While I don't know if this'll refute what he says, I think this comes close:

John Conway's Game of Life

While it's not really a single player game (you set up a particular configuration and let it go from there, the player doesn't actually do anything after it starts), the possibilities are quite endless. It's actually quite amazing what you can do with a few simple rules. No points, no win/loss situation, you just create. It's like art. Heck, it is art.

talking_chicken

I'm more than willing to settle for more two player games with comparable depth to Go.

Demonbird

@Cyrian: Or any adventure game really.

Ashurahori

What he says kind of reminds of of the original Shadowgate. The first time you play it it takes forever and is super fun, but after you've got it all figured out it's just not worth playing again until five or 10 or however many years later after you've forgotten everything.

Cyrian

"randomness-free" is where he essentially ends the argument. As he goes on to explain w/o some form of randomness the actions that worked once will work again.

Allowing randomness you can get Tetris which is arguably nearly as complex as go, but if the pieces were the same every game it wouldn't have any depth what so ever.

wswordsmen

The problem is he's chosen, specifically, an impossible task: Go is only solveable through brute force.

In chess, there's a fairly straightforward system to assess who has a better position - you can, mathematically, tell who's winning. This is important, because you need to be able to evaluate possible moves, as a computer AI, and make an appropriate play.

There is no computer solution to Go. None. There's not even a good way to tell who's winning from one move to the next.

Does that mean that all single player games are totally broken? No. Chess is very solveable, and beatable only by the best players in the world. There are sophisticated mechanisms at work for Chess that recognize, in many cases, true strategy, and often don't use brute-force to approach solutions to the problem.

In short, he's built the whole thing on a false premise; taking as the foundation the one game no computer on earth can play against even a child and win.

If you set it up that way, sure, he proves his point. But that's an argument reductio ad absurdum.

gblock

The problem is that he is trying to find the depth of a multiplayer game (Go) in a singleplayer game. How will that work?

Tyheam

@stevesan: Irrelevant. Starcraft is horrible as a single player game, which basically proves this article's point.

wormguy

Starcraft >>>> Go

stevesan

Roguelikes are an example of single-player games that have nearly endless depth and replayability but rely on randomness to achieve that result.

Perhaps the reason it's so difficult to create inexhaustible depth in a single-player game with the criteria Rohrer outlines is that a multiple-player game will always involve outside elements, such as emotions, play styles and relative skill-levels. These elements, and the intrinsic complexity they give a game, can only be mimicked in a single-player game through randomization or AI.

concrete_d

Also, there's the other argument: You can talk to me about how intellectually superior Go is, but...

Where are the millions of Go players? Are there more of them than the number of people who played Halo? How about Uno for Xbox Live Arcade?

Go may be deep, but it's not accessible. If every game were Go, we wouldn't have millions of game players; you can call Go a casual game, sure, but only if you're willing to ignore the fact that it's one of the most challenging games left around, one which takes people a lifetime to master in many cases.

I don't want to take a lifetime to master Halo. Hell, I didn't spend more than a week playing it in the first place.

gblock

@gblock: I actually jusy finished higschool with a thesis about computer go and go in general.

I'm very sorry to say it's in italian, meaning no one here except me would comprehend it.

However, it's not about the "30 years of trying".

Our technology is not even close to what we would need, an average computer would have to compute over 1.5 years to answer to one move in a reasonably strong way.

The fact is, you're right. This guy uses Go to prove he's right, which isn't really the best way.

It would be like using GTA to prove all games are violent, or Megaman on the NES to prove that all games are hard.

It's not always true, Go is hands-down the most complex game ever invented. It's hard enough for humans to keep track of certain high level games, and the game is so open ended, it can only be solved by brute force.

Each "problem" that can be solved mathematically requires an algorithm of a certain complexity. In go's case, it's exponential complexity, meaning it is solvable, but almost impossible.

CiciColino

For a fairer treatise on go, try the Wikipedia entry for Computer Go, which explains the difficulties faced by modern programmers in attempting to build good computer AI for go, if you really care.

gblock

@talking_chicken: If you put winning conditions or a scoring system on the game, however, it quickly becomes an optimization problem based on the initial setup. i45hg is actually relatively similar to a cellular automaton.

To some extent he is captain obvious. In a game without another player, randomness, or physical challenge, the only question is what strategy to implement. Once the player has that knowledge, there's no challenge. Still, that doesn't mean finding that strategy is always easy.

Capitalist_Bagel

In fact, I find his whole premise fairly paper-thin. Single player games inherently pit you against the rules of the game; multiplayer games pit you against the other player via the rules. They are, in fact, fundamentally different from one another, which is why they appear to be fundamentally different.

When pitting player versus player via rules, why add randomness? When pitting the player versus any static set of rules, why not do so? What does a single-player puzzle look like? How can that single player puzzle ever, in any straightforward way, be reshaped to easily pit two players versus each other via a set of rules instead of just being a time-based exercise?

They're different, at their core. The fundamental change is rules-as-opponent.

Other types of single-player game, including Starcraft, are based on simulated multiplayer - you versus computer via the set of rules. And depending on how those rules are set up, they can be achieved within varying degrees of success. A game like chess - with a straightforward evaulation mechanism to determine a 'good' versus 'bad' move, and with a limited playable space and clear rules of interaction, is SUPER EFFECTIVE (sorry, had to do that) when played by an AI opponent.

Now take a game like Go, where the computer can't evaluate as well as an 8 year old, where only a complete novice will ever be beaten by the computer.

That's a game whose rules were designed to be playable by a human, and build any of those games in a multiplayer environment and you'll find that the computer is a poor AI opponent; you need to ensure that the rules end up more like chess - possible to evaluate within a reasonable amount of computational effort.

Efforts like those give us AI pathing structures, hitboxes around our characters, and a host of other game optimisations whose purpose is to give the AI a fighting chance at being able to get from point A to point B, let alone actually win.

These are early days for computing; we're still getting our computational and algorithmic footing, still trying to figure out how to make some of our hardest problems solvable within reasonable amounts of time and computing power, and we're getting better at it - we're not just building faster machines, we're building better algorithms.

So I don't agree. At all. On any level.

Because quite frankly, there aren't many other human games like Go. And that's been around for thousands of years. Games with Go's magical depth and power to hold the minds of thousands, tens, maybe even hundreds of thousands don't even come once a generation; maybe once every few hundred years at best, if you take the big greats and line them all up - Mahjongg, Go, Hearts, etc.

And even now, only Go is unsolvable, unplayable by machine. And that's in what, 30 years of trying?

Give us time. I'm only asking for another 30 years, not 3000.

gblock

@gblock: Uhm, there are Go computers and you can make solutions for Go. It's just way harder.

The real Reductio ad absurdum is that he takes Go and says:"Make a game with the same depth but no AI and no more than one player." Now this is the thing that makes so deep. If there is no one to play against you already won before the game started thus make the game not deep at all.

Tyheam

taking as the foundation the one game no computer on earth can play against even a child and win.

That's a bit of an exaggeration unless the child in question is at least a somewhat habitual player of go (which, let's be honest, is less than likely if by child you mean elementary-school-aged. seriously how many 12 year olds do you know; now how many of them play go?). It's true that no computer can play go as well as a computer can play chess, but that's not to say that a computer couldn't trounce a complete novice at the game.

juv3nal

@CiciColino: Actually, I'd love to read your thesis. :) I'll leave you a private note via your user page with my email address; I'm always interested in new papers on Go.

gblock

So he was trying to see if you could make a deep single player without all the things you would need to make one?

That's like asking someone to make a deli style sandwich, except without their hands.

Thorax

I disagree strongly with his ideas and points, though I understand his meaning. It's just a flawed way to discuss or think about in all regards.

That said, I'm glad he has referenced Go. A wonderful game, an in-depth game, and the kind that only comes along once every thousand years or so.
Sigh, if only Go was more popular in the West.

What was that one saying? The possibility of different Go games is higher then that then the estimated number of atoms in the universe?

Kenofthedead

@gblock: I think if you are to head into China, Korea and Japan Go is more commonly played than chess. There are a lot more professional Go players in the world than there are professional video gamers. Heck, they have an organisation. Maybe you WOULD want to take a lifetime to master Halo if it won you such acclaim.

In regards to the article, it's interesting, but his point feels a bit lost. Turning to Go, intrinsically a two-player game, for lessons in single-player game design? He seemed to be on the right path with solitaire and Tetris, but in the end, it felt like a lot of packaging to promote i45hg. Which in retrospect, I guess it was.

ParanoidIndividual

Another point that compounds the limitations of Rohrer's statement is that there are external ways of injecting depth into single player games that he does not account for. Some examples of these are:

1) Using games to run commentary or otherwise create a form of entertainment (Let's Play on SA, Machinima, and gaming marathons for charity)

2) User imposed challenges (namely, in Nethack, which features both in-game voluntary challenges, such as dietary restrictions, as well as newer and more cruel challenges, such as beating the entire game blindfolded)

3) User created modification of games.

I suppose it all comes down to your definition of a "game". Nowadays, they don't seem to be finished upon publication. In this, I think there are few games you could not examine as "human games".

unfathomablej

I play go (semi)competivily...
I would love a game designed around the core basics of GO...

Off topic...
But GO for the DS would be amazing!

Aetherium

@wswordsmen: I think Solitaire would be a better example, since there is a definite and definable end-goal instead of just a mad march towards higher scores.

In any case, I'm glad that the author realized that he can't, in all actuality, square a goddamn circle. I would think that it'd be fairly obvious that nearly all games require a certain degree of randomness to be entertaining, whether it's from design, the AI, or a human opponent.

And before anyone else brings it up: I'm hesitant to think of adventure games as 'games' in the traditional sense. I tend to believe they occupy some strange space between films and storybooks, since actual interaction tends to be minimal, at best.

JustThisGuy

ummm, everyone who seems to be implying that few people play go has clearly never been to anywhere in southeast asia...

in terms of sheer worldwide numbers of players, the go playing population EASILY outstrips the number of WoW players, much less halo players

@gblock
actually i'd say go is possibly the MOST accessible game ever invented... there are few rules, which is NOT to say go is easy to master, just to play

zerobus

I like Go, I hope they add a Go game to XBLA soon.

Isn't it true that even the best computer programs can only play Go at the equivalent of human's intermediate skill level?

XbhaskarX

It's not about multiplayer, it's about randomness.

Go and Chess and not very deep games if the person you are playing plays exactly the same every time.

So basically his whole point is: if something is the same every time, it's the same every time.

Maggie just posts anything that seems like it is an intelligent discussion of games, even if it is actually just simple and weak, but with some faux-intellectual words like "completionism".

I mean FFS "completionism"??
For future reference, when trying to sound intelligent, *don't make up words*.

Chriky

@gblock: "When pitting player versus player via rules, why add randomness? When pitting the player versus any static set of rules, why not do so?"

I disagree. Look at a game like Company of Heroes -- there are elements of randomness present in CoH when you order your troops to move and attack that do not exist in Starcraft, and because of it different strategies evolve to account for it. Your bazooka shell might, by chance, not blow up a tank. Compensate by planning for failure and having another bazooka team ready, or pull the troops back so they can reload in safety.

A single player game is not really a player versus the rules; the rules of a game encapsulate everything and so 'the rules' cannot be an entity to be opposed. 'The rules' is another way of saying 'game'. Of course you're fighting the game you're playing, that's what constitutes a game -- a player therefore cannot be an entity either. In a multiplayer game you are not fighting another player through the rules, you are following rules that now include a live player. Having a game be singleplayer or multiplayer does not explicitly create a dichotomy; for instance you could create rules for a multiplayer puzzle game -- Puzzle Fighter is a prime example.

Snake726

@Kenofthedead: Precise

The number of possible go games exceeds (by far) the number of estimated atoms in the entire known universe.

That's why a proverb says "A game of go was never played twice"

CiciColino

Interesting concept.

I have to agree with previous posters that randomness is essential for single player competition.

If we through competition out the window, we could still loosly fit a game like SimCity to his criteria. I would never play a game like that though; Competition is essential to hold my interest, whether artificial or biological.

ComradeLewis

@Aetherium: "But GO for the DS would be amazing!"

[www.play-asia.com]

I've got this one. It has no region coding so it should work anywhere (works on my American DS), and it's not too hard to figure out what all the functions do, it just requires a little experimentation. I can't speak as to the strength of the A.I., as I'm still a beginner, but the description claims it uses one of the strongest engines out there to date (that doesn't require a supercomputer to use).

cableshaft

Very interesting. I would like to point out that he didn't mention when his game ends, but I'm assuming it's when the board has filled up completely or you can no longer place 9 stones in a turn. Also, I understand his argument on searching for depth in a single-player game, but he didn't mention that he was excluding "fun" for a replay option. If a person doesn't enjoy a game, single or multiplayer, they won't play it again unless certain circumstances bring cause to play again.

And I'd like to state that I believe humans are the peak of randomization. Therefore, every game involving a human has randomization. The counter-point to this argument is that unless the human playing ignores all rules and does not desire any particular outcome, a human is never random.

Ryodestined