Tobold's Blog
Friday, October 31, 2014
Predictability of games

Azuriel is talking about Civilization: Beyond Earth and complains that after an interesting start the end game becomes a formality, where you already know you won, but still need to play for hours to actually win. Meanwhile Zubon talks about a board game where the better player always wins. And Stubborn mentions: "Okay, most recently I’ve been playing X-Com (I haven’t fully rage-quit just yet, though I was close the other night when two 90% to hit rocket attacks missed their mark followed by an 88% sniper shot missing, causing one of my people to be killed the following round, but I held it together and played on).".

What do these posts have in common? They are all about predictability of games. Games tend to start out in a state of maximum unpredictability: You usually don't know who is winning before the game has developed a bit further. At some point it becomes very clear who is winning, but unless a player concedes (and an AI player frequently isn't programmed to do so), the game goes on in a very predictable manner. And then it comes down to the amount of randomness in the game whether the game becomes totally boring, or there is still a chance for a reversal of fortunes.

Having said that, a lot of people like knowing early that they are going to win. Not everybody is playing games in a competitive manner. Most people are quite happy for example doing quests in MMORPGs for hundreds of hours, where they always "win", and only rarely encounter minor setbacks. Other players manipulate, cheat, or pay money in order to make a game more predictably a win. Even mild-mannered Stubborn can get close to rage-quitting if his 90% win chance turns into a loss.

That poses a challenge to game design. Do we really want a "No Longer Delay the Inevitable button" as suggested by Azuriel? Or do we want games where up to the very end it isn't predictable who is winning? (There are actually a number of board games with hidden scoring systems that work like that). Do we want more randomness in games, so they become less predictable, or do we prefer less randomness and more predictability?

I prefer games where you cannot lose. Where victory is inevitable, it's just about how long it takes to get there.

Consider a game like simcity or factorio. There's no real way to simply lose the game, at worse you can get set back a bit and need to redo things. Or even games as punishing as the Dark Souls series. There is no point in dark souls where the game says, 'You died to many times, so you lose. New game?' From the moment you start, the game is pushing you towards victory.

My experience is that there are very few games where losing is even an option. It's all just win fast or win slow. And I'm fine with that.
*too many times, gah
To Stubborns defense we have to say that he was not close to rage quitting because his 90% win ration turned into a loss but because the streak of bad luck with a chance of 1,2‰ (per mille) (0.1 * 0.1 * 0.12 = 0.0012 or 1,2‰) got his team member killed.
Which I must admit I can perfectly understand if lady luck screws you big time in a heavily tactical game, despite having done nothing wrong. I would quit playing that game for 24 hours to calm down.

On a sidenote back in Classic WOW but especially in Burning Crusade the upper example (a unusual long streak of undodged blows (sometimes in conjunction with parry haste) caused from time to time some funny raid wipes out of the blue for heavy avoidance tanks (which where not heal-able by their assigned healers)
We're all jaded on Civ games.

For a long time, only the early part of such games has caught my interest.

Sometimes the win condition of a game distracts from the adventure.
I do like having the Concede button in Hearthstone. Once you know you can't win, it is best to Concede and get into another game where maybe you can win.
I've experienced the same clean up stage in every resource based strategy game. It's just inevitable really. You are smarter than the computer. Once you achieve production parity it's a done deal. Once you achieve production superiority it's just stomping bugs.

A lot of games now offer various victory conditions that benchmark the point when the game is mostly over, which I like. But you can always walk away.
One missing possibility : the game is totally predictable (not random) but the decision of who will win is at the end. Or whatever combinaison of those three possibilites. (wining at the begining, wining at the end, or wining by luck).

Compare Trackmania with Mario Kart. In Mario Kart, the winner is not really the best player (the 1# during the whole course) due to the "blue shell attack", but a big part is chance. In trackmania, the only way to win is to not do error : you never know if you have beat the gold score until then end because you can always do one more error. But there is no luck. In both game, you do not know if you win before the end of the game, but one is due to random chance, the other flawless play.

Interestingly, I prefer TrackMania in solo, and Mario Kart in multi.
Well, this is where getting the difficulty level for yourself just right comes into play (and why games should offer multiple difficulty levels). The "randomness"/"unpredictability" there comes from how well one can play during any given attempt. By keying it just right, you shouldn't be able to predict it.
This situation is most obviously a problem with strategy games, just because they are so long. How bothersome it is tends to depend on how much more work needs to be done to grind out the win. I probably dropped more Civ 4 games than I finished just because the modern era was often incredibly bogged down with massive stacks of units attacking each other.

Overall I think it depends on environment.

In single player, AI catch up mechanisms are kind of irritating. The best solution is to improve recognition of insurmountable player leads, giving the player the option to declare victory at that point.

In party and social multiplayer environments, like the aforementioned Mario Kart, aggressive catch up seems to work well. You want everyone to have a chance to do decently.

In competitive multiplayer environments, perhaps mild catch up is often appropriate. You don't want people to resign every game halfway through, but getting a lead should mean something.
I really don't care much at all if I can win/beat a game. For me it is all about the fun of playing it. If the game isn't fun then I will certainly never win or beat it, unless it is too short for me to realize it isn't fun before it's done.
That actually reminds me of the WoT XVM win chance calculator discussions about people quitting the game if their predicted win-chance is lower than 50% ...making it even harder for the people who stay.

Thanks for the link love (;

I've long said of physical games (as in not digital) that I vastly prefer games that require skill over luck, but want both to be a factor.

Purely skill games will always be won by more experienced (and thus, theoretically, knowledgeable and skillful) players, leaving no reason for new players to try (think EVE here). As a result, new players are forced to endure a series of humiliating defeats before they can even really be said to be "competing."

Purely luck games are completely random and thus cannot really be taken seriously as no metagame can really exist, driving away players looking to become deeply immersed and skillful.

Games that use both offer the best opportunity to introduce new players while still rewarding those who put in more thought and time.

This is why I hate playing Yahtzee with my wife (who routinely scores more than one Yahtzee) and she hates playing St. Petersburg with me (as I routinely trounce her since I was playing it years before she was).

Our solution, of course, is cooperative games, so everyone can win together!

In Civ, I completely agree that often by about halfway through there's a foregone conclusion of victory (or not), and in those cases, I just start a new game. Very rarely do I play a Civ game to its conclusion; I would guess it's less than 10% of the time in single-player games. But in those games I feel the true challenge and meaning of the Civ series, making all the other "false starts" (though the falseness may be in how easy it is to win a particular game) worth the time invested (around 400 hours on Civ 5 now. What have I done with my life?).

At any rate, interesting link to Azuriel and addition of your own. Thanks!
It's not about predictability, it's about perceiving fairness. In XCOM, you sometimes invest in efficient group (both strategically - with upgrades and customization, and tactically - by taking good positions), and then watch all these investments don't pay off. The system would be better if it had some sort of "on miss" damage.

You wouldn't mind missing completely when hit chance is 50% (it's a coin flip after all and it's perceived as unpredictable), but higher percentage should probable scale damage and have 100% hit chance.
Sometimes I think games need a luck counter. Something that reminds you about all those 90% hits you've had and that a couple of misses could be just around the corner.

I would like to see more games with the option to keep things challenging as you approach the win situation. Something that makes the AIs more likely to cooperate each other, and less likely to support you as you get closer to the finish.
There's a great line in one of the Discworld novels (I think it's "Interesting Times"), where the villain is disappointed by the idiocy of the people trying to depose him, and he wishes for an enemy as brilliant as he is.

He then thinks again, and wishes for an enemy who *seems* to be as brilliant as he is, but with one subtle but critical flaw which means he will always be defeated.
Post a Comment

Links to this post:

Create a Link

<< Home
Newer›  ‹Older

  Powered by Blogger   Free Page Rank Tool