5 ms·

It sounds like this game was rushed, but it sounds like an interesting challenge: how do you design a game that takes a significant amount of time to play, whil

by zo7 9y ago

It sounds like this game was rushed, but it sounds like an interesting challenge: how do you design a game that takes a significant amount of time to play, while still ensuring that it would be balanced and engaging for all of the players throughout? There'd probably have to be some sort of computer simulation involved or some expedited way to play it, but how would you measure these things? Is it even possible?

eternauta3k 9y ago
Randomly sample game states, giving more weight to those that affect balance more strongly and/or occur more frequently.
- gwern 9y ago
  Random sampling wouldn't work because it is the essence of cheese/game-breaking to go into rare or weird edge-cases and parts of state-space. Imagine a broken fighting game where you could win by simply executing 'upper block' the entire game; random sampling will be exponentially unlikely to sample a game with all upper-blocks. But if you used a mechanism like self-play reinforcement learning to explore the game state-space, your agents might autonomously discover game-breaking strategies on their own.
hervan 9y ago
I'd never expect anything lasting 1500 hours to be engaging (or fun) the whole time.
mjaniczek 9y ago
QuickCheck sense tingling!
jerf 9y ago
Well, a certain degenerate case of "balance" can be obtained with simple symmetry. It isn't quite the totality of what people mean by the concept because there are plenty of abstract games in which once one competitor has an advantage, they can inevitably press it home to a victory, and most humans would probably call that "unbalanced", even though in some sense it isn't. But still, a symmetric game certainly starts you off with a solid start in that direction. (Technically, perfect play in a non-randomized game determines the winner very rapidly, and even in randomized games there's a clear maximum probability path. However, perfect play is generally humanly-infeasible and may even be computationally infeasible; chess is a relatively simple game compared to most and we still don't know who wins or draws with perfect play.) Engaging 10 people for 1500 hours is probably effectively impossible, though. At that point you'd basically be selecting the people and building the game to match, if a solution even existed. (You may be thinking "Oh, but jerf, there are people who play 1500 hours of Team Fortress 2 and stuff", but my reply is that if you designed TF2 for this use case and picked 10 random people, even 10 random TF2 players, you'd have to be pretty lucky to get 1500 hours for all 10 of them. I'm sure there's well more than 10 people with that much time clocked but they're still the outliers, and you'd have a hard time naming them before the fact. Even with prior information about their addictiveness and tendency to stick with games.)
- Cyph0n 9y ago
  Funny that you picked TF2 as an example. From my experience with the game over the course of several years, TF2 becomes insanely addictive only after you overcome the horrendous learning curve. This takes around 50-100 hours of pure playtime on average. Once you understand the game mechanics, how to play the classes reasonably well (fully mastering a class can take 100+ hours), how trading works, and how to play the most popular game modes (including duos, 4v4, dodgeball, bball, MGE, surf, etc.), you're basically set for life! For those who overcome the curve, the retention rate is quite high, from my experience at least. The release of Overwatch probably pulled quite a few people over though. This is primarily due to how Valve has treated the game over the years. I have over 1000 hours, my brother has twice that (at least), and I know several people I played with regularly who had 4000+. Keep in mind that the country that I was living in had an absurdly small TF2 community. I don't play anymore because: (1) I don't have a good PC, and (2) grad school sucks.
  - jerf 9y ago
    "I have over 1000 hours, my brother has twice that (at least), and I know several people I played with regularly who had 4000+." Oh, I picked that carefully. TF2, Starcraft, Diablo, the more successful MMOs... certainly people can pour thousands into those. First, I guarantee you that for every person you can find with 1000 hours in TF2, there are dozens or possibly even hundreds of people who installed it, screwed around for a few dozen hours or fewer, and then left. So if you randomly select 10 TF2 players, even if I spot you "from the set of all people with more than 5 hours in the game", the odds that you'll come up with a set of 10 people who are ready to put 1500 hours into "one game" are actually pretty dismal. Secondly, I'd also very, very seriously question whether that constitutes "one game" in the way that this game really is one game, with one team winning, after 1500 hours. TF2 is more a series of games that last about an hour. I've played far less than 1500 hours on TF2, and I "won" and "lost" dozens of times each. MMOs may lack quite the same degree of clarity but one could probably say that "one quest" is a game, or some other larger subdivision like "all the quests in one area" or "a single raid", but generally if an MMO does have a quest that lasts for even a hundred hours it's going to be considered by the players as a long-term adjunct quest rather than "the thing I do for 100 hours". There have been enough MMOs out there that somebody might be able to cite a counterexample, but it would be rare. At the very least an MMO player playing one MMO for 1500 hours is probably going to be running multiple characters, which would be reasonably called "separate games". There may be individual Civ games that ran that long, like maybe the Eternal War: https://www.reddit.com/r/theeternalwar/ https://www.reddit.com/r/theeternalwar/ (The original post the guy doesn't claim a game time, just that he'd been playing for 10 years.) But, again, trying to get 10 such people together would be a challenge.
usjehshksish 9y ago
You don't need to balance a game that takes 1500 hours, because no one will ever play it enough times to discover any imblances. And the total number of completed games will never reach the point that would open the game up to large scale analysis.
- PeterStuer 9y ago
  There are more ways to discover 'imbalance' then just stumbling across them in regular play. Many games have 'theory-crafting' communities digging into the numbers outside of actual play. I used to love reading through D&D rule books, even though I didn't play it. Now unless there were inherent time sinks that prevented zergs, if there were any serious imbalances, one side would roll over the other in far less than 1500 hrs.
Someone 9y ago
Reading https://boardgamegeek.com/thread/304828/its-details-baby https://boardgamegeek.com/thread/304828/its-details-baby, game play is fluid, so the game likely would get quite a bit faster if one used computers for doing calculations. For example, those fuel leakage calculations on 1000+ units, for 100+ turns involve carefully writing down 100,000+ numbers. At two seconds per number, that’s three days of work over the entire game, or, if that 1500 hours is accurate, around 4% of playing time (.4% if ten players shared the work load) A modern computer could easily do that in one second. I think someone (truly) dedicated could build a version where every piece had an e-ink display that gets updated automatically to show each unit’s food, fuel and ammunition status and strength, and speed up game play tremendously.

by zo7 9y ago

eternauta3k 9y ago

Randomly sample game states, giving more weight to those that affect balance more strongly and/or occur more frequently.

gwern 9y ago
Random sampling wouldn't work because it is the essence of cheese/game-breaking to go into rare or weird edge-cases and parts of state-space. Imagine a broken fighting game where you could win by simply executing 'upper block' the entire game; random sampling will be exponentially unlikely to sample a game with all upper-blocks. But if you used a mechanism like self-play reinforcement learning to explore the game state-space, your agents might autonomously discover game-breaking strategies on their own.

hervan 9y ago

I'd never expect anything lasting 1500 hours to be engaging (or fun) the whole time.

mjaniczek 9y ago

QuickCheck sense tingling!

jerf 9y ago

Well, a certain degenerate case of "balance" can be obtained with simple symmetry. It isn't quite the totality of what people mean by the concept because there are plenty of abstract games in which once one competitor has an advantage, they can inevitably press it home to a victory, and most humans would probably call that "unbalanced", even though in some sense it isn't. But still, a symmetric game certainly starts you off with a solid start in that direction. (Technically, perfect play in a non-randomized game determines the winner very rapidly, and even in randomized games there's a clear maximum probability path. However, perfect play is generally humanly-infeasible and may even be computationally infeasible; chess is a relatively simple game compared to most and we still don't know who wins or draws with perfect play.) Engaging 10 people for 1500 hours is probably effectively impossible, though. At that point you'd basically be selecting the people and building the game to match, if a solution even existed. (You may be thinking "Oh, but jerf, there are people who play 1500 hours of Team Fortress 2 and stuff", but my reply is that if you designed TF2 for this use case and picked 10 random people, even 10 random TF2 players, you'd have to be pretty lucky to get 1500 hours for all 10 of them. I'm sure there's well more than 10 people with that much time clocked but they're still the outliers, and you'd have a hard time naming them before the fact. Even with prior information about their addictiveness and tendency to stick with games.)

Cyph0n 9y ago
Funny that you picked TF2 as an example. From my experience with the game over the course of several years, TF2 becomes insanely addictive only after you overcome the horrendous learning curve. This takes around 50-100 hours of pure playtime on average. Once you understand the game mechanics, how to play the classes reasonably well (fully mastering a class can take 100+ hours), how trading works, and how to play the most popular game modes (including duos, 4v4, dodgeball, bball, MGE, surf, etc.), you're basically set for life! For those who overcome the curve, the retention rate is quite high, from my experience at least. The release of Overwatch probably pulled quite a few people over though. This is primarily due to how Valve has treated the game over the years. I have over 1000 hours, my brother has twice that (at least), and I know several people I played with regularly who had 4000+. Keep in mind that the country that I was living in had an absurdly small TF2 community. I don't play anymore because: (1) I don't have a good PC, and (2) grad school sucks.
- jerf 9y ago
  "I have over 1000 hours, my brother has twice that (at least), and I know several people I played with regularly who had 4000+." Oh, I picked that carefully. TF2, Starcraft, Diablo, the more successful MMOs... certainly people can pour thousands into those. First, I guarantee you that for every person you can find with 1000 hours in TF2, there are dozens or possibly even hundreds of people who installed it, screwed around for a few dozen hours or fewer, and then left. So if you randomly select 10 TF2 players, even if I spot you "from the set of all people with more than 5 hours in the game", the odds that you'll come up with a set of 10 people who are ready to put 1500 hours into "one game" are actually pretty dismal. Secondly, I'd also very, very seriously question whether that constitutes "one game" in the way that this game really is one game, with one team winning, after 1500 hours. TF2 is more a series of games that last about an hour. I've played far less than 1500 hours on TF2, and I "won" and "lost" dozens of times each. MMOs may lack quite the same degree of clarity but one could probably say that "one quest" is a game, or some other larger subdivision like "all the quests in one area" or "a single raid", but generally if an MMO does have a quest that lasts for even a hundred hours it's going to be considered by the players as a long-term adjunct quest rather than "the thing I do for 100 hours". There have been enough MMOs out there that somebody might be able to cite a counterexample, but it would be rare. At the very least an MMO player playing one MMO for 1500 hours is probably going to be running multiple characters, which would be reasonably called "separate games". There may be individual Civ games that ran that long, like maybe the Eternal War: https://www.reddit.com/r/theeternalwar/ https://www.reddit.com/r/theeternalwar/ (The original post the guy doesn't claim a game time, just that he'd been playing for 10 years.) But, again, trying to get 10 such people together would be a challenge.

usjehshksish 9y ago

You don't need to balance a game that takes 1500 hours, because no one will ever play it enough times to discover any imblances. And the total number of completed games will never reach the point that would open the game up to large scale analysis.

PeterStuer 9y ago
There are more ways to discover 'imbalance' then just stumbling across them in regular play. Many games have 'theory-crafting' communities digging into the numbers outside of actual play. I used to love reading through D&D rule books, even though I didn't play it. Now unless there were inherent time sinks that prevented zergs, if there were any serious imbalances, one side would roll over the other in far less than 1500 hrs.

Someone 9y ago

Reading https://boardgamegeek.com/thread/304828/its-details-baby https://boardgamegeek.com/thread/304828/its-details-baby, game play is fluid, so the game likely would get quite a bit faster if one used computers for doing calculations. For example, those fuel leakage calculations on 1000+ units, for 100+ turns involve carefully writing down 100,000+ numbers. At two seconds per number, that’s three days of work over the entire game, or, if that 1500 hours is accurate, around 4% of playing time (.4% if ten players shared the work load) A modern computer could easily do that in one second. I think someone (truly) dedicated could build a version where every piece had an e-ink display that gets updated automatically to show each unit’s food, fuel and ammunition status and strength, and speed up game play tremendously.