Yeah, if you do this you’re undercutting the player’s attempt to understand how the game works. I guess that’s fine if you’re making Mario Kart and you’ve got an overriding goal. Otherwise be proud of the mechanics you actually used and present them clearly.
The problem is that 90% won’t work out to 90 hits every 100 times if it’s random.
More importantly, a 10% chance will repeatedly hit or miss a successful number over a larger number of turns than a 50% chance will.
For example, I just randomly generated a very long string of numbers between 1-100. There should be a 10% chance of getting a number between 10-20. But there were lots of sequences with about twenty rolls until I hit it, and lots of sequences where there were five rolls until it I hit it.
Treating it as a draw of cards as @pelle suggested doesn’t necessarily solve the problem. Fudging the numbers in the player’s favor kind of works.
There’s another way. It’s fairly common in tactics RPGs to represent odds as a range. So that supposedly 10% chance would be better represented as 5-20%. Sometimes, the 50% odds would just be represented as 50% because the range there would be too wide to be useful information for the player.
(I’m not a math person, so hopefully I didn’t overlook something.)
1 Like
That Mario Kart is famously fudging the odds of things is one thing, but that the designers of games like UFO Enemy Unknown and Civilization are admitting that they are faking the numbers is what I find really disturbing. I wish at least games could have an option to disable faked randomness. Or at LEAST be up front about doing it.
4 Likes
Right, that’s not a great system. You probably shouldn’t use it. You could pretend to use it, but in actually “fudge” it. But since you did, in fact, implement a different system, you should just say what it is you did.
Sure, how to convey your system to the player is another question. There may be some use in providing a summary, and it may include percentages. For instance, Brogue will say,
“The kobold has a 50% chance to hit you, typically hits for 8% of your current health, and at worst, could defeat you in 8 hits.”
Random chance is confusing, and presenting as a percentage probably seems friendlier, since people believe they understand percentages reasonably well. But you’re not helping anybody understand how to play your game by misrepresenting how it functions.
3 Likes
But since you did, in fact, implement a different system, you should just say what it is you did.
I’m not sure what you mean.
“The kobold has a 50% chance to hit you, typically hits for 8% of your current health, and at worst, could defeat you in 8 hits.”
The 50% chance is the only thing that’s relevant here. Everything else in the statement doesn’t convey how the randomness works, so it doesn’t solve the problem of helping the player estimate how often the hit/miss will happen.
It’s useful information in the game’s context, but it doesn’t solve the problem.
1 Like
I mean this is game design doublethink:
- Game designer selects die rolls as a method for determining success.
- Designer discovers that die rolls are not fun.
- Designer instead designs a fun system.
- Rather than appreciating the system they have created on its own merits, designer believes they are just “fudging” die rolls, which are the “real” system in use.
- Designer asks the world to participate in the delusion that game uses a die-roll system to determine success.
It’s just an example of a game interpreting the odds a bit to explain them to the player.
2 Likes
If you mean that changing the way percentage-based odds are visually represented (ie. either as a range of percentages, or as numbers fudged in the player’s favor) changes the entire system… I disagree. The same random dice roll is beneath each of those representations.
If you mean that all of the representations are incorrect or inaccurate in some way, I guess I’d agree. But they’re meant to be simplifications, and I prefer some simplifications to others.
1 Like
I think this was about “if you come up with a system, but then discover that players don’t like it and replace it with a different system, don’t pretend that you’re still using the original system.” So if you introduce “luck mitigation” or similar by ensuring a fixed distribution of die rolls (e.g. many games where each attack has a 5% crit chance will force roughly every twentieth hit to be a crit even though that’s not remotely how binomial probability works), don’t claim that every action is still resolved by an independent random roll.
5 Likes
So, there’s something interesting about this psychologically. Humans tend to remember negative events better than positive ones by a factor of 4 if I recall correctly. Now, I’m not a mathematician, nor a psychologist, but an argument can be made that 96% chance of success is experienced more like 88%. The 4%, when it triggers feels more like it happens 12% of the time.
Now, what I just said is in no way scientific, and I’m quite sure that 50% is experienced differently as well because you’re mentally more prepared for failure, but this means an argument could be made for fudging rolls in the player’s advantage. If only to make the experience match the representation better.
Are you sure about UFO? It seems that the mechanics of the original XCOM/UFO are pretty well documented, and as far as I can tell it does not “cheat”: https://www.ufopaedia.org/index.php?title=Accuracy_formula
The 2016 XCOM 2 is known to fudge the percentages on lower difficulty levels. The only way to “switch that off” is to play on the highest difficulty.
EDIT: Somebody actually tested the accuracy of the percentages in the 2012 XCOM: EU, and it also seems accurate.
1 Like
“Scientists have calculated that the chances of something so patently absurd actually existing are millions to one. But magicians have calculated that million-to-one chances crop up nine times out of ten.”
― Terry Pratchett, Mort
3 Likes
I’d instead make the argument that probabilities are not a great way to express to the player what may happen.
What I like about Brogue’s phrasing is that it tells you that the monster “typically” hits for a certain amount of damage, and also explains that “at worst, it could defeat you in 8 hits”. “At worst” is a good, intuitive description.
What’s not so great is to say it “has a 50% chance to hit you”. Maybe a more usable description would be something like: “The kobold has a 50% chance to hit you, which means it may well hit you 8 times within 13 turns”. “May well” indicating a 30% chance in this instance, implying an outcome which is not to be expected, but which we should be ready for. If we were talking about success of a player action, we might use phrasing like “you are likely to miss at least three times in a row”. I’d argue this is the proper place to account for the psychological bias you’ve described, not by sneakily misrepresenting the odds.
I would also argue that, ethically, you should not be actively teaching the player to misinterpret probabilities to a greater extent than they are inherently prone to. Probability is an important mathematical tool which is used to communicate life-and-death information. If you choose to communicate in terms of independent random events, you have a responsibility to at least make sure you’re actually describing independent random events. At best you’d use this teachable moment to improve the player’s intuition for probability. It’s one thing for a human GM to fudge a roll. Everybody knows this is a possibility. It’s another for a computer game to systematically “fudge” rolls while misleading people to think it isn’t. I think players expect fictional information in games, but not fictional mathematics, so they will apply the understanding games have given them to real-life situations.
4 Likes
I’ll be honest, it kind of annoys me how often video game probabilities are given in percentages with how common it is for games to be coded so the probabilities are of the form n-in-256 and often 1-in-2^n. Granted, I’m also a number nerd, have a degree in computer science, and know the default rand function in c++ can’t really handle probabilities that aren’t m-in-2^n exactly, and know people generally don’t understand what random actually means, to the point uniform, non-random data can look intuitively more random than clumpy, truly random data, so maybe rounding the raw probability to the nearest 5% is a reasonable compromise between accuracy and making things comprehensible to non-mathsy players.
I do agree with the advice to GMs to only call for a roll when it matters, though admittedly, that only works with there being a GM to make that judgment, something that is tricky to program into a game, though perhaps the way most video games hide when RNG is being used and don’t explain the nitty gritty of their mechanics’ underlying maths is actually beneficial to the average player’s experience.
If a game does expose it’s inner workings to the player, I do agree they should be honest about them, and unless it’s explicitly a game about teaching probabilities, I don’t think the burden should be on the game devs to correct any misunderstandings a player might have about how probabilities work.
Though yeah, presenting the player a probability for a hit or miss that is just a flat percentage versus a damage estimate where there is a random value in the damage formula adds another wrinkle, especially if the distribution of different damage outputs isn’t flat.
1 Like
While it’s technically true that rand() % 100 + 1 isn’t quite uniform, aren’t the odds of the non-uniformity rearing its head on the order of 1 : 2^57 against? (Since 100/2^64 is approximately 2^7/2^64 = 2^57.)
Which, for non-CS people, is approximately a 1 in 100,000,000,000,000,000 chance. For anything except cryptography, I don’t begrudge developers not worrying about something that rare!
1 Like
So, just for my further understanding (and potential integration in my current project), instead of presenting the flat percentage, we instead present the percentage, and then some boilerplate description which would prime the player on how to understand this chance? Or am I misunderstanding?
I keep wondering why people expect to see explicit probabilities in narrative games.
If the PC has been established as “a wimp” during the exposition and is facing off against the school bully, who has been described as “rather burly”, the player should be able to asses the advisability of trying to knock down the bully as opposed to just handing over their lunch money.
These very brief descriptions would suffice, at least if your narrative is consistent.
3 Likes
I feel like Choice of Games had a great article on this, but of course I can’t find it now…
EDIT: Aha! This one!
But in my experience, the issue is that it’s very easy for the author (who knows exactly how the subsequent check is going to work) to think that they’ve made it perfectly obvious what’s going to be tested, when to the player it’s very far from obvious. In a game with the classic six D&D ability scores, will choosing to “get up in that guy’s face and force him to back down” test my Strength or my Charisma? The author knows whether the subsequent narration is about social or physical intimidation, but the player doesn’t.
2 Likes
That’s the choice of games style. It is very mechanical and deliberate. Risk vs. reward. Trade one stat against the other. It’s probably just me but I’m not convinced making the numbers game explicit is the best fit for a story game.
To me it feels the less tangible the stats in your game are the better the narrative. Phillip Marlowe knows his reputation with prospective clients will suffer if he discloses some piece of evidence to the DA, but then again he won’t have to spend the night in a cell (and can get back to solving the case much earlier). He surely can’t put numbers on this and in fact, checking for abilities and weighting the costs and benefits are exactly what the player might want to think about for themselves.
Differences in taste, I know.
1 Like
The trick there is, the player can’t weigh the costs and benefits if they’re not given all the necessary information. You can do that explicitly or implicitly, but explicitly is a lot easier; if the player doesn’t have enough information to make an informed choice, then the choices don’t really matter.
2 Likes
I don’t think anyone here disagrees with this. Moving on from this point I remember some GM advice from long ago: don’t let the players roll the dice in situations where you care about the outcome. You don’t actually want the player to defeat their nemesis in act one by ridiculous luck because that would ruin a good life-long vendetta campaign. I’ve never played a Choice of Games title where this was possible, either. Granted, I haven’t played that many.
2 Likes