to be different each time. Most random numbers are generated off of a clock, but since the gameboy doesn't have a clock, they needed something else. Tetris uses the number of rotations of the pices as the seed for the random number generator. If you rotate the same block pattern the same number of times, you'll eventually start to see patterns over time (the algorithim does have a bit of true randomness, so you won't be able to duplicate a complete game).

As for unexpected results from 'patterns', that's what's called emergence in the AI field. As an AI designer, you see it quite frequently. When you're writing an AI for a first person shooter, for example, when you add a bit of randomness to the possable outcomes of a scenario you will often find your AI doing unexpected things.

For example, an enemy sees the player and starts shooting. The enemy misses the player but hits another enemy. Since the AI is programmed to attack the last entity that damaged them, the two enemies start fighing giving the player a chance to sneak away. This may be something the programmer had not intended, but since it adds an extra dimension to the game, they'll leave it in.

Emergence is something you see quite often as a designer, but is usually something the player will never notice as they believe it was done intentionally.

[cyberfluxor]

ProgrammingAce is quite correct on the subject. I'm not familiar with what seed variables they used in Tetris and I wouldn't doubt rotations was the key they used. Older systems didn't have the internal clock (TIMER) to be used for random occurances in games and it took delicate processing to dedicate one to run in the background so algorithms are dictated due to movements the environment provides. Since you have the ability to change the state the environment is in you can almost force a more predictable probability that makes it appear the game is fixed. Lastely, if the complexity isn't high then it just makes it more comprehendable for the player to build skills to master the game easier.

[FantasiaWHT]

This all reminds me of the TV show Numb3rs. Especially the episode about the not-so-random random card shufflers used in Vegas casinos... a bunch of crooks blackmailed the designer of the algorithms into spilling his guts and they used that to consistently win at Blackjack

[ubikuberalles]

Many games use pseudo-random number generators and, if not implemented very well, can exhibit patterns that are easy to figure out. Games and computers in the 1970's and 1980's didn't have very good implementations of pseudo-random number generators. As mentioned already, that was mostly becaue the machines didn't have a real-time clock to generate a sufficiently random seed for the generator.

For example, Michael Larson was able to figure out the sequence of the big board on the Press Your Luck game show and was able to net over $100,000 in prize money (he did that by spinning 45 times without a whammy!).


@FantasiaWHT: That Numb3rs episode was probably based on real events (except for the blackmailing part, perhaps). I remember hearing about people exploiting the poor pseudo-random generators in early electronic slot machines. Ron Harris, an employee of the Nevada Gaming commision at the time, wrote a program that exploited the poor randomness of the pseudo-random number generator of an Casino Keno game. He got caught, however, and went to jail (he also put back door programs in the slot machines he inspected for the commission and was able to scam the casinos of thousands of dollars).

So, people finding patterns in games that are supposed to be totally random? I'm not surprised.

[theshizzle3000]

Yeah I think most games have a pattern to how one should kill a boss I mean you can look at a game like Shadow of Collossus which is supposed to be based around only boss battles. And due to programming they are only able to kill a boss a certain couple of ways. I cannot wait till the day when graphics and technology are so good that a boss will actually be able to be beaten by a person depending on what they want to do not on what a program says....and after that rant this will probably never happen I mean I would hope we would cure cancer or something before trying this...algright I quit.

[Sweater Fish Deluxe]

Yeah I think most games have a pattern to how one should kill a boss I mean you can look at a game like Shadow of Collossus which is supposed to be based around only boss battles. And due to programming they are only able to kill a boss a certain couple of ways. I cannot wait till the day when graphics and technology are so good that a boss will actually be able to be beaten by a person depending on what they want to do not on what a program says....and after that rant this will probably never happen I mean I would hope we would cure cancer or something before trying this...algright I quit.

I think it depends on the kind of game. I don't know what Shodow Of Colossus is, but when I think of patterns in games, shooters are the first thing that come to my mind and shooters with predefined and very complicated bullet and enemy patterns are really a lot more enjoyable than ones with random fire all over the place. Discovering the pattern for perfectly weaving in and out of among bullets and enemies is an art in itself in a good shooter. In fact, it's really the whole point of the game, actually--you know---shooting is secondary.

Which is not to say that there aren't certain games--many games, in fact--that totally missed the point of what make patterns enjoyable. The bosses that require you to perform a complicated set of maneuvers to open up defenses to the point where you can get in one shot and then have to repeat everything exactly the same ten to fifteen times before beating them bother me a lot. But that's just a poor implementation of patterns, it doesn't mean that the whole concept is no good. I think that in most games--not just shooters--a well conceived pattern is better than just random fire, especially in boss sequences.

I guess that what you're suggetsing is not random, but very complicated AI that goes beyond patterns (yes, yes, someone will want to jump in and point out that AI will always reduce to patterns at some level just like computer generated randomness will never be truly random, that's just pedantic bullshit, at some point we need to realize that we're just playing video games and things don't have to be perfect to work when you're dealing with a game). That might be better than just random fire, I guess. At some point that sort of AI may be possible on newer systems, but if it means that the idea of patterns gets abandoned, I'd see that as a great loss in video games. It wouldn't be the first time that improving technology led game designers to move away from something enjoyable towards something more realistic, though. I guess I've just never been the kind of person that thinks a better video game is one that comes closer to duplicating reality.


...word is bondage...