What's the point of randomness?

[SCO]

Only because most games are cargo cult shit that shy away from a branch in a narrative like a vampire from garlic, and there are many exceptions, for example, Jagged Alliance 2 and being captured.

 

[vean]

Surrendering in JA2 is a completely useless gimmick.

 

[Calcium]

That is usually bad design but sometimes that is intended.
The world doesn't always align with what you want.
The Role Playing part of RPGs implies some sort of narrative, and sometimes that is just how the story goes.
To get the moment of clawing your way out of a desperate situation you must have a system that put you into that situation. The player not being fully in control is the way to do it.

Really? You honestly believe that losing a combat scenario in an RPG is a way to progress the narrative, when in almost every RPG losing combat is a failure state which requires the player to do the combat again? I suppose that if you're a level 60 LARP master you might find a modicum of enjoyment in pondering the consequences of your party/character losing a battle, but the actual narrative of the game you're playing does not change. You win or you lose, and only one of those results actually progresses the 'narrative'.

Go play FTL and get back to us.

If you had read my post you would have seen that I was talking about RPGs.

The only RPG that I could see applying in this sense would be Mount and Blade - a game with a deliberate lack of predefined narrative.

 

[Clockwork Knight]

They are a singular intellectual challenge, not an adaptive one. Here, the Player (or rather, the Participant) never reacts, never has to adapt, because it's a puzzle, not a challenge. Figure out the solution, implement it, win. The other side moves, but that's just fakery, not actual challenge. There is a given optimum solution, and you figure it out.

This has nothing to do with the presence or absence of randomness.

Also the player has to adapt and react every single turn since there's another player working against him. We really need to stop using chess in examples in those rpg discussions. I'm guilty of it too since it's the first thing that comes to mind when you think "tactical game" but it doesn't work.

The point of randomness is that most fights are done by people or other living beings, and unless they're actual Gods they're not perfect. In other words, yes sometimes people trip on a rock.

And sometimes people are critically hit for 300 damage.

 

[Telengard]

Puzzle Chess is something else than regular chess. It is a limited set of moves with a pre-defined optimum outcome. That's why it says "White mates in 2" on the box. This is the very definition of the destruction of all randomness.

Because:
- Given that the player must be able to succeed in combat (because rpgs are not a competition)
- And given that randomness has been removed
Then the player must be given a set of tools that will allow him to succeed.

And so :
- Given that the player has a set of tools that allow him to succeed
- And given that these tools are different from one another
Then there be an optimal use of said tools that will achieve optimum success.

In rpg terms, those tools are moves. Success in a non-random version of rpg will then automatically be defined as the discovery of that optimum set of moves that will achieve optimum success. And that, my dear children, is the definition of Puzzle Chess.

FYI: you can get books of these things, kind of like Sudoku books, but better.

 

[odrzut]

From simulation POV randomness is OK (real world is random too, even on the lowest physical level). But of course games don't need to be simulations (and usually aren't). I'm just writing this to debunk the argument "simulate everything accurately and we don't need randomness".

From game design POV randomness can play many different roles, and can be used well or badly in each of these roles.

One important role (that I think most people criticizing ranomness in rpgs mean) is to include risk management elements. For example rpg trope "Your character can miss with given probability."

If you know full well that your character is going to kill this enemy in 3 turns and will be able to escape from the other enemy in time - after you design the optimum plan you just implement it and watch it unfold. It requires almost no decision making after the first turn. When you can only predict that he will manage to kill first enemy in time with 86% probability, and escape with 60% probability - you have to adapt and to think about risks upfront. Over several rounds you some of the events you estimated are bound to happen the way that was less probable. So you will have to adapt, and depending on how well you managed risk (and how lucky you were) - you can win or lose.

BTW the important fact isn't "real" randomness (games don't usually have real randomness anyway - they fake it using pseudo random number generators). The important fact is - the imperfect knowledge. For player it doesn't matter if the reason he has to change strategy mid-game is because RNG did sth strange, because human oponent did sth unexpected, or just because the game had sth hardcoded to happen at this moment. What matters is - he has to think upfront about unexpected, and to estimate and manage risks. It makes games much more fun.

Randomness has the advantage, that unlike hardocded events it can stay unexpected after reload/on second playthrough. And unlike playing against human player it works in single-player games too.

 

[Calcium]

If you know full well that your character is going to kill this enemy in 3 turns and will be able to escape from the other enemy in time - after you design the optimum plan you just implement it and watch it unfold. It requires almost no decision making after the first turn. When you can only predict that he will manage to kill first enemy in time with 86% probability, and escape with 60% probability - you have to adapt and to think about risks upfront. Over several rounds you some of the events you estimated are bound to happen the way that was less probable. So you will have to adapt, and depending on how well you managed risk (and how lucky you were) - you can win or lose.

Executing an optimum play should result in optimum rewards. As you said - It is still possible to introduce unpredictability through the actions of your opponents rather than "86% chance to hit" said enemies. Probably the most ridiculous implementation of this system I have seen was Arcanum's "critical miss". That was a game which would have been better off without ever having combat.

 

[belowmecoldhands]

Could this all boil down to people who like randomness are just:
* immersion-fags
* exploration-fags
* system-fags

I put system because most games using RNG require you to know the system well to "rig" the dice. It's posible people who don't like RNG never liked systems either, so were woefully unprepared for the dice rolls. This could also mean players who like RNG don't actually like randomness, they just don't notice its ill-affects as much because they're system-fags.

I put immersion because randomness might make the world seem more realistic to the player. That's what immersion-fags want. They expect their character to miss or make mistakes and for others to do the same.

Exploration is there because it's possible randomness is just another version of exploration. The thrill of turning a corner and not knowing whether you'll find treasure or danger or secrets or nothing. The surprise. So fundamentally the explorer isn't motivated exclusively by new things but by new things AND surprises--or randomness. Also explorers are better able to tolerate negative outcomes or dead ends. This is all part of the surprise. It makes it worth it.

EDIT: I probably do like randomness, but only if it doesn't go too far. Too far means too much death and/or no way to prevent with foreknowledge. Fallout is the best example. I HATED the crits in it. I musth ave died at least 10 times. Another example would be Jagged Alliance 2 when I stumbled on some tigers. Like 20 of them. Was no way to win or escape. And that I know of, ther was no reason to suspect they were there beforehand. If an obscure NPC or piece of information DOES exist then it's still annoying. Maybe I can tolerate it, but I would have preferred a way to at least escape. Not necessarily an easy way.

Ranomdness is nice in creating the illusion of quantity and probability. Immersion. To some extent, I like it for this reason most. I usually favor quantity over quality, but I think mostly because games which focus on quality are almost always linear. When I was a young gamer, i hated those games. I have liked a couple. For the most part, I don't like to feel corridored. I will go off the beaten path and check on things. If I feel like there's enough side-quests and free roaming, I'll usually like. I guess I want to feel like my character is what I do and it's unique to me. I want the story to be my own.

Some linear-ish games I liked:
* Anachronox
* Myst
* Albion (this had some open world)

Most of the games I've played are either strategy/tactical/4x/simulation. I prefer open world RPGs.

What I like isn't any better than others. It's just my own preference.

 

[belowmecoldhands]

My a second post. Try to stay brief. I think randomness is like items. Both aren't needed in games. They're just exterior dressing, like hte paint or styling on a vehicle. I think the reason they're there is principally for immersion. We want items and randomness merely because they exist in RL and that's what we want in the game. It's like when girls play dress up or have a tea party with their dolls. They're mimicing their parents or things they see others do. It's make believe based on reality.

But immersion is all across the board. So some players will like XYZ and some won't. We're all different. It's hard to agree. Some players want items and resting mechanics and food and so on. Some want intricate characters with romance. Some want deeply integrated and interlinking quests. Some want a resource management system. Some want lots of twitch gameplay where you jump from one thing to another and shoot and have to learn tricky fast movements.

(Some players even want negative consequences for their immersion.)

Randomizing might be useful for replayability sometimes (like in simple games where pieces might be randomly placed at the start or during play--in Tetris for example), or for generating terrain or textures or clouds or similar. But I think that's not part of this thread, since I think the OP is addressing randomness in combat and in gameplay.

 

[DraQ]

Games aren't reproductions of reality.

That ship sailed... actually even before anyone even thought of porting PnP ruleset onto a mainframe.

Every single bit of fluff, from even tiniest bit of graphics, through textual descriptions, to even tiniest snippets of lore indicates otherwise.

Moreso, games being reproductions of *some* reality is the sole rationale justifying their existence, otherwise they are just sloppy, convoluted messes of loophole riddled diarrhoea, no sane person would ever consider playing in pace of mathematically elegant and far superior alternatives like Go (especially) or Chess.

Anyway, lack of randomness is antithetical to RPG. Without randomness every character is perfect at what they do, always achieving exactly the intended goal, while the very idea of RPG revolves around differences and thus tradeoffs between characters, and usually personal development.

Risk management also makes constructing exact solutions impractical, and favours flexibility and robustness instead.
Sure, excessive RNG is bad, but if your strategy relies on ever single move being executed flawlessly, it's simply put a losing strategy.
Wining strategy is the one that includes all manner of contingencies and that can account for and adapt to changing circumstances.

 

[Guy de Incognito]

The point of randomness is to level the playing field, by giving those less skilled a chance at success and giving those more skilled a chance of failure. Randomness also helps account for complexity from real life situations without having to model that complexity.

That said, my game has random encounters but there is no randomness in combat. I'm still not sure if that's a flaw or a strength.

 

[aweigh]

one of the best games to exemplify how randomizing gameplay elements can increase atmosphere, tension and logistics is Wizardry.

the trapped chests being the most obvious example and then the loot lottery afterwards.

randomizing gameplay elements can be a VERY good thing up until a point; and of course true randomness does not exist but things can't seem stacked too much against the player or it's bad game design.

i recommend the OP go google some texts on casino game theory and skinner-boxes.

 

[Glop_dweller]

Randomness for random-sake is pretty bad design, but most RPG's I have played have used a weighted system that tilts the odds. Those that did not use a weighted system ~tended not to actually be RPGs per say [IMO]. I would not rank a game like N.O.L.F 2 as an RPG, but these days some publishers [erroneously] might.

The random roll can be seen as the sum of all minutia and circumstance at play in the situation; it defines the bar of difficulty for that ~specific attempted action in that specific moment. On the opposing side, the character's level of expertise [the aforementioned weight], can be seen both as the character's ability to influence the situation, and [as result of that] the character's level of confidence at the given task. Failure is effect of the unexpected. For instance, we have a silver-tongued devil of a diplomat PC, one who can talk a prospector out of his gold teeth... and he tries it; and he fails... (consider if the prospector has a toothache and is not really listening in earnest to the babbling suit; the expert fails ~and most unexpectedly). Casey has struck out.

Elsewhere we have the Phantom himself; consummate extraordinaire cat burglar... on the rooftops at night, and he get's noticed because his shoe sticks to some roofing tar, and comes off after a jump... He falls flat on his face, and his tools rattle and clank. This is not a common occurrence, but it shouldn't be impossible. In a weighted random system, the novice makes more mistakes than the expert ~but neither is infallible. Disheartened, our cat burglar goes home, and fails to open his own front door lock, using his own set of keys... How could this be?
He dropped his keys.

But consider a different night, different job; a hotel with a hallway of 25 identical door locks he plans to pick. Should these locks be each and every one a cinch? [aka threshold skills]... even the one where a kid stuck a paper match stick into it? Or the other that had to be rebuilt after a rockstar stayed in the room, and it always seemed a bit stuck after that? These are potential hap & circumstance; but so is a finger cramp; and so is a lockpick that was strained on the last job ~and broke under light use on this job. The actual verbiage [the details] of the situation do not really matter, all that does is that in this moment the attempt failed ~for whatever reason.
*This goes both ways too. Consider the one door in that hall that is unlocked; and another that seems to be locked, but its lock is defective, and a child could pick it with nailclippers. For that matter one could have tried, and broke the lock in the first place, and the expert thief is picking a lock that cannot be opened... but after ten failed attempts, he dislodges the debris, and the lock would have opened on the eleventh attempt... but he moved on to the next door ~which he easily opened.

This concept applies to any and every skill in the game(s); it applies to brain surgery, to document forgery, to hand to hand combat; to public speaking; gem assessment. Things can and do go wrong [often at inopportune times]. The weighted random system ensures this to be the case for the character; just as the character's level of skill ensures their chance to overcome misfortune. Any alternative system that lacked this would seem to make for infallible characters that only act when success is assured.

 

[Galdred]

Leechmonger, let's continue this discussion there, as I don't think the XCOM2 thread is the place for this

I'm generalizing of course, but there's a clear correlation between a game (PC or boardgame) using die-based outcome resolution and it being shit. There are outliers like the original X-COM, which has so many die rolls resolved nearly in parallel (countless recruits shooting one after another) that the statistics almost always converge to the mean. nuXcom lacks this of course due to its limited squad size and is much worse because of it (and other reasons). In general I would say a game's design is worse and less elegant the more luck has a role in determining success (as opposed to skill). Think of Monopoly and Risk as classic luck-based boardgames that are now popular to bash.

Also I'd like to point out that randomness is not the same as luck. A game can inject randomness in several ways that have varying degrees of luck associated with them; the ratio of randomness to luck varies with each implementation. The goal should be to achieve the level of randomness (and hence in a sense replayability) the designer is looking for while minimizing the control that luck has over the outcome of the game.

It is more complicated:
Monopoly sucks because there is zero non trivial decision.
A bot could play monopoly by buying every street he stumbles upon.
Randomness has nothing to do with it. If you allowed the players to choose the results of the die, the game would be as bad as it is with the dice rolls.
I cannot comment on Risk, because I have never played it, but I played a lot of Dude on a Map games (Nexus Ops, Starcraft: The Board Game, A Game of Thrones, Heroscape, Conquest of Nerath), or 4X games (Eclipse, Twilight Imperium) or area control games (Conquest of the Roman Empire, Chaos in the Old World). Most of them had lots of issue, but I never felt that randomness was any of them (granted AGOT and Starcraft don't use dice for resolution).
Diskwars has no randomness btw, and is much weaker than Warhammer.

Risk management is an essential part of strategy, and implementing it without randomness will result in convoluted designs.

It also helps making imperfect information games more interesting (because they are not purely Rock Paper Scissor then, as the moves your opponent can make depend on his hand).

BTW, luck and skills are orthogonal concepts: chess is simple enough that it has been "solved" by computers a decade ago, while 6 players No Limit Hold Em Poker still has not.
Tic Tac Toe has zero randomness. It is a game of pure skill...
Avoiding randomness for the sake of avoiding it is as bad as going overboard with it.
It is true that random events that don't allow any player to react to them are usually bad, but it is not a problem of randomness itself, but a problem of implementation:
In Space Hulk, one of the mission has the players roll a dice at the end: if the Space Marine player has got more Space Marine out than the roll of the die, he wins. It is an exemple of stupid randomness as no one can play after the outcome.

 

[Leechmonger]

I'd like to start by saying that I don't have a problem with randomness, I have a problem with luck. I can tolerate some amount of luck, simply because it leads to an enjoyable level of randomness and therefore variety, but I don't enjoy it for it's own sake. Additionally, the longer the game, the less luck I can tolerate.

Risk management is an essential part of strategy, and implementing it without randomness will result in convoluted designs.

I disagree on both points. Assuming by "risk management" you mean making optimal choices in the face of RNG elements, then no it is not an essential part of strategy as chess and go demonstrate. Risk management is an essential element of reality. It can be implemented in strategy games with varying levels of success. Furthermore, generally I find that convoluted designs tend to have more randomness, not less (epic Ameritrash boardgames). Is Starcraft convoluted because risk management arises out of fog of war? Would it be less convoluted if units' damage variance were increased?

It also helps making imperfect information games more interesting (because they are not purely Rock Paper Scissor then, as the moves your opponent can make depend on his hand).

I'm assuming "it" refers to risk management due to the use of RNG like die rolls or randomized card draw. If so, I think it's important to recognize that some types of randomness offer greater room for variety and skill than others. For example, let's say the optimal move (which has been proven mathematically to yield the greatest win%) is to shoot at an enemy unit with an 80% to hit. It's true that risk management made the process of evaluating the optimal move more difficult (required more skill -> good), but now that the optimal move has been selected, there's still only an 80% chance that the player will be rewarded for it. In other words, a skill ceiling has been placed on that turn (and depending on the turn's importance, even the whole match) because no matter how good of a player you are, you'll still miss 20% of the time. A good game would of course involve enough die rolls that it all evens out in the end, but many popular games frequently involve a crucial roll that gives one player a significant advantage.

Further food for thought: card games. A randomized hand challenges players to play new configurations every game. Already there is potential for luck (mana screw/flood in Magic), but I think worse than that is when the game has stalled and both players are hoping to topdeck their game-winning card. At that point if one player draws the board wipe, removal, or whatever there is zero skilled involved. You either drew it or you didn't.

chess is simple enough that it has been "solved" by computers a decade ago

Chess has never been solved. On the contrary, chess is complex enough that it has had unimaginable amounts of computational power thrown at it by mathematicians and computer scientists world-wide for decades and still has not been cracked. Even when it is finally solved, the amount of time and effort it will have taken speaks volumes as to the decision-making space the game presents. Not to mention that the solution would almost certainly be impossible for a human to implement without the aid of a computer.

Avoiding randomness for the sake of avoiding it is as bad as going overboard with it.

I'll refer back to the beginning of my post: I want randomness due to the variety it provides at the cost of luck. I like games that maximize variety while minimizing luck through intelligent design decisions.

In Space Hulk, one of the mission has the players roll a dice at the end: if the Space Marine player has got more Space Marine out than the roll of the die, he wins. It is an exemple of stupid randomness as no one can play after the outcome.

That's an extreme example of the downside to die-based outcome resolution. Take my earlier example of having an 80% to hit and place it in a context where landing the hit means winning the game or gaining an insurmountable advantage.

 

[Galdred]

I'd like to start by saying that I don't have a problem with randomness, I have a problem with luck. I can tolerate some amount of luck, simply because it leads to an enjoyable level of randomness and therefore variety, but I don't enjoy it for it's own sake. Additionally, the longer the game, the less luck I can tolerate.

Risk management is an essential part of strategy, and implementing it without randomness will result in convoluted designs.

I disagree on both points. Assuming by "risk management" you mean making optimal choices in the face of RNG elements, then no it is not an essential part of strategy as chess and go demonstrate. Risk management is an essential element of reality. It can be implemented in strategy games with varying levels of success. Furthermore, generally I find that convoluted designs tend to have more randomness, not less (epic Ameritrash boardgames). Is Starcraft convoluted because risk management arises out of fog of war? Would it be less convoluted if units' damage variance were increased?

You got a point, it is not essential, but it is an important part. Positioning could also be considered non essential, given that some strategy games don't have a board where position matter. But it is also an important part. And yes, chess does not have risk management. Not all games need to implement all dimensions of strategy.
The problem if luck in long games is something else entirely, and it also afflicts game without randomness:
It is the time between the moment a winner has emerged, and the moment he fulfills the victory conditions.
In monopoly, it takes about half of the game.
Of course, it can also be a problem if a very long game is decided by a single die roll at the end of the game, but as long as this one was decisive only because the game was very balanced before (like a 16th turn unlikely TD in Blood Bowl), it is not inherently a problem (nor is not liking this kind of game obviously).

It also helps making imperfect information games more interesting (because they are not purely Rock Paper Scissor then, as the moves your opponent can make depend on his hand).

I'm assuming "it" refers to risk management due to the use of RNG like die rolls or randomized card draw. If so, I think it's important to recognize that some types of randomness offer greater room for variety and skill than others. For example, let's say the optimal move (which has been proven mathematically to yield the greatest win%) is to shoot at an enemy unit with an 80% to hit. It's true that risk management made the process of evaluating the optimal move more difficult (required more skill -> good), but now that the optimal move has been selected, there's still only an 80% chance that the player will be rewarded for it. In other words, a skill ceiling has been placed on that turn (and depending on the turn's importance, even the whole match) because no matter how good of a player you are, you'll still miss 20% of the time. A good game would of course involve enough die rolls that it all evens out in the end, but many popular games frequently involve a crucial roll that gives one player a significant advantage.

Further food for thought: card games. A randomized hand challenges players to play new configurations every game. Already there is potential for luck (mana screw/flood in Magic), but I think worse than that is when the game has stalled and both players are hoping to topdeck their game-winning card. At that point if one player draws the board wipe, removal, or whatever there is zero skilled involved. You either drew it or you didn't.

The optimal move is not only taking the shot, but planning for the 20% of times you will miss it.
That is why you don't go all in for a 80% chance to kill in X-COM, or why you try to do the 80% after having secured a "correct" backup position in Blood Bowl. In other words.
And actually, the randomness is part of why there usually is no optimal choice:
It will often depend on the relative skill level of your opponent: if you opponent is better, then chosing an low variance strategy is a bad move, but if he is weaker, then chosing a safer one, where you only take a 60% shot, but in which the result of failure won't be catastrophic (because you took the shot from a safer position) can make more sense.
So estimating the skill level of your opponent (or the risk you can afford to take) can be critical in such games, so it is usually not as simple as choosing an optimal move (because the best expected result is not always the best move to make).

Regarding top decking, you are correct that it is not a very interesting situation, but it is mostly a tie breaker for a game that went for too long (and there is some resource management to avoid it in the first place in CCG), like the penalty shootings at the end of a soccer game.

Chess has never been solved. On the contrary, chess is complex enough that it has had unimaginable amounts of computational power thrown at it by mathematicians and computer scientists world-wide for decades and still has not been cracked. Even when it is finally solved, the amount of time and effort it will have taken speaks volumes as to the decision-making space the game presents. Not to mention that the solution would almost certainly be impossible for a human to implement without the aid of a computer.

I know, hence the use of the quotation marks. My point is that chess is calculatory enough that it can be brute forced.

I'll refer back to the beginning of my post: I want randomness due to the variety it provides at the cost of luck. I like games that maximize variety while minimizing luck through intelligent design decisions.

Luck also allows to simulate a lot of factors that would be tedious to track. If you prefer euro games, that may not be a problem for you, but you cannot have any strategic simulation without luck. But to each his own. The game not suiting your preference does not make it bad.
I don't like eurogame because I don't like solving optimization problems: they usually just feel too puzzly for my tastes. That doesn't make me consider them bad (and I can even tolerate playing them with friends!)

In Space Hulk, one of the mission has the players roll a dice at the end: if the Space Marine player has got more Space Marine out than the roll of the die, he wins. It is an exemple of stupid randomness as no one can play after the outcome.

That's an extreme example of the downside to die-based outcome resolution. Take my earlier example of having an 80% to hit and place it in a context where landing the hit means winning the game or gaining an insurmountable advantage.[/QUOTE][/QUOTE]
More often than not, a good part of the outcome will have been decided by what led to this 80% shot.
I never had any issue with close game going one way or another. Why is it so important that the better player win 100% of the time?
In poker or MTG, for instance, what counts is you long term winnings. Winning a single hand is irrelevant.
In Blood Bowl, most leagues use a championship system, where winning most games out of 10 is what matters, not each individual game.
My point is that as long as the game leaves enough room for player decision (so no monopoly, or tic tac toe), the better player will still win in the long run, and there will be a format better suited to that game.
Note that Risk having lot of randomness is not that important, because the game is mostly decided by your kingmaking skills (same for Game of Thrones which have little randomness).

 

[Leechmonger]

The optimal move is not only taking the shot, but planning for the 20% of times you will miss it.

You've missed my point entirely. I'm saying that after accounting for all possible situations (making the shot, missing the shot, missing both this shot and the next, and so on) it is possible to deduct that taking the shot will win you more games than not taking the shot. You should therefore take the shot, and missing the shot afterwards will not retroactively make your choice a bad one.

And actually, the randomness is part of why there usually is no optimal choice:

There is always an optimal choice for a given context (game state, opponent, etc.) that is most likely to win you the game in the long run, either because of RNG probabilities, number of paths to victory (chess), or both. Sometimes there are multiple choices that are equally good (equally optimal), but then it doesn't matter which of these optimal choices you go with.

I know, hence the use of the quotation marks. My point is that chess is calculatory enough that it can be brute forced.

Luck also allows to simulate a lot of factors that would be tedious to track.

If you're not tracking them you're not really simulating them, they've been abstracted away. It's a tradeoff between realism and simplicity. I think oftentimes if you're comfortable reducing a complex system down to a die roll, go ahead and remove the die roll as well.

Why is it so important that the better player win 100% of the time?

I believe that, all else being equal, the higher the winrate of the better skilled player the better the game. As noted previously I'm willing to sacrifice some amount of winrate to increase variety.

In poker or MTG, for instance, what counts is you long term winnings.

That's nice and all but losing an MTG championship to mana screw must suck a huge dick.

My point is that as long as the game leaves enough room for player decision (so no monopoly, or tic tac toe), the better player will still win in the long run, and there will be a format better suited to that game.

If the better player wins 51% of games, would you be satisfied?

 

[Galdred]

The optimal move is not only taking the shot, but planning for the 20% of times you will miss it.

You've missed my point entirely. I'm saying that after accounting for all possible situations (making the shot, missing the shot, missing both this shot and the next, and so on) it is possible to deduct that taking the shot will win you more games than not taking the shot. You should therefore take the shot, and missing the shot afterwards will not retroactively make your choice a bad one.

And actually, the randomness is part of why there usually is no optimal choice:

There is always an optimal choice for a given context (game state, opponent, etc.) that is most likely to win you the game in the long run, either because of RNG probabilities, number of paths to victory (chess), or both. Sometimes there are multiple choices that are equally good (equally optimal), but then it doesn't matter which of these optimal choices you go with.

You can pick the option with the best expected value ("optimal choice") or the one with the least worst outcome in case of failure. In Blood Bowl, many players usually go for the second option.
And having randomness makes the branching factor get so large that determining the optimal move is more difficult (ie best expected value in this case) than in a system where you simply trade deterministic hitpoints.
+ Randomness forces you to plan for all the possible outcomes of your action: You have to both plan for the 20% chance that the shot will miss, and the 80% it will hit. Many players consider that 80% = 100%.
XCOM is a good experience that makes you feel how wrong it is.
It is very possible that the best move is rewarded with failure indeed, but I don't see how it is a problem. It is part of the skillset: to determine whether you played wrong, or whether you were unlucky. Most people tend to attribute way too much of their losses to the luck factors.

Luck also allows to simulate a lot of factors that would be tedious to track.

If you're not tracking them you're not really simulating them, they've been abstracted away. It's a tradeoff between realism and simplicity. I think oftentimes if you're comfortable reducing a complex system down to a die roll, go ahead and remove the die roll as well.

I still fail to see how to make compelling wargames without the dice roll, but a determinist hitpoint system.
You cannot simulate conflict if the outcome of the actions you have planned is deterministic, because it is in the nature of warfare that your units won't always perform the task as you hoped they would. Do you know of any wargame with a deterministic system?

Why is it so important that the better player win 100% of the time?

I believe that, all else being equal, the higher the winrate of the better skilled player the better the game. As noted previously I'm willing to sacrifice some amount of winrate to increase variety.

In poker or MTG, for instance, what counts is you long term winnings.

That's nice and all but losing an MTG championship to mana screw must suck a huge dick.

Of course I got angry losing 100$ all in with a pair of Ace versus something else preflop in poker, but you learn to let it go
And some unlikely screw ups make for quite memorable stories:
I still remember my company of Shadowsword super heavy tanks being destroyed by a single volley of an exarch platoon in Space Marines (I lost 3 instead of 1 on average). But I got my revenge a few games later when a single Terminator platoon destroyed the reactor of the Eldar titan by an unlikely volley.
As it averages out in the long run, it doesn't bother me so much.

If the better player wins 51% of games, would you be satisfied?

The only game where the better player will always win 51% of the time regardless of the skill difference (ie not 55% if he is much better) is some weird mutation of chess or purely skill based game where you roll a D100 on checkmate and invert the winner on 49-100. But if we assume a less extreme : it never bothered me in poker and some CCG. What matters to me is the number of significant decisions per unit game time.
But of course, I agree with your point that a very long game that can snowball because of a single random event in the end is a bad thing. But it is usually not so clear cut.

 

[Leechmonger]

You can pick the option with the best expected value ("optimal choice") or the one with the least worst outcome in case of failure.

Or you can pick the option that wins the most games which will, by definition, which you the most games.

There's no point in continuing this conversation if we can't agree on this.

 

[Galdred]

You can pick the option with the best expected value ("optimal choice") or the one with the least worst outcome in case of failure.

Or you can pick the option that wins the most games which will, by definition, which you the most games.

There's no point in continuing this conversation if we can't agree on this.

The mathematical optimal move assumes that the opponent plays optimally, so I assumed you were speaking about this one.
The optimal move against a specific opponent is not always the same as the optimal move that wins the most games against a random opponent.
A sub optimal opponent can be exploited by using a strategy that would be sub optimal against an opponent playing optimally.
Playing safe is exploiting the fact that your opponent plays sub optimally, which can be better than doing what wins you more games on average against a random opponent, but there are many variants of it that can also work in a deterministic situation.

A simple exemple with randomness is the following:
Suppose that we are playing some kind of assassination game with random chance to hit:
You optimal move is to expose your commander to take out the opposing big guy.
But if you do so, the opponent can make a move with his own commander that would result in his getting killed 80% of the time, and yours getting killed 20% of the time.

It is still optimal to take the chance (ie expose your commander to a 20/80 desperate strike by the opponent) against an optimal opponent, because taking the 20/80 chance would make no sense for him.
But against your moronic opponent that likes to win through suicidal strategies, it is better not to take the "optimal move that wins the most game", because even though not risking the 20% chance of successful kamikaze strike with your commander is sub optimal, you are confident your chances of winning would be more than 80/20 in the long run against this particular opponent.

Of course, this kind of opponent specific actions can occur in non random games provided you have a strong read on your opponent strategy too. My point was that this risk vs rewards decisions occur more naturally in games with randomness (and imperfect information but it is another subject altogether).

 

[Leechmonger]

The optimal move against a specific opponent is not always the same as the optimal move that wins the most games against a random opponent.
A sub optimal opponent can be exploited by using a strategy that would be sub optimal against an opponent playing optimally.

I'll go ahead an quote myself:

There is always an optimal choice for a given context (game state, opponent, etc.) that is most likely to win you the game in the long run, either because of RNG probabilities, number of paths to victory (chess), or both. Sometimes there are multiple choices that are equally good (equally optimal), but then it doesn't matter which of these optimal choices you go with.

 

[Galdred]

The optimal move against a specific opponent is not always the same as the optimal move that wins the most games against a random opponent.
A sub optimal opponent can be exploited by using a strategy that would be sub optimal against an opponent playing optimally.

I'll go ahead an quote myself:

There is always an optimal choice for a given context (game state, opponent, etc.) that is most likely to win you the game in the long run, either because of RNG probabilities, number of paths to victory (chess), or both. Sometimes there are multiple choices that are equally good (equally optimal), but then it doesn't matter which of these optimal choices you go with.

I didn't see you mentionned the opponent before, but I assumed that an optimal move was a move that was part of an optimal strategy (which is a strategy that cannot be exploited regardless of the oppnent, not the strategy to counter a specific opponent) from the earlier post where you didn't mention the opponent at all.

Even in a deterministic game, you can lose by playing an optimal move as long as there is hidden information: If you play Starcraft and know that your opponent will rush 95% of the time, you should build turrets 100% of the time. You will be at a strong disadvantage the 5% of the time he didn't rush, because his economy will be much better than yours(we'll assume that scouting is not possible to make things simpler).
Given that the game state your move is based on can only be comprised of the things you know, so it was correct of you to prepare for a rush, even if it costed you this specific game where he didn't do that.

So yes, your current reward will usually not be your expected reward if the game is random or has hidden information. It can make it harder to determine what caused the final outcome of the game, but you can also consider that as part of the skillset (ie finding whether you played well or not regardless of the result).

But my point was that randomness (and/or imperfect information) make it more important to know your opponent to determine the best move against him, while going for the optimal move (ie the best move regardless of your opponent) is usually "sufficient" in deterministic games with complete information(if you know that your opponent always goes with the same opening, or is bad with some endings in chess, it may also make the best strategy against him differ from the optimal one, but it is not nearly important as sizing your opponent in poker for instance).

 

[Elemental_wanamingo]

There are many reasons why randomness is part of games. Some of the reasons are based on simulations and ultimate knowledge of the system while others are purely psychological, gameplay technical or just ideological.

Real life has randomness but the randomness is just stuff we don't know exactly. If we could know and measure everything we would get very close to 100% accuracy in our predictions. But knowing everything is expensive, takes a lot of time and after a point doesn't even have any effect anymore. When firing a bullet we can measure each and every bullet shape, material properties and analyze the air composition, wind and all that to try to get a perfect prediction. We can similarly do a similar analysis about the foe who is being shot at. What kind of armor exactly is he wearing, the exact thickness, layering and angle of the armor in various places. The coins he has in his pockets, the amulets he is wearing. The wear on the armor plates, the smell of the sweat of the slaves that mined the ore to make that material. What does he do when he is shot. And even then we don't end up with clear prediction. We end up with a propability.

While this probability comes directly from measurements and information it is very complex and hard to understand and read system. And even if you know what all the things are that effect the result you are still left with a probability. It is a bit like weather forecast. The system we are reading is extremely complex. We can make predictions with good accuracy and over large sample of tries we may get good accuracy. But individually we sometimes get it really wrong while other times we get it exactly right. Even if we understand the reasons why we can't make precise prediction - it doesn't make our prediction more accurate for that attempt. Over large sample of attempts we probably still get it right more than wrong but it is still a probability. Be that result of measurement of real world or simulation of the game world.

From game perspective we have two different choises. Either code all those variables in and calculate them all the time. Or just use numbers to give typical probability for each situation. Coding it all in is a lot of work though. And in the end would anyone notice if the trajectory calculation code had 5000 lines of psychology code, 5000 lines of physics calculation, 5000 lines of biology and human anatomy, 5000 lines of material properties and weapon tolerances and smaller amount of randomness than just something that is 50 lines of code and rng generator?

One of the problems with randomness in rpgs, especially turn based ones is that it looks bad. It is hard to represent the chaos and movement of combat when each unit moves one by one and shoots at stationary target 5 feet away and the result seems like just a dice roll. That xcom example suffers so much from this. Stationary units blazing away at stationary targets. And the most accurate guns don't even hit the same postal code as the target. It looks completely ridiculous. The easyness of the hit is insanely exaggerated while the chance to avoid being hit is almost comically understated. There is no zigzagging of the alien that makes you miss it with your shotgun. There are no difficult angles, unpenetrable steep angles or armor nor weight of the weapon and its aiming fighting against you. The alien just fucking sits there and you miss. Or so it looks. It is a representation issue. Not rng issue. It looks too easy. The difficulty of the shot is not there graphically. It is there only in the numbers.

Also I'd argue that in most cases some amount of randomness can make a dull game look more exciting than it is. Even in basic dull fight people still want to see what kind of numbers they roll. It adds some value to an event that otherwise would be uninteresting. The difficulty is managing frustration while still keeping the numbers dynamic. It is certainly annoying to miss at 95% chance but similarly it is kinda fun when you hit that 5% neck beard shot and are rewarded with disintegrating npcs. Not so fun when you are the one disintegrating though so it is a difficult balance sometimes.

 

[Galdred]

One of the problems with randomness in rpgs, especially turn based ones is that it looks bad. It is hard to represent the chaos and movement of combat when each unit moves one by one and shoots at stationary target 5 feet away and the result seems like just a dice roll. That xcom example suffers so much from this. Stationary units blazing away at stationary targets. And the most accurate guns don't even hit the same postal code as the target. It looks completely ridiculous. The easyness of the hit is insanely exaggerated while the chance to avoid being hit is almost comically understated. There is no zigzagging of the alien that makes you miss it with your shotgun. There are no difficult angles, unpenetrable steep angles or armor nor weight of the weapon and its aiming fighting against you. The alien just fucking sits there and you miss. Or so it looks. It is a representation issue. Not rng issue. It looks too easy. The difficulty of the shot is not there graphically. It is there only in the numbers.

Good point. XCOM misses would make a lot more sense if there was a dodge animation actually. It means that it is pretty hard to convey what these rolls really mean indeed.