"Fairness is the unspoken promise of most video games. "
February 5, 2017 10:14 PM Subscribe
This bothers me a lot. People really need to get a better understanding of risk.
Why would this bother you? Video games aren't simulations of reality (except for those that explicitly are), they're supposed to be fun experiences. Designers make unrealistic choices in games to improve player experience all the time.
Did you know that stomping on a turtle doesn't yield a weapon that can be kicked to side along the ground forever? Or that very few ninjas can shoot fireballs and run up walls?
posted by Sangermaine at 10:44 PM on February 5, 2017 [13 favorites]
Why would this bother you? Video games aren't simulations of reality (except for those that explicitly are), they're supposed to be fun experiences. Designers make unrealistic choices in games to improve player experience all the time.
Did you know that stomping on a turtle doesn't yield a weapon that can be kicked to side along the ground forever? Or that very few ninjas can shoot fireballs and run up walls?
posted by Sangermaine at 10:44 PM on February 5, 2017 [13 favorites]
I am very skeptical about Olaf's magical last throw.
posted by corb at 10:47 PM on February 5, 2017 [4 favorites]
posted by corb at 10:47 PM on February 5, 2017 [4 favorites]
No-one in the world can be persuaded that weapons appear then you stand on turtles, and it would not matter anyway if they could.
Many people in the world fail to understand risk, and this is hugely important for the decisions we make personally and nationally, from global warming to policing.
A (made-up statement) like "Immigrants make an economy stronger, but Meier discovered that people found this unbelievable, so he changed the game to match players' cognitive biases: immigrants now ate up resources and racially-pure civilisations had lower crime rates" would bother me too.
Also, and maybe more to the point, I feel cheated: the game lied to me. It's not a 33% chance of winning. That's also "unfair".
posted by alasdair at 11:17 PM on February 5, 2017 [36 favorites]
Many people in the world fail to understand risk, and this is hugely important for the decisions we make personally and nationally, from global warming to policing.
A (made-up statement) like "Immigrants make an economy stronger, but Meier discovered that people found this unbelievable, so he changed the game to match players' cognitive biases: immigrants now ate up resources and racially-pure civilisations had lower crime rates" would bother me too.
Also, and maybe more to the point, I feel cheated: the game lied to me. It's not a 33% chance of winning. That's also "unfair".
posted by alasdair at 11:17 PM on February 5, 2017 [36 favorites]
It's just that these biases toward distorted estimation of probability are so difficult to overcome, and so material to forming public policy and personal decisions, that having yet another thing reinforcing said biases is hard.
posted by amtho at 11:21 PM on February 5, 2017 [9 favorites]
posted by amtho at 11:21 PM on February 5, 2017 [9 favorites]
This sort of thing is kind of why I've sort of stopped caring about pretty much any game that doesn't involve multiplayer with a level playing field. I want to play a game that simply does not care whether I win or lose. Granted, I understand that that's not exactly a popular subcategory of video games overall compared with single-player stuff that's designed to eventually be beaten by the player, but.
posted by DoctorFedora at 11:55 PM on February 5, 2017 [4 favorites]
posted by DoctorFedora at 11:55 PM on February 5, 2017 [4 favorites]
Humans didn't learn to be bad at evaluating risk from playing Civ III, we are bad at it because the kind of risk our evolution optimized us for is fundamentally different than what we encounter today.
Blaming video games is... misguided.
Also the real problem is that the game in question bases its victory decisions on probability at all rather than something more deterministic. Like when I attack the AI's spearmen with my tank unit... and lose. It is perfectly natural for me to yell at my cheating ass computer, even if this only happens 1 in 10 times (as dictated, "correctly", by the odds).
posted by danny the boy at 12:18 AM on February 6, 2017 [3 favorites]
Blaming video games is... misguided.
Also the real problem is that the game in question bases its victory decisions on probability at all rather than something more deterministic. Like when I attack the AI's spearmen with my tank unit... and lose. It is perfectly natural for me to yell at my cheating ass computer, even if this only happens 1 in 10 times (as dictated, "correctly", by the odds).
posted by danny the boy at 12:18 AM on February 6, 2017 [3 favorites]
I realized a while ago that for all the complaints about stupid AI the priority in most games is to make the AI seem "real," not come up with the best strategy. An AI that did "gamey" things to win would be really annoying; better to make the AI act like a human, even a stupid human, then compensate with bonuses if it's not competetive.
For the life of me I can't remember the "replay the same battle for increasing cost" mechanic in Civ. The only replay was save and reload I thought?
posted by mark k at 12:21 AM on February 6, 2017 [2 favorites]
For the life of me I can't remember the "replay the same battle for increasing cost" mechanic in Civ. The only replay was save and reload I thought?
posted by mark k at 12:21 AM on February 6, 2017 [2 favorites]
Skewing probabilities to satisfy player perception has been a series staple in Fire Emblem since the Game Boy Advance games, and the fandom calls it "True Hit". Suppose the game says you have an 80% chance to hit an enemy. You might imagine that the game would draw an integer from 0 to 99, and if the number is less than 80, you hit, and otherwise you miss. Actually, it draws two integers from 0 to 99, and if the average is less than 80, you hit, and otherwise you miss. There are 780 out of 10,000 ways to miss, which means your true probability of hitting is 92.2%. It cuts both ways; if the game says you have a 20% chance to hit, your real probability of hitting is 8.2%. The effect of the average is to reduce the variance by making very high and very low numbers much less likely. In a series like Fire Emblem where a few bad rolls of the dice mean one of your characters dies and you're restarting the level, that reduced variability is a rare mercy.
posted by J.K. Seazer at 1:02 AM on February 6, 2017 [24 favorites]
posted by J.K. Seazer at 1:02 AM on February 6, 2017 [24 favorites]
Normal mode is easy, difficult mode is normal, and hardcore mode is difficult, and I wouldn't want it any other way. I'm not looking to prove anything. I just want to have fun.
Example, my favorite video game is Borderlands 2. For those who keep up with such things, one of the game's expansion classes had a very powerful ability, which was specifically included for players who's reaction time isn't very good; any bullets fired at a target still had a chance to ricochet and hit the target even if their shot missed. This made the game much easier/satisfying for people who aren't very good at first person shooters, which is why it's my favorite class by a mile. I wish other game studios would do this. Not everyone is good at twitch games, but we still want to play, and we have the money, so make a place for us at the table even if the purist howl.
posted by Beholder at 1:09 AM on February 6, 2017 [12 favorites]
Example, my favorite video game is Borderlands 2. For those who keep up with such things, one of the game's expansion classes had a very powerful ability, which was specifically included for players who's reaction time isn't very good; any bullets fired at a target still had a chance to ricochet and hit the target even if their shot missed. This made the game much easier/satisfying for people who aren't very good at first person shooters, which is why it's my favorite class by a mile. I wish other game studios would do this. Not everyone is good at twitch games, but we still want to play, and we have the money, so make a place for us at the table even if the purist howl.
posted by Beholder at 1:09 AM on February 6, 2017 [12 favorites]
This sort of thing is kind of why I've sort of stopped caring about pretty much any game that doesn't involve multiplayer with a level playing field. I want to play a game that simply does not care whether I win or lose. Granted, I understand that that's not exactly a popular subcategory of video games overall compared with single-player stuff that's designed to eventually be beaten by the player, but.
Even this is sometimes rubber banded/handicapped to keep it close. The common example of this is Mario Kart, but there's quite a few games where the devs have obviously done something that weights the probabilities on random drops etc.
posted by jaduncan at 1:32 AM on February 6, 2017 [2 favorites]
Even this is sometimes rubber banded/handicapped to keep it close. The common example of this is Mario Kart, but there's quite a few games where the devs have obviously done something that weights the probabilities on random drops etc.
posted by jaduncan at 1:32 AM on February 6, 2017 [2 favorites]
Yeah, I'd meant more Street Fighter than Mario Kart. ; )
posted by DoctorFedora at 1:47 AM on February 6, 2017
posted by DoctorFedora at 1:47 AM on February 6, 2017
In fairness, it's hard to get more equal than Ryu and Ken.
posted by jaduncan at 1:52 AM on February 6, 2017 [1 favorite]
posted by jaduncan at 1:52 AM on February 6, 2017 [1 favorite]
Simple answer to luck in video games - Don't play on my team.
posted by Samizdata at 1:54 AM on February 6, 2017 [2 favorites]
posted by Samizdata at 1:54 AM on February 6, 2017 [2 favorites]
Beholder: "Normal mode is easy, difficult mode is normal, and hardcore mode is difficult, and I wouldn't want it any other way. I'm not looking to prove anything. I just want to have fun.
Example, my favorite video game is Borderlands 2. For those who keep up with such things, one of the game's expansion classes had a very powerful ability, which was specifically included for players who's reaction time isn't very good; any bullets fired at a target still had a chance to ricochet and hit the target even if their shot missed. This made the game much easier/satisfying for people who aren't very good at first person shooters, which is why it's my favorite class by a mile. I wish other game studios would do this. Not everyone is good at twitch games, but we still want to play, and we have the money, so make a place for us at the table even if the purist howl."
Salvador? (Been a long old while since I played 2.)
posted by Samizdata at 1:54 AM on February 6, 2017 [1 favorite]
Example, my favorite video game is Borderlands 2. For those who keep up with such things, one of the game's expansion classes had a very powerful ability, which was specifically included for players who's reaction time isn't very good; any bullets fired at a target still had a chance to ricochet and hit the target even if their shot missed. This made the game much easier/satisfying for people who aren't very good at first person shooters, which is why it's my favorite class by a mile. I wish other game studios would do this. Not everyone is good at twitch games, but we still want to play, and we have the money, so make a place for us at the table even if the purist howl."
Salvador? (Been a long old while since I played 2.)
posted by Samizdata at 1:54 AM on February 6, 2017 [1 favorite]
Fascinating article. I hadn't heard of those particular tricks.
This is maybe a type of sausage we don't want to see made. But we humans really have lousy intuitions about probability, so it's not surprising that developers fudge things a bit. Few customers write in to complain that they never get terrible results 8 times in a row.
Heck, this goes back to AD&D— the manual itself had some strategies for "rolling 3d6" which created better rolls.
posted by zompist at 1:56 AM on February 6, 2017
This is maybe a type of sausage we don't want to see made. But we humans really have lousy intuitions about probability, so it's not surprising that developers fudge things a bit. Few customers write in to complain that they never get terrible results 8 times in a row.
Heck, this goes back to AD&D— the manual itself had some strategies for "rolling 3d6" which created better rolls.
posted by zompist at 1:56 AM on February 6, 2017
Salvador?
Mechromancer, pet plus ricochet ability. Fingers crossed that BL3 has a pet class.
posted by Beholder at 2:04 AM on February 6, 2017 [3 favorites]
Mechromancer, pet plus ricochet ability. Fingers crossed that BL3 has a pet class.
posted by Beholder at 2:04 AM on February 6, 2017 [3 favorites]
There's also the flipside of this; Civ also *heavily* cheats on higher levels with AI that can see the whole map etc. It's mostly about maintaining the illusion of an equal chance, at least until AI gets a lot better.
posted by jaduncan at 2:37 AM on February 6, 2017 [2 favorites]
posted by jaduncan at 2:37 AM on February 6, 2017 [2 favorites]
I can't believe Peggle lied to me. I feel dirty now.
posted by parm at 3:35 AM on February 6, 2017 [6 favorites]
posted by parm at 3:35 AM on February 6, 2017 [6 favorites]
> "Like when I attack the AI's spearmen with my tank unit... and lose. It is perfectly natural for me to yell at my cheating ass computer, even if this only happens 1 in 10 times."
1 in 10 times, they are Ewoks.
posted by kyrademon at 4:07 AM on February 6, 2017 [15 favorites]
1 in 10 times, they are Ewoks.
posted by kyrademon at 4:07 AM on February 6, 2017 [15 favorites]
In World of Warcraft, every time players defeat a foe, they hope to receive a “Legendary”—one of the game’s highly powerful weapons. Legendaries have an infinitesimally small chance of being “dropped,” but are also on a pity timer. “Fatigue can set in where a player is just waiting for the pity timer to kick in,” Sottosanti said. “The primary emotion they feel upon finally finding a Legendary is often not joy but relief, perhaps tinged with sadness.”
That's in part because several of the Legendaries suck and it's unfun to get them in place of something that could've significantly improved your character.
posted by Pope Guilty at 4:23 AM on February 6, 2017
That's in part because several of the Legendaries suck and it's unfun to get them in place of something that could've significantly improved your character.
posted by Pope Guilty at 4:23 AM on February 6, 2017
"Fairness is the unspoken promise of most video games. "
With all due respect, this is a false premise. Video games are not true games (chess/monopoly) and the unspoken promise is really just fun or at least "don't bore me too much".
There certainly can be a fair game in video format but like a correctly written tic-tac-toe game there is no actual way to win unless the game algorithm includes a calculated (possibly randomized) move where the compute looses a certain percentage. That can be fun, exciting, a good challenge, a method of teaching new skill, but in no way is it "fair".
posted by sammyo at 4:37 AM on February 6, 2017 [2 favorites]
With all due respect, this is a false premise. Video games are not true games (chess/monopoly) and the unspoken promise is really just fun or at least "don't bore me too much".
There certainly can be a fair game in video format but like a correctly written tic-tac-toe game there is no actual way to win unless the game algorithm includes a calculated (possibly randomized) move where the compute looses a certain percentage. That can be fun, exciting, a good challenge, a method of teaching new skill, but in no way is it "fair".
posted by sammyo at 4:37 AM on February 6, 2017 [2 favorites]
With all due respect, this is a false premise. Video games are not true games (chess/monopoly) and the unspoken promise is really just fun or at least "don't bore me too much".
Monopoly literally just features an RNG and some decision making per turn, complete with other random events via card decks (themselves essentially a RNG, in the video game context). It isn't fair at all. What is the difference from a video game here, and how is that fairer than, say, games designed to depend on individual skill against the other player such as Street Fighter n or Quake y?
A fight between Ken and Ryu is, as noted above, about as equal as things can get. They are essentially the same character with different graphics. That is much, much more a true (although fast-paced) game than Monopoly, if true games are defined as fair.
posted by jaduncan at 4:43 AM on February 6, 2017 [6 favorites]
Monopoly literally just features an RNG and some decision making per turn, complete with other random events via card decks (themselves essentially a RNG, in the video game context). It isn't fair at all. What is the difference from a video game here, and how is that fairer than, say, games designed to depend on individual skill against the other player such as Street Fighter n or Quake y?
A fight between Ken and Ryu is, as noted above, about as equal as things can get. They are essentially the same character with different graphics. That is much, much more a true (although fast-paced) game than Monopoly, if true games are defined as fair.
posted by jaduncan at 4:43 AM on February 6, 2017 [6 favorites]
DOTA has had since its inception one of the better implementations I've seen - it's called pseudo random distribution.
For example, an ability that would normally have 25% to trigger per attack, in most games would normally just be a regular 1/4 dice roll each time.
However under pseudo random distribution they instead choose another value - in this case, 8.5% as the chance to trigger - and increase the chance each time it fails to trigger until a trigger occurs, at which point it resets back to 8.5%. So the first hit would have an 8.5% chance - and if it fails, the second hit has a 17.0% chance, then if it fails the third hit has 25.5% chance and so on.
Basically this reduces outliers at either end - it's much rare to trigger several times in a row, and also much rarer to go for long periods without a trigger. It's applied selectively to certainly classes of abilities which they don't want large variance in.
posted by xdvesper at 4:47 AM on February 6, 2017 [4 favorites]
For example, an ability that would normally have 25% to trigger per attack, in most games would normally just be a regular 1/4 dice roll each time.
However under pseudo random distribution they instead choose another value - in this case, 8.5% as the chance to trigger - and increase the chance each time it fails to trigger until a trigger occurs, at which point it resets back to 8.5%. So the first hit would have an 8.5% chance - and if it fails, the second hit has a 17.0% chance, then if it fails the third hit has 25.5% chance and so on.
Basically this reduces outliers at either end - it's much rare to trigger several times in a row, and also much rarer to go for long periods without a trigger. It's applied selectively to certainly classes of abilities which they don't want large variance in.
posted by xdvesper at 4:47 AM on February 6, 2017 [4 favorites]
Samsung Roulette, anyone? Samsung's battery blowup makes me more likely to buy the next Note -- not less
posted by fairmettle at 4:48 AM on February 6, 2017
posted by fairmettle at 4:48 AM on February 6, 2017
Man, folks out there saying Ryu vs. Ken is fundamentally a mirror match have clearly not played a Street Fighter game since SF2 Championship Edition ; )
(Though to some extent or other, they've generally done a pretty good job of asynchronous balance without too many wildly uneven matchups)
posted by DoctorFedora at 5:01 AM on February 6, 2017 [5 favorites]
(Though to some extent or other, they've generally done a pretty good job of asynchronous balance without too many wildly uneven matchups)
posted by DoctorFedora at 5:01 AM on February 6, 2017 [5 favorites]
It would seem that engineered luck has adjusted the baseline of gameplay kindness expectations: something like Flappy Bird might have once felt normal, but instead it felt cruel.
(Of course, Flappy Bird was also hugely addictive, so maybe we should give players a little more credit instead of pandering?)
posted by thejoshu at 5:22 AM on February 6, 2017 [3 favorites]
(Of course, Flappy Bird was also hugely addictive, so maybe we should give players a little more credit instead of pandering?)
posted by thejoshu at 5:22 AM on February 6, 2017 [3 favorites]
For the life of me I can't remember the "replay the same battle for increasing cost" mechanic in Civ. The only replay was save and reload I thought?
I'm guessing it's a poor explanation of the 'stack of axemen attacking three archers in a city' style scenario - you get at least three attempts at effectively the same battle (fresh axeman vs fresh archer).
posted by Dysk at 5:24 AM on February 6, 2017 [3 favorites]
I'm guessing it's a poor explanation of the 'stack of axemen attacking three archers in a city' style scenario - you get at least three attempts at effectively the same battle (fresh axeman vs fresh archer).
posted by Dysk at 5:24 AM on February 6, 2017 [3 favorites]
It always did annoy me in Shadowrun Returns that a character with an 85% chance to hit misses five times in a row... and subjectively it felt like it was always the same character who had that kind of luck.
Along with the intuitive misunderstanding of probability, there's also a suspicion of buggy software, hidden variables that are not factored into the odds that the game tells you, and the less rational feeling that the game (or designer) fudges the numbers because it wants you to lose this time for some reason.
posted by Foosnark at 5:28 AM on February 6, 2017 [6 favorites]
Along with the intuitive misunderstanding of probability, there's also a suspicion of buggy software, hidden variables that are not factored into the odds that the game tells you, and the less rational feeling that the game (or designer) fudges the numbers because it wants you to lose this time for some reason.
posted by Foosnark at 5:28 AM on February 6, 2017 [6 favorites]
Luck plays a big part in real games like monopoly too. Like when I am lucky enough to get picked as banker and get to steal all the money.
posted by sexyrobot at 5:46 AM on February 6, 2017 [3 favorites]
posted by sexyrobot at 5:46 AM on February 6, 2017 [3 favorites]
Just going to drop a link for the 8 reasons people like to play games (as far as it is understood).
It's important to remember that one of the reasons gaming is fun is to challenge yourself, but also that there are 7 other things that also make games worth playing for many people!
posted by Nutri-Matic Drinks Synthesizer at 6:14 AM on February 6, 2017 [5 favorites]
It's important to remember that one of the reasons gaming is fun is to challenge yourself, but also that there are 7 other things that also make games worth playing for many people!
posted by Nutri-Matic Drinks Synthesizer at 6:14 AM on February 6, 2017 [5 favorites]
Well, nothing beats Football Manager deciding on a Europa League semi-final second leg match my forwards could not stop hitting the woodworks more than a handful of times, and my opponents scoring on the only time they reached my goal. Reloaded, same thing. Reloaded again, now with a slightly different tactic and the same happened. Uninstalled.
(Cue in a raged out Miles Jacobson storming in calling me a liar and implying I have a pirate copy)
I wish other game studios would do this. Not everyone is good at twitch games, but we still want to play, and we have the money, so make a place for us at the table even if the purist howl.
A lot of games now have such modes now, one of them even calls it "just give me the story".
posted by lmfsilva at 6:51 AM on February 6, 2017
(Cue in a raged out Miles Jacobson storming in calling me a liar and implying I have a pirate copy)
I wish other game studios would do this. Not everyone is good at twitch games, but we still want to play, and we have the money, so make a place for us at the table even if the purist howl.
A lot of games now have such modes now, one of them even calls it "just give me the story".
posted by lmfsilva at 6:51 AM on February 6, 2017
Well, nothing beats Football Manager deciding on a Europa League semi-final second leg match my forwards could not stop hitting the woodworks more than a handful of times, and my opponents scoring on the only time they reached my goal. Reloaded, same thing. Reloaded again, now with a slightly different tactic and the same happened. Uninstalled.
I wonder if matches are predetermined based on a starting random seed (possibly specifically to prevent save-scumming).
posted by EndsOfInvention at 7:00 AM on February 6, 2017 [1 favorite]
I wonder if matches are predetermined based on a starting random seed (possibly specifically to prevent save-scumming).
posted by EndsOfInvention at 7:00 AM on February 6, 2017 [1 favorite]
This bothers me a lot. People really need to get a better understanding of risk.
I agree entirely, and it bothers me too.
Why would this bother you? Video games aren't simulations of reality (except for those that explicitly are), they're supposed to be fun experiences.
If a video game is just supposed to be a fun experience, then it is intellectually empty. In truth, video games can teach, inform, prepare, and give the benefits of living other lives without the risk. I, at least, am not interested in games where the dev is just massaging a player's pleasure centers, saying "Who's a good boy." That kind of thing might be okay once in a while, but it's everywhere.
Having spent years writing about roguelikes, as you might imagine, I have rather a lot to say about randomness in games. A good illustration of the uses, and problems, with randomness, is the differences between Pac-Man and Ms. Pac-Man.
Pac-Man has little randomness (the only pseudorandom element is the movement of vulnerable ghosts), and what it has can be manipulated. This is why patterns work, which is generally a major flaw in the game.
Ms. Pac-Man has randomness in the first few seconds of each level, as some monsters move randomly. That means patterns do not work, even though most of the time they use the same AI as in Pac-Man, because their early random movements change the state of the board in a way independent of the player's actions. That's good randomness because it gives the player a fair chance to react to it, and forces him to improvise.
However, after board 7, Ms. Pac-Man gives the player random fruit, worth random points. This means a good player's score becomes more about which awards the machine chooses to allow than how well he plays. Two players of equal skill could have greatly different scores because one player got lots of bananas while the other was stuck with cherries. That's bad randomness because the player cannot affect it.
I am very skeptical about Olaf's magical last throw.
If the Super Bowl were an arcade football game, you'd have good reason to be! Old arcade sports games like NBA Jam and NFL Blitz were long-known to bias the odds of making shots by whether you're head or behind, as a form of rubberbanding.
Monopoly literally just features an RNG and some decision making per turn, complete with other random events via card decks (themselves essentially a RNG, in the video game context). It isn't fair at all.
I am not a fan of Monopoly, but there are some things it does well. One thing about it is that is fair, in the sense that, whatever biases the dice have, each of the players has an equal chance to get it. And believe it or not, it can be fun to play an unfair game, if the unfairness is presented to you honestly. It can be challenging to overcome the odds, and satisfying if you succeed. Ideally, the trailing players will collude against the leading players to even out the game, although few games of Monopoly I've ever played had this happen.
But anyway, part of what is interesting about Monopoly is that the dice are not *entirely* random, but follow a probability curve, which, along with doubles, cards and Jail, are why some spaces are landed on more often than others, which is in turn part of why some properties are better buys than others. If all the properties were equally good then it'd be a null choice, you wouldn't have sufficient knowledge to make a good decision and thus all the choices would be the same to you.
posted by JHarris at 7:08 AM on February 6, 2017 [13 favorites]
I agree entirely, and it bothers me too.
Why would this bother you? Video games aren't simulations of reality (except for those that explicitly are), they're supposed to be fun experiences.
If a video game is just supposed to be a fun experience, then it is intellectually empty. In truth, video games can teach, inform, prepare, and give the benefits of living other lives without the risk. I, at least, am not interested in games where the dev is just massaging a player's pleasure centers, saying "Who's a good boy." That kind of thing might be okay once in a while, but it's everywhere.
Having spent years writing about roguelikes, as you might imagine, I have rather a lot to say about randomness in games. A good illustration of the uses, and problems, with randomness, is the differences between Pac-Man and Ms. Pac-Man.
Pac-Man has little randomness (the only pseudorandom element is the movement of vulnerable ghosts), and what it has can be manipulated. This is why patterns work, which is generally a major flaw in the game.
Ms. Pac-Man has randomness in the first few seconds of each level, as some monsters move randomly. That means patterns do not work, even though most of the time they use the same AI as in Pac-Man, because their early random movements change the state of the board in a way independent of the player's actions. That's good randomness because it gives the player a fair chance to react to it, and forces him to improvise.
However, after board 7, Ms. Pac-Man gives the player random fruit, worth random points. This means a good player's score becomes more about which awards the machine chooses to allow than how well he plays. Two players of equal skill could have greatly different scores because one player got lots of bananas while the other was stuck with cherries. That's bad randomness because the player cannot affect it.
I am very skeptical about Olaf's magical last throw.
If the Super Bowl were an arcade football game, you'd have good reason to be! Old arcade sports games like NBA Jam and NFL Blitz were long-known to bias the odds of making shots by whether you're head or behind, as a form of rubberbanding.
Monopoly literally just features an RNG and some decision making per turn, complete with other random events via card decks (themselves essentially a RNG, in the video game context). It isn't fair at all.
I am not a fan of Monopoly, but there are some things it does well. One thing about it is that is fair, in the sense that, whatever biases the dice have, each of the players has an equal chance to get it. And believe it or not, it can be fun to play an unfair game, if the unfairness is presented to you honestly. It can be challenging to overcome the odds, and satisfying if you succeed. Ideally, the trailing players will collude against the leading players to even out the game, although few games of Monopoly I've ever played had this happen.
But anyway, part of what is interesting about Monopoly is that the dice are not *entirely* random, but follow a probability curve, which, along with doubles, cards and Jail, are why some spaces are landed on more often than others, which is in turn part of why some properties are better buys than others. If all the properties were equally good then it'd be a null choice, you wouldn't have sufficient knowledge to make a good decision and thus all the choices would be the same to you.
posted by JHarris at 7:08 AM on February 6, 2017 [13 favorites]
> "It's important to remember that one of the reasons gaming is fun is to challenge yourself, but also that there are 7 other things that also make games worth playing for many people!"
Yup. I have no interest in being challenged by a videogame. Challenging videogames bore me.
posted by kyrademon at 7:10 AM on February 6, 2017 [1 favorite]
Yup. I have no interest in being challenged by a videogame. Challenging videogames bore me.
posted by kyrademon at 7:10 AM on February 6, 2017 [1 favorite]
XCOM (the modern game) has a bunch of chance-to-hit cheating going on in the background. On Easy or Normal it makes every 84% or higher Chance To Hit shot hit automatically, and if you miss a >50% CTH shot, your next shot gets more likely to hit. XCOM also uses the predetermined random seed value that prevents save-scumming (so if for any given shot, the hit/miss result will be the same however many times you reload the game).
I remember when it was released a lot of people complaining about missing "too many" shots with 85% CTH, although I'm not sure whether that was just because the game wasn't compensating for player bias on the harder difficulties.
posted by EndsOfInvention at 7:11 AM on February 6, 2017 [8 favorites]
I remember when it was released a lot of people complaining about missing "too many" shots with 85% CTH, although I'm not sure whether that was just because the game wasn't compensating for player bias on the harder difficulties.
posted by EndsOfInvention at 7:11 AM on February 6, 2017 [8 favorites]
(Note: Oops -- I didn't watch the Super Bowl last night [sports are not my thing] and assumed the thing about Olaf had to do with the Patriots upset victory. My point still holds, it's just not relevant to the comment I quoted, heh.)
posted by JHarris at 7:12 AM on February 6, 2017
posted by JHarris at 7:12 AM on February 6, 2017
BTW in the article there's a bit of a problem. Besides being filled with the kind of "brain chemical" stuff I think is a plague on gaming, their use of the term "ball saver" in pinball has nothing to do with randomness!
If a ball saver is on the board it's visible, it's in play and can be taken advantage of. Larry DeMar's statement is probably more about taking away player crutches. A better statement about randomness in pinball would be about random awards awarded by software, which often are manipulated according to how the player's been doing. That is a known an recognized thing in pinball circles, to the extend that machine tournament modes have the effect of derandomizing those awards so they always pay out the same things.
posted by JHarris at 7:24 AM on February 6, 2017 [5 favorites]
If a ball saver is on the board it's visible, it's in play and can be taken advantage of. Larry DeMar's statement is probably more about taking away player crutches. A better statement about randomness in pinball would be about random awards awarded by software, which often are manipulated according to how the player's been doing. That is a known an recognized thing in pinball circles, to the extend that machine tournament modes have the effect of derandomizing those awards so they always pay out the same things.
posted by JHarris at 7:24 AM on February 6, 2017 [5 favorites]
Why would this bother you? Video games aren't simulations of reality (except for those that explicitly are), they're supposed to be fun experiences. Designers make unrealistic choices in games to improve player experience all the time.
But people play games a lot, and when you are getting trained, that thinking will spill over to other parts of your life. People form habits, and habits form expectations. We take this way too lightly, thinking we can handle silent and subtle alterations in our expectations and thinking.
People do not understand rigs. They take rigs as lessons in life and then believe those rigs are reality. It is no different than people reading or watching stories that impose certain ideas of what reality is and then they make decisions based on what they learned.
There is a simple way to confirm or refute these ideas, and there may be experimental psychologists who would have more knowledge about the role of learning on our expectations.
posted by Alexandra Kitty at 7:38 AM on February 6, 2017 [10 favorites]
But people play games a lot, and when you are getting trained, that thinking will spill over to other parts of your life. People form habits, and habits form expectations. We take this way too lightly, thinking we can handle silent and subtle alterations in our expectations and thinking.
People do not understand rigs. They take rigs as lessons in life and then believe those rigs are reality. It is no different than people reading or watching stories that impose certain ideas of what reality is and then they make decisions based on what they learned.
There is a simple way to confirm or refute these ideas, and there may be experimental psychologists who would have more knowledge about the role of learning on our expectations.
posted by Alexandra Kitty at 7:38 AM on February 6, 2017 [10 favorites]
And believe it or not, it can be fun to play an unfair game, if the unfairness is presented to you honestly.
I'm with you on this -- I don't think it's a problem for a game to gate how many misses you can have, but I think it is a problem for the game to lie to you about this. For the Civ example, you'd get the exact same effect if the game just said to you, "this will take 1-3 attempts to succeed", instead of claiming each event was precisely a 1/3 chance (with the third being guaranteed, behind the curtain).
Having important aspects of a game be a lottery that occasionally takes pity on you and breaks its own rules to keep you coming back is a bit too pavlovian for my tastes.
posted by tocts at 7:41 AM on February 6, 2017 [4 favorites]
I'm with you on this -- I don't think it's a problem for a game to gate how many misses you can have, but I think it is a problem for the game to lie to you about this. For the Civ example, you'd get the exact same effect if the game just said to you, "this will take 1-3 attempts to succeed", instead of claiming each event was precisely a 1/3 chance (with the third being guaranteed, behind the curtain).
Having important aspects of a game be a lottery that occasionally takes pity on you and breaks its own rules to keep you coming back is a bit too pavlovian for my tastes.
posted by tocts at 7:41 AM on February 6, 2017 [4 favorites]
League of Legends has a considered approach to randomness that I think is very shrewd, particularly for a skill-based PvP esport. They've basically removed almost all randomness from game interactions. Attacks always hit, the damage they do is completely deterministic, no dice rolls. (Critical strike is the one exception.) When you get in a fight in LoL it mostly feels like the players' actions determine the outcome, not luck.
But they do have randomness in the environment. What Ghostcrawler calls "input randomness" in the video. For instance there's four different kinds of dragon that can spawn in the game, an epic monster for people to kill. It's completely random which one spawns. But everyone sees ahead of time what dragon it's going to be and can make a decision about whether they want to contest it. The randomness gives the game variance, but players control how they react to the randomness.
Also LoL players talk about luck all the time. "Unlucky" is an actual meme in the community. But what LoL folks mean by "luck" isn't a coin-flip. It's used to describe the outcome of decisions players make with imperfect information. Like if I go to face-check a brush where I have no vision, it's "luck" whether there's an enemy there waiting to ambush me or not. The situation is entirely deterministic, the enemy chose whether to hide in that brush or not, but I don't know what the situation is. So if I get killed by an ambush I say "that was unlucky". (Mostly as a rueful joke.)
The magic part about League is even this kind of imperfect information situation is under player control. Getting better at the game makes you have more information. Good players have enough map awareness they can often predict whether there's an ambush or not. The luck you make is the best kind of luck.
posted by Nelson at 7:43 AM on February 6, 2017 [8 favorites]
But they do have randomness in the environment. What Ghostcrawler calls "input randomness" in the video. For instance there's four different kinds of dragon that can spawn in the game, an epic monster for people to kill. It's completely random which one spawns. But everyone sees ahead of time what dragon it's going to be and can make a decision about whether they want to contest it. The randomness gives the game variance, but players control how they react to the randomness.
Also LoL players talk about luck all the time. "Unlucky" is an actual meme in the community. But what LoL folks mean by "luck" isn't a coin-flip. It's used to describe the outcome of decisions players make with imperfect information. Like if I go to face-check a brush where I have no vision, it's "luck" whether there's an enemy there waiting to ambush me or not. The situation is entirely deterministic, the enemy chose whether to hide in that brush or not, but I don't know what the situation is. So if I get killed by an ambush I say "that was unlucky". (Mostly as a rueful joke.)
The magic part about League is even this kind of imperfect information situation is under player control. Getting better at the game makes you have more information. Good players have enough map awareness they can often predict whether there's an ambush or not. The luck you make is the best kind of luck.
posted by Nelson at 7:43 AM on February 6, 2017 [8 favorites]
very few ninjas can shoot fireballs and run up walls
Pffft. Next you'll be trying to convince me that bricks don't actually disappear when you bounce balls off them.
posted by flabdablet at 7:44 AM on February 6, 2017 [1 favorite]
Pffft. Next you'll be trying to convince me that bricks don't actually disappear when you bounce balls off them.
posted by flabdablet at 7:44 AM on February 6, 2017 [1 favorite]
Also, and maybe more to the point, I feel cheated: the game lied to me. It's not a 33% chance of winning.
Yeah, but in real life, anyone who tells you you have a 33% chance of winning some battle is lying, because it's not possible to know that. The most unrealistic (and maybe the most pernicious) aspect of civ games and city builders is the ridiculously comprehensive and accurate information the game gives you about the state of the world. Mayors' and city managers' jobs would be radically different if they had the kind of data available that SimCity gives you.
posted by straight at 7:56 AM on February 6, 2017 [10 favorites]
Yeah, but in real life, anyone who tells you you have a 33% chance of winning some battle is lying, because it's not possible to know that. The most unrealistic (and maybe the most pernicious) aspect of civ games and city builders is the ridiculously comprehensive and accurate information the game gives you about the state of the world. Mayors' and city managers' jobs would be radically different if they had the kind of data available that SimCity gives you.
posted by straight at 7:56 AM on February 6, 2017 [10 favorites]
However under pseudo random distribution they instead choose another value - in this case, 8.5% as the chance to trigger - and increase the chance each time it fails to trigger until a trigger occurs, at which point it resets back to 8.5%.
This isn't unique to video games; it's done elsewhere, too. (Although video games make it easy to twist the probability dials behind the scenes, compared to dice or card-based games.) In particular, this exact mechanic is used in bingo—which I find interesting as a betting "game", because it's literally nothing but randomness—in order to produce occasional big jackpots but still be unpredictable. This is called a "progressive jackpot", at least as far as I've seen it.
Typically a progressive jackpot game will have two payoff values, a high number which is obviously what you want to win (say, $500), and a consolation value (say, $50). And you only get the higher payout if you win in under a certain number of calls — starting with something that's just on the raw edge of possible (say 5 calls, since most winning bingo combinations require a minimum of 4 numbers called). Each time the game is played—every day or every hour or whatever—both numbers, the jackpot pool and the number of calls allowable to win it in, go up. Eventually you get to something that's both a big prize and seemingly-attainable (35 or 40 calls is nearly certain to produce a winner given enough players, depending on the pattern being sought). The players obviously know this, so as you start to get into games that have a multi-thousand-dollar payout and give you dozens of calls to win, the game suddenly becomes more popular.
From the perspective of someone running the game, this is nice because it doesn't require any particularly complex math or up-my-sleeve chicanery, and it generally results in more players (which is to say, people paying into the pool) as the possible downside risk of paying out a big prize increases. And psychologically it produces a much more "exciting" game than if you just had 1 game in 25 or whatever pay out 20x more than normal.
It doesn't surprise me that video games do the same thing, in fact it would be a bit odd if they didn't. People don't like purely mechanistic probability. I mean, it would be easy enough to run a "bingo hall" where people walked in, paid their entire budget for the evening, and then got handed back their winnings immediately, without the need to actually 'play'. You could even run through the games in simulation, using a TRNG, and know to a very high degree of confidence that the whole thing was fair, and probably a lot more random than pulling physical plastic balls from an air hopper. But it would be terribly unsatisfying. And betting games involving a house edge—as virtually all do—are arguably "unfair" by definition. But most people, at least those who are playing, don't care, or rather they accept that unfairness as a condition of the game; the issue of "fairness" really only comes in when it's one player vs. another player.
It's interesting that there are games, both traditional and video-based, which also manage to juke the player-vs-player relationship in a way that people don't mind. That seems like a much more delicate proposition (and, unsurprisingly, you don't see it very often in games that are being played for money). E.g. some modern board games are designed expressly to limit the role of skill and experience so that new players have a reasonable chance of winning vs. experienced players. If they didn't, the games might be pretty unappealing to new players and perceptively less "fun". Other games that are expressly skill-based solve this through the use of handicapping (e.g. golf). Again, the issue seems to be a tradeoff in fairness in order to make the game not-unappealing to less skilled people, and allow friendly competition across skill levels—although the handicapping, being done in the open, also makes clear who the more-skilled player actually is. I can't think offhand of any video games that have a hard handicapping system like that, although lots of games offer character selection options that make life intentionally harder (I believe that Mortal Kombat only included "Jax" for the purpose of giving older siblings a character they could play as to handicap themselves vs. whiny younger siblings who wouldn't play any other way, not that I'm salty about that at all).
posted by Kadin2048 at 8:10 AM on February 6, 2017 [7 favorites]
This isn't unique to video games; it's done elsewhere, too. (Although video games make it easy to twist the probability dials behind the scenes, compared to dice or card-based games.) In particular, this exact mechanic is used in bingo—which I find interesting as a betting "game", because it's literally nothing but randomness—in order to produce occasional big jackpots but still be unpredictable. This is called a "progressive jackpot", at least as far as I've seen it.
Typically a progressive jackpot game will have two payoff values, a high number which is obviously what you want to win (say, $500), and a consolation value (say, $50). And you only get the higher payout if you win in under a certain number of calls — starting with something that's just on the raw edge of possible (say 5 calls, since most winning bingo combinations require a minimum of 4 numbers called). Each time the game is played—every day or every hour or whatever—both numbers, the jackpot pool and the number of calls allowable to win it in, go up. Eventually you get to something that's both a big prize and seemingly-attainable (35 or 40 calls is nearly certain to produce a winner given enough players, depending on the pattern being sought). The players obviously know this, so as you start to get into games that have a multi-thousand-dollar payout and give you dozens of calls to win, the game suddenly becomes more popular.
From the perspective of someone running the game, this is nice because it doesn't require any particularly complex math or up-my-sleeve chicanery, and it generally results in more players (which is to say, people paying into the pool) as the possible downside risk of paying out a big prize increases. And psychologically it produces a much more "exciting" game than if you just had 1 game in 25 or whatever pay out 20x more than normal.
It doesn't surprise me that video games do the same thing, in fact it would be a bit odd if they didn't. People don't like purely mechanistic probability. I mean, it would be easy enough to run a "bingo hall" where people walked in, paid their entire budget for the evening, and then got handed back their winnings immediately, without the need to actually 'play'. You could even run through the games in simulation, using a TRNG, and know to a very high degree of confidence that the whole thing was fair, and probably a lot more random than pulling physical plastic balls from an air hopper. But it would be terribly unsatisfying. And betting games involving a house edge—as virtually all do—are arguably "unfair" by definition. But most people, at least those who are playing, don't care, or rather they accept that unfairness as a condition of the game; the issue of "fairness" really only comes in when it's one player vs. another player.
It's interesting that there are games, both traditional and video-based, which also manage to juke the player-vs-player relationship in a way that people don't mind. That seems like a much more delicate proposition (and, unsurprisingly, you don't see it very often in games that are being played for money). E.g. some modern board games are designed expressly to limit the role of skill and experience so that new players have a reasonable chance of winning vs. experienced players. If they didn't, the games might be pretty unappealing to new players and perceptively less "fun". Other games that are expressly skill-based solve this through the use of handicapping (e.g. golf). Again, the issue seems to be a tradeoff in fairness in order to make the game not-unappealing to less skilled people, and allow friendly competition across skill levels—although the handicapping, being done in the open, also makes clear who the more-skilled player actually is. I can't think offhand of any video games that have a hard handicapping system like that, although lots of games offer character selection options that make life intentionally harder (I believe that Mortal Kombat only included "Jax" for the purpose of giving older siblings a character they could play as to handicap themselves vs. whiny younger siblings who wouldn't play any other way, not that I'm salty about that at all).
posted by Kadin2048 at 8:10 AM on February 6, 2017 [7 favorites]
I don't think I could ever happily play a video game knowing that the probabilities that the game reports to me are not what are used internally. It'd be like people seeing 2+2 to everywhere, but the game rounding it up to 5 just to make them feel better. *shiver*
posted by Alex404 at 8:20 AM on February 6, 2017 [1 favorite]
posted by Alex404 at 8:20 AM on February 6, 2017 [1 favorite]
Recently I've been playing a strategy RPG on a Nintendo DS emulator on my computer. The game reports hit percentages to you. So if you attack a beefy, slow enemy it might report a 99% chance to hit. If you attack a fast, agile enemy it might report a 60% chance to hit.
Playing on an emulator means that you can use what are called "save states". A save state takes a snapshot of the game at that moment, giving you the ability to save progress beyond what the game allows. You can obviously use this to cheat, if you want. I typically don't abuse save states, but sometimes if a battle is boring and I don't feel like replaying it, then I'll make save states at different points and allow myself to replay critical parts, making different moves.
In doing this one time, I found out that the percent chance to hit is actually complete bullshit. Certain enemies seem to get randomly assigned a turn where they cannot be hit by an attack, and certain members of your party get designated rounds where they cannot hit any enemies regardless of their hit percentage. I found it out when I was doing a "speed battle" which has a time limit on it, and my last shot at winning the battle got a miss-round dumped on her.
I actually don't particularly care about this specific mechanic, but the fact that the game is telling me one thing and then doing another was uniquely infuriating. It is ironic that I only figured this out by cheating, however.
posted by codacorolla at 8:28 AM on February 6, 2017
Playing on an emulator means that you can use what are called "save states". A save state takes a snapshot of the game at that moment, giving you the ability to save progress beyond what the game allows. You can obviously use this to cheat, if you want. I typically don't abuse save states, but sometimes if a battle is boring and I don't feel like replaying it, then I'll make save states at different points and allow myself to replay critical parts, making different moves.
In doing this one time, I found out that the percent chance to hit is actually complete bullshit. Certain enemies seem to get randomly assigned a turn where they cannot be hit by an attack, and certain members of your party get designated rounds where they cannot hit any enemies regardless of their hit percentage. I found it out when I was doing a "speed battle" which has a time limit on it, and my last shot at winning the battle got a miss-round dumped on her.
I actually don't particularly care about this specific mechanic, but the fact that the game is telling me one thing and then doing another was uniquely infuriating. It is ironic that I only figured this out by cheating, however.
posted by codacorolla at 8:28 AM on February 6, 2017
A fight between Ken and Ryu is, as noted above, about as equal as things can get. They are essentially the same character with different graphics. That is much, much more a true (although fast-paced) game than Monopoly, if true games are defined as fair.
But to the point of the article, we're relying on the computer to be a neutral arbiter of player actions and that's not necessarily true. If I were designing a PVP game, I might chose to allow players with 15% or less health more leeway in timing combo moves in order to ratchet up the end of fight drama and try to convince the losing player that they were just starting to get good so they should pump another quarter in.
posted by Candleman at 8:29 AM on February 6, 2017
But to the point of the article, we're relying on the computer to be a neutral arbiter of player actions and that's not necessarily true. If I were designing a PVP game, I might chose to allow players with 15% or less health more leeway in timing combo moves in order to ratchet up the end of fight drama and try to convince the losing player that they were just starting to get good so they should pump another quarter in.
posted by Candleman at 8:29 AM on February 6, 2017
In doing this one time, I found out that the percent chance to hit is actually complete bullshit. Certain enemies seem to get randomly assigned a turn where they cannot be hit by an attack, and certain members of your party get designated rounds where they cannot hit any enemies regardless of their hit percentage.
It's not bullshit, it's still random. Kind of. Read this article for details.
posted by EndsOfInvention at 9:16 AM on February 6, 2017
It's not bullshit, it's still random. Kind of. Read this article for details.
posted by EndsOfInvention at 9:16 AM on February 6, 2017
Legendaries have an infinitesimally small chance of being “dropped,” but are also on a pity timer.
This is true, for this expansion. In the recent past, they were at the end of long, arduous quest-chains. This was more fair, in that if you do the quest, you get the loot. And they had some really interesting elements and stories. But, they could take months and months to complete, so as to prevent the more hardcore from getting them the first week and that meant if you started too late, or took too long, you couldn't complete the quest chain. Before that, they were purely random drops, and you'd have to farm for them - and when they were current, you'd also be competing against everyone else in the group to win it. And when they are old content, you can only run once a week and it's not at all uncommon for people to dedicate years to getting a legendary just so they can say they got it.
So, the problem WoW has is really a problem of gating. They want people to all have the same(ish) chance to get the best gear, but time/effort gating automatically excludes those of us who didn't move to Colorado to be full time WoW players, and the RNG gating means when RNG hates you, you lose. And because the gear is so good, it's a disadvantage to not have it.
Blizzard had implemented the Pity RNG a long time ago, to prevent the problem of having to kill 6000 yaks to get 8 eyeballs because RNG hates you (or all yaks are blind). Now, the drop rate is a progressive percentage, the more you kill the better the rate becomes - and the quests are so much less frustrating as a result.
I think they missed the mark applying this system to the legendaries. It might be better if they measured effort as a ratio of time spent in game, rather than a total X number of quests done or dungeons run, and used that to drive their gating. The current method is pretty badly flawed, and I don't think it's an improvement over the quest lines.
They also screwed up because I've gotten 4 legendaries on my 2 farming alts and only 1 on my main, and that's doubling frustrating because I don't really do much with the alts and would really prefer the best gear on my main. But, that is a separate rant.
posted by Pogo_Fuzzybutt at 9:19 AM on February 6, 2017 [1 favorite]
This is true, for this expansion. In the recent past, they were at the end of long, arduous quest-chains. This was more fair, in that if you do the quest, you get the loot. And they had some really interesting elements and stories. But, they could take months and months to complete, so as to prevent the more hardcore from getting them the first week and that meant if you started too late, or took too long, you couldn't complete the quest chain. Before that, they were purely random drops, and you'd have to farm for them - and when they were current, you'd also be competing against everyone else in the group to win it. And when they are old content, you can only run once a week and it's not at all uncommon for people to dedicate years to getting a legendary just so they can say they got it.
So, the problem WoW has is really a problem of gating. They want people to all have the same(ish) chance to get the best gear, but time/effort gating automatically excludes those of us who didn't move to Colorado to be full time WoW players, and the RNG gating means when RNG hates you, you lose. And because the gear is so good, it's a disadvantage to not have it.
Blizzard had implemented the Pity RNG a long time ago, to prevent the problem of having to kill 6000 yaks to get 8 eyeballs because RNG hates you (or all yaks are blind). Now, the drop rate is a progressive percentage, the more you kill the better the rate becomes - and the quests are so much less frustrating as a result.
I think they missed the mark applying this system to the legendaries. It might be better if they measured effort as a ratio of time spent in game, rather than a total X number of quests done or dungeons run, and used that to drive their gating. The current method is pretty badly flawed, and I don't think it's an improvement over the quest lines.
They also screwed up because I've gotten 4 legendaries on my 2 farming alts and only 1 on my main, and that's doubling frustrating because I don't really do much with the alts and would really prefer the best gear on my main. But, that is a separate rant.
posted by Pogo_Fuzzybutt at 9:19 AM on February 6, 2017 [1 favorite]
> Designers make unrealistic choices in games to improve player experience all the time.
This isn't really an example of an unrealistic choice, or a realistic choice, at all. This is simply poor problem analysis. Suppose that rather than changing the game mechanics to model what players thought the numbers meant, you instead decline to tell the player the probabilities? The numbers aren't significant because the sample size is too small, so why not just dispense with them?
Endless Legend has an interesting take on this, where rather than giving a probability of success, it shows the two armies' relative strength on something like a pie chart. Much more intuitive, and doesn't set up an expectation that requires too many iterations to obtain.
posted by I-Write-Essays at 9:20 AM on February 6, 2017 [2 favorites]
This isn't really an example of an unrealistic choice, or a realistic choice, at all. This is simply poor problem analysis. Suppose that rather than changing the game mechanics to model what players thought the numbers meant, you instead decline to tell the player the probabilities? The numbers aren't significant because the sample size is too small, so why not just dispense with them?
Endless Legend has an interesting take on this, where rather than giving a probability of success, it shows the two armies' relative strength on something like a pie chart. Much more intuitive, and doesn't set up an expectation that requires too many iterations to obtain.
posted by I-Write-Essays at 9:20 AM on February 6, 2017 [2 favorites]
Beholder: "Salvador?
Mechromancer, pet plus ricochet ability. Fingers crossed that BL3 has a pet class."
I guess I never followed that skill tree. Funny story, I was playing 2 and wrapping up old quests. I do the one with Moxxi's family and I go to the bar. I summon D3TH-TR4P, he does the explosive handclap and slaughters the entire bar in one shot. Yeah, so I was a little OP.
posted by Samizdata at 9:25 AM on February 6, 2017
Mechromancer, pet plus ricochet ability. Fingers crossed that BL3 has a pet class."
I guess I never followed that skill tree. Funny story, I was playing 2 and wrapping up old quests. I do the one with Moxxi's family and I go to the bar. I summon D3TH-TR4P, he does the explosive handclap and slaughters the entire bar in one shot. Yeah, so I was a little OP.
posted by Samizdata at 9:25 AM on February 6, 2017
And in any case, I think the real problem is that people don't understand how random number algorithms work on computers, more than their lack of understanding of risk. The actual meaning of saying you have a 75% chance of success breaks down on a single attack. Is that on d100 or 2d10?
posted by I-Write-Essays at 9:27 AM on February 6, 2017
posted by I-Write-Essays at 9:27 AM on February 6, 2017
I actually don't particularly care about this specific mechanic, but the fact that the game is telling me one thing and then doing another was uniquely infuriating. It is ironic that I only figured this out by cheating, however.
It's only happening because "random" numbers on computers aren't really random, but really a sequence of numbers generated by an algorithm. Every time you reload your save state, the next "random" value in the sequence is the same.
posted by neckro23 at 9:36 AM on February 6, 2017 [3 favorites]
It's only happening because "random" numbers on computers aren't really random, but really a sequence of numbers generated by an algorithm. Every time you reload your save state, the next "random" value in the sequence is the same.
posted by neckro23 at 9:36 AM on February 6, 2017 [3 favorites]
If it turns out the stats are off for the weapons in ADOM and Nethack, I'll be very upset.
When a game gives statistics about your chances, it should be right, dammit.
Also, does this mean my uber-cautious RTS / sim-game strategy was doomed from the start because the game rewards repeated futile attacks? Sad.
posted by steady-state strawberry at 9:41 AM on February 6, 2017
When a game gives statistics about your chances, it should be right, dammit.
Also, does this mean my uber-cautious RTS / sim-game strategy was doomed from the start because the game rewards repeated futile attacks? Sad.
posted by steady-state strawberry at 9:41 AM on February 6, 2017
It's only happening because "random" numbers on computers aren't really random, but really a sequence of numbers generated by an algorithm. Every time you reload your save state, the next "random" value in the sequence is the same.
This is a somewhat severe over-simplification. Putting aside true randomness, it takes extra work to result in a saved game having the same pseudorandom sequence -- you have to choose to keep (and presumably serialize/deserialize) your PRNG in the save state, instead of the game just having an PRNG that gets initialized when the game starts. If reloading a save results in the same random numbers every time, that's something the developer is doing on purpose.
posted by tocts at 9:46 AM on February 6, 2017 [3 favorites]
This is a somewhat severe over-simplification. Putting aside true randomness, it takes extra work to result in a saved game having the same pseudorandom sequence -- you have to choose to keep (and presumably serialize/deserialize) your PRNG in the save state, instead of the game just having an PRNG that gets initialized when the game starts. If reloading a save results in the same random numbers every time, that's something the developer is doing on purpose.
posted by tocts at 9:46 AM on February 6, 2017 [3 favorites]
(and, responding to point out that I'm dumb because this is an emulator save state, not in-game, so whatever's going on in memory is going to revert 100% back to what it was; though, depending on how far back the save state is, it could still be the game messing with you -- e.g. if you have a save state 5 minutes back, the likelihood of you doing all the same things that move the random number sequence forward the same way is very low)
posted by tocts at 9:49 AM on February 6, 2017 [2 favorites]
posted by tocts at 9:49 AM on February 6, 2017 [2 favorites]
It's not bullshit, it's still random. Kind of. Read this article for details.
I read the article. It's interesting, and explains the mechanic, but that doesn't mean that the number the game is reporting to me isn't bullshit. My issue is with the information that the game is giving me. That one character had 2 different targets, and 4 different directionalities to attack from. The percent chance to hit changed with each different target, and each different direction. But all eight of those combinations were misses. That means my actual percent chance to hit was 0%, because my character was in a miss round. Just tell me that, if that's how the system works, and don't give me a completely fake percent chance.
posted by codacorolla at 9:59 AM on February 6, 2017
I read the article. It's interesting, and explains the mechanic, but that doesn't mean that the number the game is reporting to me isn't bullshit. My issue is with the information that the game is giving me. That one character had 2 different targets, and 4 different directionalities to attack from. The percent chance to hit changed with each different target, and each different direction. But all eight of those combinations were misses. That means my actual percent chance to hit was 0%, because my character was in a miss round. Just tell me that, if that's how the system works, and don't give me a completely fake percent chance.
posted by codacorolla at 9:59 AM on February 6, 2017
The actual meaning of saying you have a 75% chance of success breaks down on a single attack. Is that on d100 or 2d10?
I assume by "2d10", you mean "1d10 for tens digit, 1d10 for ones digit", but how is that substantively different from 1d100?
posted by Etrigan at 10:02 AM on February 6, 2017
I assume by "2d10", you mean "1d10 for tens digit, 1d10 for ones digit", but how is that substantively different from 1d100?
posted by Etrigan at 10:02 AM on February 6, 2017
If reloading a save results in the same random numbers every time, that's something the developer is doing on purpose.
Some games do store the PRNG seed as part of the save game, precisely to discourage savescumming.
posted by escape from the potato planet at 10:05 AM on February 6, 2017 [2 favorites]
Some games do store the PRNG seed as part of the save game, precisely to discourage savescumming.
posted by escape from the potato planet at 10:05 AM on February 6, 2017 [2 favorites]
I was hoping this was entirely about XCOM and sadly it wasn't.
posted by GuyZero at 10:31 AM on February 6, 2017 [1 favorite]
posted by GuyZero at 10:31 AM on February 6, 2017 [1 favorite]
Some games do store the PRNG seed as part of the save game, precisely to discourage savescumming.
IIRC, the new XCOMs have an option during campaign creation that either enables or disables this. I can't recall which.
It's interesting because there are actually a lot of ways to exploit your knowledge of the PRNG seed with a we-all-move-you-all-move turn system. If you know that the first shot of your turn will hit and the second will miss (within certain hit chance parameters), you can cheese it out and make sure the first shot goes to a high-damage unit while the second goes to a unit with the weapon attachment or ability that does guaranteed damage on a miss. Or give the first shot to a unit with a lower chance to hit, so they make the "lucky" shot, and then give the second shot to somebody with extremely high innate accuracy, so they still might make the hit even with a "bad" roll.
posted by tobascodagama at 10:33 AM on February 6, 2017 [2 favorites]
IIRC, the new XCOMs have an option during campaign creation that either enables or disables this. I can't recall which.
It's interesting because there are actually a lot of ways to exploit your knowledge of the PRNG seed with a we-all-move-you-all-move turn system. If you know that the first shot of your turn will hit and the second will miss (within certain hit chance parameters), you can cheese it out and make sure the first shot goes to a high-damage unit while the second goes to a unit with the weapon attachment or ability that does guaranteed damage on a miss. Or give the first shot to a unit with a lower chance to hit, so they make the "lucky" shot, and then give the second shot to somebody with extremely high innate accuracy, so they still might make the hit even with a "bad" roll.
posted by tobascodagama at 10:33 AM on February 6, 2017 [2 favorites]
To be clear, I wasn't arguing that games don't save the PRNG state with save data, I was simply saying that if that's happening it's on purpose. It's less effort to not do it, so anyone doing it is doing it because that's how they want the player to experience the game.
posted by tocts at 10:47 AM on February 6, 2017
posted by tocts at 10:47 AM on February 6, 2017
I-Write-Essays: "The actual meaning of saying you have a 75% chance of success breaks down on a single attack. Is that on d100 or 2d10?"
A reported chance of 75% shouldn't be affected by the shape of the curve of absolute values of the die roll.
posted by Mitheral at 11:07 AM on February 6, 2017 [2 favorites]
A reported chance of 75% shouldn't be affected by the shape of the curve of absolute values of the die roll.
posted by Mitheral at 11:07 AM on February 6, 2017 [2 favorites]
> This isn't really an example of an unrealistic choice, or a realistic choice, at all. This is simply poor problem analysis. Suppose that rather than changing the game mechanics to model what players thought the numbers meant, you instead decline to tell the player the probabilities?
I think I-Write-Essays has it. The thing that bothers me about the hypothetical Civ example isn't manipulating probabilities to be more fun, or presenting the player with "33% chance of success", it's the lie in doing both at the same time. You can either just say "Chances don't look great", or display the true increased probability.
(As an aside, what would Fire Emblem's True Hit distribution be called?)
posted by lucidium at 11:16 AM on February 6, 2017 [3 favorites]
I think I-Write-Essays has it. The thing that bothers me about the hypothetical Civ example isn't manipulating probabilities to be more fun, or presenting the player with "33% chance of success", it's the lie in doing both at the same time. You can either just say "Chances don't look great", or display the true increased probability.
(As an aside, what would Fire Emblem's True Hit distribution be called?)
posted by lucidium at 11:16 AM on February 6, 2017 [3 favorites]
To use XCOM as an example:
XCOM shows you the odds in order to inform your strategy. On the macro level, seeing the chance-to-hit percentage gradually educates you on the conditions you need to have for accurate shots (penalty for shooting at someone in cover, bonus for shooting at someone from a flanking position, etc.). On the micro level, it helps you decide whether you're better off trying a borderline shot or saving your shot for Overwatch. Sometimes you think an enemy has been flanked, but the engine doesn't, because you're one square over from where you'd need to be, so technically the enemy's cover is still blocking your shot a bit. And so on.
But the developers also want you to keep playing the game and not Alt-F4 angrily after your sixth miss in a row. If you flipped a coin 100 times and logged the results, you'd notice that random behavior is prone to a surprising amount of streakiness. Random events don't always feel random. This is one of the reasons why the original X-COM from the DOS days is remembered as being cruel; just one of those outlier streaks can ruin an entire mission.
So I don't think it's a problem for a XCOM to show you one percentage but act on another percentage. The first percentage is still meaningful; it represents the quality of the shot you take. The second percentage just incorporates extra situational factors that wouldn't make sense to show to the user unless you also explained them. If XCOM said “this is ordinarily a 15% chance-to-hit, but right now it's 41% because you've had some unlucky misses lately,” then they'd be opening up that mechanic to manipulation by the player. (Purposefully miss three times in a row on low-quality shots with your rookies, thus guaranteeing that the veteran with the sniper rifle will get a critical hit.)
If they cheated too much, such that players noticed how little their outcomes matched up with the stated percentages, then that itself would frustrate players more than the original randomness Firaxis was trying to guard against.
It seems to me this is only a problem with games that actually show you the odds. If they don't show you the odds in explicit terms, then they're free to monkey around with them behind the scenes without anyone complaining about a violation of trust.
posted by savetheclocktower at 11:35 AM on February 6, 2017 [4 favorites]
XCOM shows you the odds in order to inform your strategy. On the macro level, seeing the chance-to-hit percentage gradually educates you on the conditions you need to have for accurate shots (penalty for shooting at someone in cover, bonus for shooting at someone from a flanking position, etc.). On the micro level, it helps you decide whether you're better off trying a borderline shot or saving your shot for Overwatch. Sometimes you think an enemy has been flanked, but the engine doesn't, because you're one square over from where you'd need to be, so technically the enemy's cover is still blocking your shot a bit. And so on.
But the developers also want you to keep playing the game and not Alt-F4 angrily after your sixth miss in a row. If you flipped a coin 100 times and logged the results, you'd notice that random behavior is prone to a surprising amount of streakiness. Random events don't always feel random. This is one of the reasons why the original X-COM from the DOS days is remembered as being cruel; just one of those outlier streaks can ruin an entire mission.
So I don't think it's a problem for a XCOM to show you one percentage but act on another percentage. The first percentage is still meaningful; it represents the quality of the shot you take. The second percentage just incorporates extra situational factors that wouldn't make sense to show to the user unless you also explained them. If XCOM said “this is ordinarily a 15% chance-to-hit, but right now it's 41% because you've had some unlucky misses lately,” then they'd be opening up that mechanic to manipulation by the player. (Purposefully miss three times in a row on low-quality shots with your rookies, thus guaranteeing that the veteran with the sniper rifle will get a critical hit.)
If they cheated too much, such that players noticed how little their outcomes matched up with the stated percentages, then that itself would frustrate players more than the original randomness Firaxis was trying to guard against.
It seems to me this is only a problem with games that actually show you the odds. If they don't show you the odds in explicit terms, then they're free to monkey around with them behind the scenes without anyone complaining about a violation of trust.
posted by savetheclocktower at 11:35 AM on February 6, 2017 [4 favorites]
Mitheral: "I-Write-Essays: "The actual meaning of saying you have a 75% chance of success breaks down on a single attack. Is that on d100 or 2d10?"
A reported chance of 75% shouldn't be affected by the shape of the curve of absolute values of the die roll."
Yeah, well, the granularity of possible results will have an effect on the die roll, assuming of course, we are simulating a dice rool.
posted by Samizdata at 11:45 AM on February 6, 2017
A reported chance of 75% shouldn't be affected by the shape of the curve of absolute values of the die roll."
Yeah, well, the granularity of possible results will have an effect on the die roll, assuming of course, we are simulating a dice rool.
posted by Samizdata at 11:45 AM on February 6, 2017
XCOM shows you the odds in order to inform your strategy.
It would probably be better to express the bonuses and maluses in terms of some abstract value, call them "pips". Flanking is plus five pips, long range is minus two pips, etc., etc. Then obscure how the shot is actually resolved while ensuring that more pips means the shot is more likely to resolve successfully.
But, then, that little word "likely" is still a problem, because probabilities expressed as likelihoods are only relevant over a large number of repeated trials, while a strategy gamer really only wants to know the result of the specific granular action they're undertaking at that particular moment. There's basically no single, distilled number that could possibly encapsulate this. But randomness, paradoxically, is only exciting if you know the odds, so it's necessary to distill a complex idea down to a single number you can display on the screen so the player knows to curse his foul luck or say a prayer of thanks to the RNG odds when they see the outcome of the action.
One approach that I find fascinating but don't have much personal experience with is board games that use a number-of-successes system. Roll some number of dice where a result greater or lower than X is a "success", then count the number of successes. Either each success represents a packet of damage or else there's a required number of successes to complete the action being performed. This allows the number of dice thrown, success threshold per die, and the number of successes required to all vary independently, modeling different things.
In an XCOM example, the number of dice thrown might represent the weapon's damage potential, the per-die success threshold would represent the soldier's accuracy and positioning relative to the target, and the number-of-successes target represents the enemy's defences, ie. armour, cover, etc. And then you might say that hitting the number-of-successes target deals 1HP damage to the enemy, with every success beyond that doing an additional 1HP of damage.
I think this is a system that's potentially more intuitive, statistically, than what the game currently does, which is to roll a single number for each attack and resolve everything about the attack from that single roll. (Technically, EU/EW would roll for a hit, then make a separate roll for crit chance, then roll separately for damage. XCOM2 uses a single roll for graze-hit-crit, then a separate roll for damage. I don't think the details are all that important for my argument, but I present them here for completeness.) Knowing that each of the six dice you're throwing on this attack has an 80% chance of success gives you a lot more actionable info than knowing that the one die you're throwing on this attack has an 80% chance of success.
posted by tobascodagama at 12:36 PM on February 6, 2017 [1 favorite]
It would probably be better to express the bonuses and maluses in terms of some abstract value, call them "pips". Flanking is plus five pips, long range is minus two pips, etc., etc. Then obscure how the shot is actually resolved while ensuring that more pips means the shot is more likely to resolve successfully.
But, then, that little word "likely" is still a problem, because probabilities expressed as likelihoods are only relevant over a large number of repeated trials, while a strategy gamer really only wants to know the result of the specific granular action they're undertaking at that particular moment. There's basically no single, distilled number that could possibly encapsulate this. But randomness, paradoxically, is only exciting if you know the odds, so it's necessary to distill a complex idea down to a single number you can display on the screen so the player knows to curse his foul luck or say a prayer of thanks to the RNG odds when they see the outcome of the action.
One approach that I find fascinating but don't have much personal experience with is board games that use a number-of-successes system. Roll some number of dice where a result greater or lower than X is a "success", then count the number of successes. Either each success represents a packet of damage or else there's a required number of successes to complete the action being performed. This allows the number of dice thrown, success threshold per die, and the number of successes required to all vary independently, modeling different things.
In an XCOM example, the number of dice thrown might represent the weapon's damage potential, the per-die success threshold would represent the soldier's accuracy and positioning relative to the target, and the number-of-successes target represents the enemy's defences, ie. armour, cover, etc. And then you might say that hitting the number-of-successes target deals 1HP damage to the enemy, with every success beyond that doing an additional 1HP of damage.
I think this is a system that's potentially more intuitive, statistically, than what the game currently does, which is to roll a single number for each attack and resolve everything about the attack from that single roll. (Technically, EU/EW would roll for a hit, then make a separate roll for crit chance, then roll separately for damage. XCOM2 uses a single roll for graze-hit-crit, then a separate roll for damage. I don't think the details are all that important for my argument, but I present them here for completeness.) Knowing that each of the six dice you're throwing on this attack has an 80% chance of success gives you a lot more actionable info than knowing that the one die you're throwing on this attack has an 80% chance of success.
posted by tobascodagama at 12:36 PM on February 6, 2017 [1 favorite]
I love discovering the edges of probability in video games. My wife and I have recently been playing Divinity: Original Sin, a turn-based cooperative melee and magic combat RPG, and the dice rolls are brutally wide.
It's possible to fail to buff yourself with a spell and possible to kill a big monster with a dramatically-damaging critical hit. Like most games, going up against baddies that are much lower level is easy, but still, occasionally, I miss or a spell fails. Going up against bigger critters is very hard, but possible, thanks to the potential of landing a solid hit and getting lucky on the riposte.
It feels like there are no edges on the probability in this game or the edges are far enough out that they're not experienced by players and not heavily controlled by the game creators. I know every video game is a controlled universe within the parameters of the design, but the ability to make a game feel like there's broad swaths of luck outside of the bell curve is more impressive to me than the ability to make a game let me win because it thinks I expect it. This feels like a game while more sculpted games feel like a hallway.
posted by Revvy at 1:03 PM on February 6, 2017 [1 favorite]
It's possible to fail to buff yourself with a spell and possible to kill a big monster with a dramatically-damaging critical hit. Like most games, going up against baddies that are much lower level is easy, but still, occasionally, I miss or a spell fails. Going up against bigger critters is very hard, but possible, thanks to the potential of landing a solid hit and getting lucky on the riposte.
It feels like there are no edges on the probability in this game or the edges are far enough out that they're not experienced by players and not heavily controlled by the game creators. I know every video game is a controlled universe within the parameters of the design, but the ability to make a game feel like there's broad swaths of luck outside of the bell curve is more impressive to me than the ability to make a game let me win because it thinks I expect it. This feels like a game while more sculpted games feel like a hallway.
posted by Revvy at 1:03 PM on February 6, 2017 [1 favorite]
It would probably be better to express the bonuses and maluses in terms of some abstract value, call them "pips".
You're just replacing numbers with symbols. Players will still feel cheated if they are told to expect one thing (high chance of success) but get something else. How could that NOT feel like a con? I don't think you can interface design your way out of the fundamental issue--you tell them A but they get B.
I think you have to create the game around that truth--get to a point that it's either fun to get B instead of A, or alternately make the fun arise somehow out of everyone having perfect info.
Which are both pretty tough asks... like, counting cards in blackjack isn't fun, it's a job...
posted by danny the boy at 1:38 PM on February 6, 2017 [1 favorite]
You're just replacing numbers with symbols. Players will still feel cheated if they are told to expect one thing (high chance of success) but get something else. How could that NOT feel like a con? I don't think you can interface design your way out of the fundamental issue--you tell them A but they get B.
I think you have to create the game around that truth--get to a point that it's either fun to get B instead of A, or alternately make the fun arise somehow out of everyone having perfect info.
Which are both pretty tough asks... like, counting cards in blackjack isn't fun, it's a job...
posted by danny the boy at 1:38 PM on February 6, 2017 [1 favorite]
That one character had 2 different targets, and 4 different directionalities to attack from. The percent chance to hit changed with each different target, and each different direction. But all eight of those combinations were misses. That means my actual percent chance to hit was 0%, because my character was in a miss round. Just tell me that, if that's how the system works, and don't give me a completely fake percent chance.
Well, I'm willing to bet that the way the system actually works is something like this:
1. For each action the player may take, calculate a chance to succeed S(A), 0-100%, based on the game's rules. Display those to the player.
2. Let the player choose an action.
3. Draw a pseudorandom number R, 0-100.
4. If R ≤ S(A), the action succeeds, otherwise it fails.
When you reload an emulator's saved state, the pseudorandom number generator is in the same state every time, which means R is the same every time. Reloading and changing which action you choose varies only S(A). So if you happen to be "unlucky" in that R is, say 100, then you will indeed find that your action fails no matter which action you choose after reloading the save state; conversely if you're "lucky" in that R is 0, then you'll succeed no matter what.
So in one sense you're right that in your test, your actual chance to hit was 0%, given the state of the pseudorandom number generator. But all video games with an element of chance rely on PRNGs (unless they do something fancy to pull in external sources of entropy), so the only meaningful sense in which a video game can report a probability is one which is marginalized over PRNG states. The probability is "fake" in that the outcome actually is predetermined by the PRNG algorithm, but "real" in that if you don't know (or artificially fix) the state of the PRNG, multiple encounters with the same reported probability will turn out as described. This "fakeness" is exactly what the "pseudo-" in "pseudorandom" is meant to describe.
posted by biogeo at 1:48 PM on February 6, 2017 [4 favorites]
Well, I'm willing to bet that the way the system actually works is something like this:
1. For each action the player may take, calculate a chance to succeed S(A), 0-100%, based on the game's rules. Display those to the player.
2. Let the player choose an action.
3. Draw a pseudorandom number R, 0-100.
4. If R ≤ S(A), the action succeeds, otherwise it fails.
When you reload an emulator's saved state, the pseudorandom number generator is in the same state every time, which means R is the same every time. Reloading and changing which action you choose varies only S(A). So if you happen to be "unlucky" in that R is, say 100, then you will indeed find that your action fails no matter which action you choose after reloading the save state; conversely if you're "lucky" in that R is 0, then you'll succeed no matter what.
So in one sense you're right that in your test, your actual chance to hit was 0%, given the state of the pseudorandom number generator. But all video games with an element of chance rely on PRNGs (unless they do something fancy to pull in external sources of entropy), so the only meaningful sense in which a video game can report a probability is one which is marginalized over PRNG states. The probability is "fake" in that the outcome actually is predetermined by the PRNG algorithm, but "real" in that if you don't know (or artificially fix) the state of the PRNG, multiple encounters with the same reported probability will turn out as described. This "fakeness" is exactly what the "pseudo-" in "pseudorandom" is meant to describe.
posted by biogeo at 1:48 PM on February 6, 2017 [4 favorites]
The RNG in Fallen London actually sends you a Christmas present each year to make up for the prior year's abuses.
posted by praemunire at 2:33 PM on February 6, 2017 [4 favorites]
posted by praemunire at 2:33 PM on February 6, 2017 [4 favorites]
Darkest Dungeon is an example of a game that seems to pull no punches when it comes to chances and randomness. Seriously, no other game has made me rage quit harder at its "unfairness". And yet I keep playing it.
posted by Doleful Creature at 4:17 PM on February 6, 2017 [1 favorite]
posted by Doleful Creature at 4:17 PM on February 6, 2017 [1 favorite]
In the Fallout games, "Luck" is literally an attribute that you can level up, and results in all sorts of different benefits in-game. More damage, bonus caps, better rolls, all kinds of goodies. But it's very behind-the-scenes, unlike the other attributes, which have tangible effects. Like, upgrade your Strength, you can carry more loot. Bump your Charisma, more settlers.
So what I'm basically saying is I have nothing to add to this conversation except I'm playing Fallout 4 finally and really enjoying it!
posted by turbid dahlia at 5:11 PM on February 6, 2017 [3 favorites]
So what I'm basically saying is I have nothing to add to this conversation except I'm playing Fallout 4 finally and really enjoying it!
posted by turbid dahlia at 5:11 PM on February 6, 2017 [3 favorites]
Luck is an actual attribute in Nethack too but it functions more like Karma. Your luck score is hidden but somewhat query-able. And then some things are only possible if your luck is either positive or negative. And some things are easier if you have positive luck. And as part of Nethack's theory of including everything your luck is dependent on the phase of the moon (Full moon = more luck) and special days of the week (Friday the 13th = less luck)
posted by Mitheral at 6:07 PM on February 6, 2017 [2 favorites]
posted by Mitheral at 6:07 PM on February 6, 2017 [2 favorites]
My favorite strategy game series is far and away Advance Wars, for the Gameboy Advance and Nintendo DS. I really love the battle system, which is something like rock/paper/scissors meets a sophisticated form of chess, with some very light, non-frustrating elements of chance incorporated. The percentage tells you what damage you'll reliably do - 80% damage means you'll do at least 8 out of 10 damage, guaranteed, though occasionally it'll be 9 or 10 out of 10, destroying the unit. The same goes for your opponents attacking you. The way certainty and slight chance come together is a shining example of thoughtful game design, complex yet simply presented, reliable yet open to surprise. It makes the game fiendishly addictive.
Playing a game like that will convince you that accurately-modeled systems of probability aren't some pure ideal, and instead are rather likely to lead to frustrated players and poor game design. A player throwing a controller in frustration isn't always the player showing weakness or entitlement - more likely it's game designers failing to make something that balances fun, depth, accessibility, and fairness. Not an easy feat.
This explains more about the Advance Wars system:
Playing a game like that will convince you that accurately-modeled systems of probability aren't some pure ideal, and instead are rather likely to lead to frustrated players and poor game design. A player throwing a controller in frustration isn't always the player showing weakness or entitlement - more likely it's game designers failing to make something that balances fun, depth, accessibility, and fairness. Not an easy feat.
This explains more about the Advance Wars system:
Another example of wise design in AW is the way that random chance is employed. No randomness at all would have simplified production and testing and would make the game more chess-like. This would not only increase fairness by guaranteeing that each attack with identical forces and situations would be resolved the same, but would make the process of game development simpler and therefore cheaper. Randomness in a game means that the testing and debugging process will be more involved and expensive. Just because something works fine one time doesn’t mean it will do so the next when different random numbers come up, so there is a temptation to eliminate randomness to shorten the testing process. But the price is lack of excitement and realism. With some randomness, the possibility of a lucky break saving the day, or an unlucky one costing a valuable unit can make a nearly even match into a thrilling back-and-forth epic event. AW gives you the basic “odds” of an individual battle, showing essentially how many points of damage a given attack will do – but it can add a few percent to that total randomly (apparently with a bell curve), so an attack that usually will only severely damage an enemy unit most of the time may destroy it occasionally, opening a hole in enemy lines. The chance of extra damage being incurred goes up as multiple attacks are made on the same unit in the same turn, providing a tactical advantage as well. These things open up possibilities that add layers of complexity to expert play without burdening the beginner with complicated calculations and record keeping. As an example of a subtle design point that would be easy to miss, the attacks ALWAYS do the minimum stated, but may do more. A less experienced design would have the randomness sometimes add, and sometimes take away, making critical attacks that are on the cusp sometimes fail to destroy the enemy by a fraction of a point. That might be more realistic, but would be intensely frustrating for the player.posted by naju at 6:09 PM on February 6, 2017 [6 favorites]
I suspect that presentation matters almost as much as actual mechanics. For example, these are all the same distribution (assuming an inclusive rand that can generate the end values):
rand(8, 12)
8 + rand(4)
10 +/- rand(2)
12 - rand(4)
but psychologically they all read differently. Probably the tendency is to implicitly assume that the largest number is the baseline, so "8 to 12" gets interpreted as "usually 12", and "10 +/- 2" gets interpreted as "usually 10".
posted by Pyry at 7:08 PM on February 6, 2017 [2 favorites]
rand(8, 12)
8 + rand(4)
10 +/- rand(2)
12 - rand(4)
but psychologically they all read differently. Probably the tendency is to implicitly assume that the largest number is the baseline, so "8 to 12" gets interpreted as "usually 12", and "10 +/- 2" gets interpreted as "usually 10".
posted by Pyry at 7:08 PM on February 6, 2017 [2 favorites]
(As an aside, what would Fire Emblem's True Hit distribution be called?)
It's basically 2d100. This is what the distribution of averages looks like.
posted by J.K. Seazer at 8:25 PM on February 6, 2017
It's basically 2d100. This is what the distribution of averages looks like.
posted by J.K. Seazer at 8:25 PM on February 6, 2017
Thinking about the Civ example more, and the visceral negative reaction some people have to it, I have to say it's nice framing by Sid Meier & the author of the article. Instead of saying "gamers got irate because single winner-take all die rolls was a bad mechanic" (which makes it Sid's fault) they mention cognitive biases (so the players fault.)
Basically the rule that is in the game for the scenario described is: If you have three units against one you are guaranteed a victory but might lose up two units. This is a valid, random rule. I did a lot of old fashioned hex-based games back in my teens and I'm actually getting a little nostalgic remember these battle tables, where you'd look at the "3:1 ratio" column and see that you rolled a 1 which sucks, but that's coded "DxA2" or something. Then you'd at the key you'd see "defender eliminated; attacker loses twice as many units as the defender removes."
What happened is Sid came up with a very simple mechanic, which is easy to understand but didn't play well, because it didn't reward numbers enough to feel realistic. With the straight "high roll wins" type approach, in the scenario mentioned you could outnumber your opponent six to one and still lose 10% of the battles--not just lose, but be wiped out while causing no damage to the defender. Basically numerical superiority with roughly equal quality troops does not help the way anyone with a passing interest in history knows it should.* It's a horrible simulation for many reason and it's why the wonkier war games always had battle tables, and it's why in play testing players hate it and they need tweak it until it's a bit closer to the traditional war game. Also why they dropped the overly simple mechanic in later Civ games and had unit health and other things to make the distribution seem more realistic.
You can still complain about the transparency issue--the game designers aren't making the algorithm clear--but that's on the designer too, not the stupid player. I think it's reasonable for a player to grade a game on whether it does a reasonable job "feeling" like it's real, without opening a spreadsheet and going the min/max, stats only or moneyball style of strategizing.
As an aside I'm 100% sure one of the later Civ games, maybe Civ IV, implemented something so if you reloaded some number of the next "rolls" would have the same outcome. Presumably saving a seed value separately for each unit or something, I never quite managed to reverse engineer it but I very much respect that.
*Please repress the urge mention Agincourt, or Belisarius defeating a rampaging Avar horde with 300 geriatric veterans**. Not only are they rare but they're different situations than what's modeled here.
**My personal favorite incident to mention, by the way. I read about it in a Robert Graves novel and it blew my mind when I found out it was real. You could look it up. Although if you do you'll find I'm lying about "geriatric".
posted by mark k at 9:21 PM on February 6, 2017 [4 favorites]
Basically the rule that is in the game for the scenario described is: If you have three units against one you are guaranteed a victory but might lose up two units. This is a valid, random rule. I did a lot of old fashioned hex-based games back in my teens and I'm actually getting a little nostalgic remember these battle tables, where you'd look at the "3:1 ratio" column and see that you rolled a 1 which sucks, but that's coded "DxA2" or something. Then you'd at the key you'd see "defender eliminated; attacker loses twice as many units as the defender removes."
What happened is Sid came up with a very simple mechanic, which is easy to understand but didn't play well, because it didn't reward numbers enough to feel realistic. With the straight "high roll wins" type approach, in the scenario mentioned you could outnumber your opponent six to one and still lose 10% of the battles--not just lose, but be wiped out while causing no damage to the defender. Basically numerical superiority with roughly equal quality troops does not help the way anyone with a passing interest in history knows it should.* It's a horrible simulation for many reason and it's why the wonkier war games always had battle tables, and it's why in play testing players hate it and they need tweak it until it's a bit closer to the traditional war game. Also why they dropped the overly simple mechanic in later Civ games and had unit health and other things to make the distribution seem more realistic.
You can still complain about the transparency issue--the game designers aren't making the algorithm clear--but that's on the designer too, not the stupid player. I think it's reasonable for a player to grade a game on whether it does a reasonable job "feeling" like it's real, without opening a spreadsheet and going the min/max, stats only or moneyball style of strategizing.
As an aside I'm 100% sure one of the later Civ games, maybe Civ IV, implemented something so if you reloaded some number of the next "rolls" would have the same outcome. Presumably saving a seed value separately for each unit or something, I never quite managed to reverse engineer it but I very much respect that.
*Please repress the urge mention Agincourt, or Belisarius defeating a rampaging Avar horde with 300 geriatric veterans**. Not only are they rare but they're different situations than what's modeled here.
**My personal favorite incident to mention, by the way. I read about it in a Robert Graves novel and it blew my mind when I found out it was real. You could look it up. Although if you do you'll find I'm lying about "geriatric".
posted by mark k at 9:21 PM on February 6, 2017 [4 favorites]
I wonder if part of the problem is that computers are pretty bad at generating random numbers: you can get a longer stretch of success or failures than even a reasonable person would expect. Sure, developers COULD try to improve whatever algorithm they use, but I can see the appeal of just rigging the system to make everyone happy.
posted by wheloc at 5:01 AM on February 7, 2017
posted by wheloc at 5:01 AM on February 7, 2017
True randomness is incredibly streaky, though! I don't think substituting a PRNG for a cryptographically-strong true RNG would actually help anything. Like mark k, I feel that randomness is often a crutch. A fun crutch, if used correctly, because it evens out differences in skill that would otherwise make PvP games almost completely pre-determined.
posted by tobascodagama at 5:29 AM on February 7, 2017
posted by tobascodagama at 5:29 AM on February 7, 2017
computers are pretty bad at generating random numbers
Humans are pretty bad at consuming random numbers: we're always relentlessly looking for patterns even when there truly are none.
posted by We had a deal, Kyle at 7:45 AM on February 7, 2017 [1 favorite]
Humans are pretty bad at consuming random numbers: we're always relentlessly looking for patterns even when there truly are none.
posted by We had a deal, Kyle at 7:45 AM on February 7, 2017 [1 favorite]
Humans didn't learn to be bad at evaluating risk from playing Civ III, we are bad at it because the kind of risk our evolution optimized us for is fundamentally different than what we encounter today.
Blaming video games is... misguided.
Actually, I completely disagree. I personally learned so much about risk and probability from playing a hundred hours or so of Final Fantasy Tactics Advance. The gameplay is centered around strategy and tactics, and every action you take carries a probability of success or failure, and the stakes were very clear. After playing for just a short period of time, I quickly learned that an action with a 70% success rate generally wasn't worth taking unless the penalty was very slight. An 80% risk was only worth it if there was no other path to victory. Generally speaking, if a particular failure has any kind of moderate to severe consequences, I felt like I could only count on actions with 93% or better success rate. Constantly making bets on 70% actions would quickly lead to losing battles and having to replay long sections of the game.
I was not at all surprised that Trump won when the polls were showing he was 80% to lose. Thanks to this game, I had an intuitive understanding that you should never, ever count on 80% odds when there is a lot at stake. I think the outcome of the election might have been different if more people had understood this simple fact.
And I think video games actually have a lot of potential to teach things like this, and it's sad when game designers deliberately mislead players, because it's a such a wasted opportunity.
posted by zixyer at 1:02 PM on February 7, 2017 [5 favorites]
Blaming video games is... misguided.
Actually, I completely disagree. I personally learned so much about risk and probability from playing a hundred hours or so of Final Fantasy Tactics Advance. The gameplay is centered around strategy and tactics, and every action you take carries a probability of success or failure, and the stakes were very clear. After playing for just a short period of time, I quickly learned that an action with a 70% success rate generally wasn't worth taking unless the penalty was very slight. An 80% risk was only worth it if there was no other path to victory. Generally speaking, if a particular failure has any kind of moderate to severe consequences, I felt like I could only count on actions with 93% or better success rate. Constantly making bets on 70% actions would quickly lead to losing battles and having to replay long sections of the game.
I was not at all surprised that Trump won when the polls were showing he was 80% to lose. Thanks to this game, I had an intuitive understanding that you should never, ever count on 80% odds when there is a lot at stake. I think the outcome of the election might have been different if more people had understood this simple fact.
And I think video games actually have a lot of potential to teach things like this, and it's sad when game designers deliberately mislead players, because it's a such a wasted opportunity.
posted by zixyer at 1:02 PM on February 7, 2017 [5 favorites]
What zixyer said. I have a friend who complains that even Nate Silver misled us about the election, when in fact he was the one who most understood what the odds he gave meant.
On the subject of Advance Wars, I just have to add that one of the COs, Nell, had as her special power Luck, she would randomly do a bit of extra damage, which could increase drastically during her CO and SCO power. "Lucky!" It made her one of the better commanders.
Fire Emblem is an interesting case, BTW. In the GBA games at least, if you attacked an opponent, then shut the game off right after seeing the result of the attack animation (say, because your character died), then reloaded, you'd be treated to the sight of your action before happening as soon as you reloaded, with the same result. This happened because the game saved the current state of the random seed in the save file!
At AGDQ this year, there was a great manipulation of the Pokemon Emerald RNG. As perhaps befitting games where randomness accounts for so much, all the Pokemon games short of the recently-released Sun and Moon have had their generators studied. It was discovered that Emerald, oddly unique among the games of its generation, initialized its RNG to 0 upon game startup. From there, there are limited opportunities to randomize the seed. Speedrunners found out that, if you saved the game right before the choice of starter then reloaded, the game would "spin" the RNG, just generated numbers to add uncertainty, in the frames where you selected your starter.
One of those frames, if hit just as you were picking Mudkip, would produce of Mudkip with very high "IVs," that is, innate potential. They managed to generate that Mudkip (named "Waluigi"), and then rode the strength of that one Pokemon for nearly the entire game. It was only then they managed to catch a legendary with the Master Ball that it ceased to be their main. They finished the whole game in less than three hours.
posted by JHarris at 8:34 PM on February 7, 2017 [3 favorites]
On the subject of Advance Wars, I just have to add that one of the COs, Nell, had as her special power Luck, she would randomly do a bit of extra damage, which could increase drastically during her CO and SCO power. "Lucky!" It made her one of the better commanders.
Fire Emblem is an interesting case, BTW. In the GBA games at least, if you attacked an opponent, then shut the game off right after seeing the result of the attack animation (say, because your character died), then reloaded, you'd be treated to the sight of your action before happening as soon as you reloaded, with the same result. This happened because the game saved the current state of the random seed in the save file!
At AGDQ this year, there was a great manipulation of the Pokemon Emerald RNG. As perhaps befitting games where randomness accounts for so much, all the Pokemon games short of the recently-released Sun and Moon have had their generators studied. It was discovered that Emerald, oddly unique among the games of its generation, initialized its RNG to 0 upon game startup. From there, there are limited opportunities to randomize the seed. Speedrunners found out that, if you saved the game right before the choice of starter then reloaded, the game would "spin" the RNG, just generated numbers to add uncertainty, in the frames where you selected your starter.
One of those frames, if hit just as you were picking Mudkip, would produce of Mudkip with very high "IVs," that is, innate potential. They managed to generate that Mudkip (named "Waluigi"), and then rode the strength of that one Pokemon for nearly the entire game. It was only then they managed to catch a legendary with the Master Ball that it ceased to be their main. They finished the whole game in less than three hours.
posted by JHarris at 8:34 PM on February 7, 2017 [3 favorites]
In the Fallout games, "Luck" is literally an attribute that you can level up, and results in all sorts of different benefits in-game. More damage, bonus caps, better rolls, all kinds of goodies. But it's very behind-the-scenes, unlike the other attributes, which have tangible effects
In Fallout 2 a luck of ten (max) would consistently land critical hits per turn, so that was the only way to play a melee character without making a beeline for end game equipment (since you get the weapon that will carry you through the midgame in the first area of the game).
posted by ersatz at 11:17 PM on February 7, 2017
In Fallout 2 a luck of ten (max) would consistently land critical hits per turn, so that was the only way to play a melee character without making a beeline for end game equipment (since you get the weapon that will carry you through the midgame in the first area of the game).
posted by ersatz at 11:17 PM on February 7, 2017
I suppose there is often a temptation to inflate one's aesthetic preferences to a moral stance.
I think though the problem is when the presentation is misleading. When it says something has an 80% chance but that doesn't really mean an 80% chance.
A lot of implementations of Tetris use a sort of virtual deck behind the scenes. Instead of picking, for each new block, one of seven shapes randomly, imagine you have seven cards that it shuffles and deals. So that guarantees you'll get an even mix of shapes, and you won't get a whole bunch of the same shape in a row. But I think that bothers me less because the game never really gives you any indication how the pieces are being picked.
posted by RobotHero at 10:22 AM on February 10, 2017 [1 favorite]
I think though the problem is when the presentation is misleading. When it says something has an 80% chance but that doesn't really mean an 80% chance.
A lot of implementations of Tetris use a sort of virtual deck behind the scenes. Instead of picking, for each new block, one of seven shapes randomly, imagine you have seven cards that it shuffles and deals. So that guarantees you'll get an even mix of shapes, and you won't get a whole bunch of the same shape in a row. But I think that bothers me less because the game never really gives you any indication how the pieces are being picked.
posted by RobotHero at 10:22 AM on February 10, 2017 [1 favorite]
Robothero, the "seven-card deck" approach is now actually mandated for all licensed implementations of Tetris as a requisite to be allowed to even use the brand name
posted by DoctorFedora at 4:13 PM on February 10, 2017
posted by DoctorFedora at 4:13 PM on February 10, 2017
It makes sense, really. Using a deck of cards (virtually) means that while the output is random, there's hard limits to how big of a streak can occur, which obviously there are no limits to with (virtual) dice.
posted by tocts at 7:23 AM on February 11, 2017
posted by tocts at 7:23 AM on February 11, 2017
I didn't know how consistently it was used. But I was reminded of it, because it seems motivated by the same sort of idea that the player could be annoyed by something rare but possible if they used direct odds.
One common thing is a player will set up a four-line score which just requires they wait for a long piece. Using the deck randomness, there's a guaranteed upper limit on how long they'll have to wait to get one.
If it were just a flat 1 out of 7 chance the next one will be a long piece, then very long waits will be rare but possible. And if you play enough Tetris, they would come up.
So it's a similar scenario to missing two 90% shots in a row for example. With flat odds, that should happen 1% of the time.
posted by RobotHero at 8:26 AM on February 11, 2017 [1 favorite]
One common thing is a player will set up a four-line score which just requires they wait for a long piece. Using the deck randomness, there's a guaranteed upper limit on how long they'll have to wait to get one.
If it were just a flat 1 out of 7 chance the next one will be a long piece, then very long waits will be rare but possible. And if you play enough Tetris, they would come up.
So it's a similar scenario to missing two 90% shots in a row for example. With flat odds, that should happen 1% of the time.
posted by RobotHero at 8:26 AM on February 11, 2017 [1 favorite]
The "piece bag" in Tetris is actually a good example of how messing with randomness can fundamentally change the game.
An essential aspect of Old Tetris is the risk/reward aspect of building the stack with just a single block width hole the right size for the Bar to fill, ready to make a Tetris. In classic versions, there was no guarantee that a Bar piece would show up in time, or indeed ever show up the whole game. This uncertainty meant that sometimes, you would have to eventually cut your losses and start clearing lines for plain survival. This aspect is deemphasized now that Bars are mandated to appear once in a while.
posted by JHarris at 8:46 AM on February 11, 2017
An essential aspect of Old Tetris is the risk/reward aspect of building the stack with just a single block width hole the right size for the Bar to fill, ready to make a Tetris. In classic versions, there was no guarantee that a Bar piece would show up in time, or indeed ever show up the whole game. This uncertainty meant that sometimes, you would have to eventually cut your losses and start clearing lines for plain survival. This aspect is deemphasized now that Bars are mandated to appear once in a while.
posted by JHarris at 8:46 AM on February 11, 2017
It makes sense, really. Using a deck of cards (virtually) means that while the output is random, there's hard limits to how big of a streak can occur, which obviously there are no limits to with (virtual) dice.
Cards instead of dice are also one of the big changes in designer board-games from American style family board games for that very reason.
posted by codacorolla at 10:21 AM on February 11, 2017 [2 favorites]
Cards instead of dice are also one of the big changes in designer board-games from American style family board games for that very reason.
posted by codacorolla at 10:21 AM on February 11, 2017 [2 favorites]
We've thought about using cards to play Settlers of Catan for that very reason.
posted by JHarris at 12:20 PM on February 11, 2017
posted by JHarris at 12:20 PM on February 11, 2017
They sell an official expansion for that, though it also adds "events" as well, which may or may not be to your tastes. The die results in the deck are just the 36 results from a 2d6 distribution, though there's a special card put in after shuffling so you reshuffle when there's 5 cards left (thus you can never be 100% sure via card counting).
(independent tabletop game design and publishing is a minor (unprofitable) side business for me; I love dice and cards both, each with their own strengths/weaknesses)
posted by tocts at 12:32 PM on February 11, 2017 [3 favorites]
(independent tabletop game design and publishing is a minor (unprofitable) side business for me; I love dice and cards both, each with their own strengths/weaknesses)
posted by tocts at 12:32 PM on February 11, 2017 [3 favorites]
My own experience with game design is that cards and dice both lend different feels to a game. It depends on what you want your players to feel.
posted by codacorolla at 5:02 PM on February 11, 2017
posted by codacorolla at 5:02 PM on February 11, 2017
« Older An interesting week of protests in Romania | NSFW \__ Newer »
This thread has been archived and is closed to new comments
"Extensive play-testing revealed that a player who was told that he had a 33 percent chance of success in a battle but then failed to defeat his opponent three times in a row would become irate and incredulous. (In Civilization, you can replay the same battle over and over until you win, albeit incurring costs with every loss.) So Meier altered the game to more closely match human cognitive biases; if your odds of winning a battle were 1 in 3, the game guaranteed that you’d win on the third attempt—a misrepresentation of true probability that nevertheless gave the illusion of fairness."
This bothers me a lot. People really need to get a better understanding of risk.
posted by amtho at 10:20 PM on February 5, 2017 [70 favorites]