The Role of Randomness
In game design, a “random element” or “random mechanic” is not always desirable. Randomness in some circles is even looked down upon as a “cheap” mechanic that degrades the strategic or tactical elements of a game. The phrase “luck based” has been used to label the random element of games to showcase that the skill of the player cannot be fully expressed in that type of game. On the flip side, many games hinge solely on a central random mechanic and in some circles “luck based” or random games are considered more fun or more inclusive of all players regardless of skill level.
So what is the role of randomness in games? Don’t be shocked now… but the role is to provide a random element to the game. *gasp... looks right then left*... did I lose anyone?
Both circles of thought about randomness in games make valid points. But what if I told you that they were both mistaken about the nature of randomness. Now before you tune me out, I don’t think either side is wrong about “how randomness gets used in games”, but they are both confused about how randomness can be used in games. I don’t feel that random and strategic elements are always mutually exclusive.
At the core, randomness in game design deals with the concept of predictability.
- Predict: “Say or estimate that (a specified thing) will happen in the future or will be a consequence of an action.”
Randomness that creates unpredictability is not very desirable in a strategy game. In most strategy games players expect a certain level of predictability to the outcome of the actions they take. If a player cannot predict the outcome of an action (or set of actions) building strategies becomes nearly impossible.
Not all randomness in games create unpredictability and the level of complexity within a game can also create unpredictability without the use of randomness.
Let’s take a step back and look at some subtypes of randomness:
- Endless random
Dwindling random
Random output (after a player makes a choice)
Random Input (before a player makes a choice)
Endless random is straightforward, like dice that give an inexhaustible supply of random numbers. This might also be a deck of cards that once depleted is shuffled (randomized) and reused. There is virtually no way for a human to accurately predict the outcome of endless "truly" random elements.
Dwindling random is like a deck of cards that is not shuffled (randomized) and reused. This might also be a set of tokens in a bag. As you draw cards or tokens from the supply the number left becomes less and less until only one predictable non-random card or token is drawn. This assumes the players are familiar with the set of random elements in the deck of cards or set of tokens in the bag.
This touches on the idea of “counting cards” as a player that knows the number and types in a set can mentally track what elements remain in the set as those elements are drawn or used. This idea is that the randomness becomes less and less as the elements of the set are used. In this design structure, an experienced player can begin to predict to some extent the outcome of events the closer to the end of the set. To eliminate or lessen this type of predictability within the dwindling random system the majority of blackjack games today use six or eight decks so that counting cards becomes much more difficult (nearing impossible for humans).
Random output deals with the timing of the random element. This is the place that randomness in strategy games is undesirable. The player takes an action, then the random force or element changes things right before the outcome of that action is decided. The outcome or output of the action is affected by the random element. This is frustrating in strategy games because the player cannot predict the outcome before the action is taken. While this type of randomness is great for many other types of games it is poorly suited for building extended “long-term” strategies. It might be “true to life” that there are random forces at work in just about everything, but the strategy game (as such) depends on predictability.
Where this type of randomness shines is in role-playing and simulation games that deal with unpredictable forces. This is the “luck” factor and the “wind” of the game world. Many times this type of randomness is measured or controlled by the mathematical and statistical formulas related to probability. This can set things up so that a player can “know” what chance they have to succeed or fail before taking an action. But in the end, the random force controls the output. So there can be some “strategy” to the way a game is played within this form of randomness, just not the type of strategy needed for traditional “strategy based games”.
Random Input is when the random element is known by the player before the decision to take and action is made. This can be done many ways but similar to output, this type also deals with the timing of the random element and not necessarily the method of randomness used. For instance, a random tile is flipped revealing the next challenge a player must overcome in a tactical way. If all of the tiles used to build a “random” game board are set out before the game even starts, the players would all have access to the same “random” information about the board. This would still give the players the ability to plan out or build a strategy for this game. This type of randomness will not work in “strategy based games” unless most if not all of the random elements happen before the game starts. But the timing of the randomness it the important part here.
This is also called “procedural generation” in the video game world. It can throw a monkey wrench into the realm predictability but it does not always need to in order to be effectively used in game design. This type of randomness, more than any other, is overlooked by board game designers. This has been changing some in the last few years so we will see how it can be used in the future.
As game designers, it is critical to understand what role we want randomness to play in the game we are designing and also how that randomness will affect player strategy. Randomness can have a place in any style or type of game if used in the right way, but the players will be the final judge of whether or not the game is “fun”.
- One final note: “Player action selection, when no single choice is better than any other choice in a set of predetermined choices, is never truly random.”
Player based “random” selection: This is a type of “randomness” not totally covered by the other subtypes above. This is when a player needs to pick one option but does not need to “care” what action is selected because all of the choices have the same potential benefit. Humans don’t do true randomness in their selection of any choice.
There is nothing all that “random” about what you select when playing rock - paper - scissors (RPS). You may “feel” like you are choosing a random selection but science has proven that humans do not naturally select true random strings. When people play this game against machines that can create true random strings, the machine wins many more games than it loses. So while the dominant strategy of RPS is to select the most random option each time, humans and computers can’t do it. Keep this in mind when taking the player based random selection into account during game design. This is Pseudo-Random vs. True Random, and while it is less important to get into this subject for board game design. Just keep in mind that balanced dice really do give random numbers and common board game dice just like people don’t.
For the board game designer, whatever method or procedure you pick for the role randomness in your games, just make sure it does the “thing” you designed it to do and everything will work out fine. (until the playtesters get their hands on it)
This is intended only as "Food for Thought". Please let me know what you think, I am by no means the authority on this subject so any input from other designers is greatly appreciated.
"Always remember to think outside the box so your games will fit inside!"
@BHFuturist
Comments
I love dice, and the more
I love dice, and the more dice & variety of dice types the better! But I also love the feeling in a game of my strategic decisions playing out the way I planned, leading me to victory!
Can one game have both of those elements? Yes! I believe one important key in such a game is giving players multiple ways of dealing with randomness. If you can prepare for randomness in multiple ways, and have multiple uses for random elements, your strategy lies in trying to make the best choices with what you have. Also, many games have "modifiers" that help you alter the randomness you face.
I'm going to use a few examples from my current design "Chrysopoeia" because "dealing with randomness" is the main focus of the game mechanics. Don't mean for this to be a plug for the game - I'll use other games as examples too.
Random input: To me this would be something like Catan, rolling to see which resources you get this turn. Do you have choices in your game as to how you deal with this? Can each resource be used for one thing only, or can it be used or saved for a variety of things? Can it be traded with the bank or other players? The more options you have, the less chance that the "randomness" has of ruining your strategies.
Random input and output in a battle: Is it possible for combat to be completely devoid of randomness? I think there will always be an element of randomness from the choices your opponent makes. Your strategies may force an opponent to make choices that they'd rather not make, like "retreat", but a good opponent might just make a surprise move that you didn't expect!
In Chrysopoeia, battles aren't about winning the game - they are a mini-game used to capture workers and sabotage your opponent's future Alchemy experiments. Players acquire Strategy cards from a face down deck (Random input). Players can choose to focus more on acquiring more strategy cards, or building upgrades that provide permanent modifiers for either attack or defense (strategy). Or they can focus more on resource benefits and equipment for Alchemy experiments, which is where the game points are won (again, this is strategy).
In a battle, players roll a combination of battle dice (I'm calling this roll "random input"). They add their respective Strike or Defense modifiers, and play Strategy cards to further modify their scores. Players can also build a Time Machine which allows them to re-roll any bad die results.
Quick note on the Time Machine: can be used to re-roll in battles, but also for resource production rolls and Alchemy dice rolls. This produces "random output", but it is used to change an already "bad" roll, so it usually ends up as an improvement.
Alchemy experiments are similar to the adventure card mechanics of Elder Sign. You have "tasks" or "experiments" that consists of a row of symbols that must be matched from a roll of several dice. Each card has 2 or 3 rows, so when you roll, you can take the "random input" and try to match the symbols to one row. If you can't make a match, you can use Apparatus cards to change one die to a specific result, or have a wild card to change one die to any result.
There are a variety of Apparatus cards, which players can decide how many to build up in their hand before attempting Alchemy experiments. Thus, preparing to be able to deal with the random input of the dice rolls. I call it input because it's what you start with, and you use your "equipment" to change that input to a better output. Also, the Time Machine can be used to re-roll the "undesirable" dice results only, while keeping the dice results you need. This random output usually helps but can also surprise you by making things worse - it is player choice that decides what order to use all these elements, and that order can make a difference in your success! So player preparations, anticipations, and choices in how they apply the dice results (trying to complete the more difficult sets of symbols first, for example) are all "strategic".
Elder Sign also has a lot of those elements. The big difference between Chrysopoeia and Elder Sign is that in Alchemy experiments, you risk losing ingredients required for the card (again, you can prepare with extra ingredients). In Elder Sign you risk losing dice, so subsequent rolls will be less and less effective. I prefer my Alchemy experiments - far less frustrating, even if you fail!
OK, hope this reply illustrates how it can be fun and rewarding to use your many choices to prepare for and deal with the tension-generating excitement of randomness in a game!
Thank you!
This is the confusion of randomness and unpredictability. The two ideas are not the same thing but are loosely related.
I may not have explained this very well, but a player's choices are not random, they are unpredictable. You can have many unpredictable forces working in a complex system but that does not make that force random in any way. This is, of course, a very large subject that I might have to write a separate article about someday.
This brings up a great point. The real difference between input and output randomness is where you place it within the "sequence of actions" in the game. The fact is the random element can be placed in front of a player's choice, between player choice and the outcome of their action or even be the result of the action or any combination of ways. Whatever we call it, understanding how where you place it effects the game is the important part! I like how you are handling the random element and giving the players more ability to react tactically to it.
There is a distinction to be made between "being able to react tactically" and "being able to build a long-term strategy of tactical reactions". It sounds like you are working to have both in your game while still having a random element. My overall point is just that! it is possible to do this if we understand the role that randomness plays in the game.
I agree with virtually everything you wrote and enjoyed hearing your way of saying the things that I also think. Bringing out the different ways we all think and say (in slightly different ways) things, is one of the reasons I wanted to start writing these articles in the first place. I don't have all the answers! Only together can we build the answers.
Thanks for taking the time to read and comment!
-Eamon
This is a fascinating subject
This is a fascinating subject for me Eamon, thanks for starting this thread!
I didn't quite get the distinction you were making before between "random" and "unpredictable". I think I get it now though. With a 20 sided die, you can predict probabilities, and the probability of rolling a "1" is the same as rolling a "20". When rolling 2D8, the probabilities are no longer equal, but still you can calculate probabilities.
Edit: not that dice are predictable, but there are a finite number of possibilities. Even more so with cards - they run out.
With players who have multiple choices for strategies and goals, you can't really calculate probabilities. You can try to predict an opponent's next move in Chess, but they might make a move you didn't see, or they might even make a bad move because of something they DIDN'T see. That's unpredictable.
Is that the way you were distinguishing the two concepts?
I believe a good strategy game needs to provide many options. Players can't plan (strategize) if only one option is available to them. The more options, the more complex of a strategy you can develop, which might also make it harder for other players to predict your moves.
That's not always the case - opponents may see the direction your choices are going and draw conclusions (predictions). But if there are enough options in a game, feignts and distractions can become another unpredictable aspect. Your opponents "think" they know what you're doing, but you really have something else planned
Sort of...
You are right about this being a fascinating subject! While I don't completely understand all the mathematical "proofs" that are involved in some of the more complicated aspects of this subject, the concepts are getting clearer to me all the time.
Let me try to say it a different way:
Randomness always leads to unpredictability
-BUT-
Unpredictability can be reached by factors other than randomness
There are many factors that can lead to unpredictability:
Player Choice
Randomness
Game Complexity
Hidden Information
and more I am sure...
No matter how you get to unpredictability, it can then be managed by the ability to calculate probabilities. Here are two statements to illustrate this.
Player choice (while not random) does insert unpredictability to a game. However, this can be managed by many known factors (if known) to calculate the probability that their next choice will be X.
A die roll (while completely random) does insert unpredictability to a game. However, this can be managed by known factors (number of sides) to calculate the probability that the next roll will be X.
Both paths lead to unpredictability from there how you manage that is up to you as a game designer. Both random and deterministic options can create unpredictability.
For strategy games, as long as the randomness is inserted into the right place and time in the game it should theoretically make no difference to the tactical and strategic decisions a player needs to make in order to cope with the unpredictability.
For player choice, this is when "tactical & strategic" thinkers ask themselves the types of things found in the monolog by Vizzini in the movie the princess bride:
More to come on Tactical vs. Strategic, I am working on that article now :)
-Eamon
To me randomness is the
To me randomness is the foundation of the strategy. Many people thing that deterministic structures are related to strategy, but this is far to be true: determination is related to technic, not strategy.
The randomness destroy, in some level, any technical or algorithmic mechanic to play a game, liberating in the process the strategy behind a mechanic, then the level on strategy increases heavily.
con-fusion
I love it that there is always confusion about randomness and unpredictability.
This is how I look at the 2 concepts:
Randomness, like a d6 roll, is predictable. 1/6th of a chance for every outcome.
However, going to roll only once, will have an unpredictable outcome.
huh, what?!
Predictable. (Predict-table)
You know what you might roll.
Unpredictable.
But you don't know what you are going to roll.
Random.
You know what you might roll, but you don't know what you are going to roll.
I bet you are even more confused now. :) Lets... do a bit more damage, shall we?
There are many, many more things that are unpredictable. And un-predictable can be read in 2 ways.
You do not have a predict-table. You don't know the possible outcomes on the die.
Or you cannot predict the table. 1 roll is less than the table with 6 outcomes. You need to roll many times more to see the table.
***
PS
I like these blogs. That was unpredictable, wasn't it? Or was it? At least it isn't random.
Confusion in Meanings
The randomness destroy, in some level, any technical or algorithmic mechanic to play a game, liberating in the process the strategy behind a mechanic, then the level on strategy increases heavily.
I have always thought that randomness could play a strong role in a well-developed strategy game. However, I am not sure I have ever thought of it as "the foundation of strategy" as you put it.
Deterministic structures are "in my mind" very important to the concept of strategy. This is evidenced by games like Chess, there are no random elements to the game and it is entirely deterministic. While you might be able to argue that "the Technique of master level play" is to memorize as many strategies as you can in order to have the knowledge to counter anything your opponent might do. However, most of the world (including myself) consider the game to be one of the greatest strategy games of all time.
It could be that I just don't fully understand what you were trying to say, it is one of my thick headed flaws. I do tend to get tunnel vision once I have an idea firm in my mind. For sure, I don’t understand how you are using the word “Technic” which is another form of the word “Technique”.
Technique: "A way of carrying out a particular task, especially the execution or performance of an artistic work or a scientific procedure."
Because the "playing" of a game, mixes the two types of "tasks" from the definition above, playing a game "well" means you will need both good “Technique” and “Strategy” to win.
"winning" is, therefore, the successful execution of a plan to win (so to speak).
-----------------
However, going to roll only once, will have an unpredictable outcome.
You are very right about there being a lot of confusion.
Predict: "Say or estimate that (a specified thing) will happen in the future or will be a consequence of something."
-able: "A suffix meaning “capable of, susceptible of, fit for, tending to, given to,”
Predictable (Predict-able): "Able to be foretold or declared in advance."
A "table of probabilities" does not (and cannot) make a prediction for the next roll. In fact, the idea of predicting a truly random number is mathematically impossible. You can roll a six-sided die as many times as you like (forever even), and you will never have more than a 1 in 6 chance of "predicting" any one roll. With anything that is truly random, the number of times the random cycle is repeated makes NO difference to increasing the chance of predicting the outcome.
Only in dwindling randomness does the probability "dwindle" until it becomes predictable to 100% as the last item from the set is revealed. (last card in the deck or last token from the bag).
Becuase the action selection of humans and computers can be based on the input of the situation and the logic of what is best in that situation there is deterministic qualities to the selection. This is why the actions of a human or computer can be predicted if the logical variables are known.
I do not love it that there is always confusion about randomness and unpredictability. I would love there to be more understanding of the words and concepts on this topic so we can all grow together and have meaningful discussions.
My comments, as always, are only ever intended as "food for thought" or to present my non-authoritative viewpoints. I really am a calm, rational and reasonable guy, once you get to know me. I do not intend offense of any kind should I seem to disagree with you or anyone else on any subject.
-Eamon
I like to juggle with words.
I like to juggle with words. It gives one a different view on things. Just sharing my thoughts, not saying that any one is incorrect (that includes me).
When predicting that the chance to roll a 6 with d6 is 16,7%. You can predict/calculate any other chance to roll multiple times a 6 in a row. Eventually this chance becomes so low. You can predict that rolling 100 times d6 will give a very low chance to roll a 100 times 6.
Your multiple rolls, still random. Will become more certain. And thus predictable.
In fact, I used so much table's to calculate the possibilities of outcomes. That the true definition by dictionaries, is a bit lost to me at this point.
I see your point...
But, your scenario is based on the desire to roll or not roll only one of the possibilities and adds other criteria to the equation. This in effect makes a situation where you are looking for a dwindling random probability within the set. This in no way gives you the ability to determine (or predict) what the outcome of a roll will be, only what it "might" not be. there is also with true random number generation no "mathematical guarantee" that the 6 won't actually in practice come up that one more extra time in a row. (the odds are just in your favor).
This does not sound true to me, the rest of your point seems valid but that last statement is not a logical leap I can make. (at least as I see things)
Just more "Food for thought"
-Eamon
It is very common in war
It is very common in war games with dice. That one can predict the outcome with a high accuracy. It completely depends on the numbers that are required though. Good war games still have a good random outcome.
However, take a look at this table for example:
http://boardgames.stackexchange.com/questions/3514/how-can-i-estimate-my...
I had the same issue. But I had an army of 36 vs 35. It was a 80% chance (back then) that the army of 36 would win. And 36 vs <30 was not uncommon (99,9% victory chances). My simple "random" mechanics where flawed, big time. It didn't matter what the next roll would be between rounds, since the total outcome was "predictable".
Thus, to keep randomness to a maximum. Keep it to a minimum.
Personal, I consider 80% more predictable than random. The turning point is at 75% for me. While 50% is true random on a yes/no situation.
For a die with 16,7% chance per side, this is true random. The turning point is for me at about 58,3%, I know it sounds strange at first. But I have my reasons.
I see and understand
However, take a look at this table for example:
http://boardgames.stackexchange.com/questions/3514/how-can-i-estimate-my-chances-to-win-a-risk-battle
We are talking across the same points. I think we are on the same page but saying the same things in different ways...
The table you are talking about is in fact, a table of probability. The table must be referenced for the "current" odds AFTER each roll. This fact alone should show you that it does not at any point "predict" the outcome of a roll... or even a battle. It just gives the raw odds of winning for each game state that could exist between two players with 10 or fewer armies each.
What a player gets from this type of table is the "odds" based on the current game state, not a prediction of what "will" happen. I have never said that in a random system a player could not know the odds of any given game state. However, if a truly random method is being used, future game states cannot be "predicted".
As each turn "updates" the game state's "known information" the dwindling randomness "odds" becomes ever more slanted in the favor of whoever is winning. No one that I know needs a table to tell them that there is "virtually" no chance of defending a territory in Risk with one army.
A note from reality: I have played hundreds of games of risk over the last 30 years... One time I successfully defended a territory that only had 1 army against an army of more than 10 (I think it was 12 or 13)! This cries in the face of a table that says there is a 0.00% chance of that happening. This type of thing is where the expression "beat the odds" comes from. In all my life this may never happen again, as the odds of it happening the first time were stupid low.
Above is a single die roll with no other factors to consider. Having a number of armies and several dice changes the "odds" some but not the truth that the next roll is unpredictable. I do agree that in many war games that have variables other than a single die roll, the outcome of a conflict can be estimated base on the "odds" but not "predicted". As long as there is a chance, the outcome is unknown until it happens.
I do understand how probability works, how it is calculated, and how valuable it is. I also, understand how random works, that it can't be calculated beyond the odds, and how valuable it is. I worked for the Air Force in the field of cryptographic security for over 10 years. I am quite sure, while we may have some idea what "might" come next, no one can "predict" the next roll of the dice... other than God.
If we could know everything about everything it is "theoretically" possible for us to predict a die roll. But who knows everything about everything... I sure don't.
As always, I love to hear what others think about things, and hope my non-authoritative perspective is taken as "food for thought".
-Eamon
Well, I guess I look
Well, I guess I look different at things since do a lot in chemistry (and a bit with radiation and quantum mechanics).
So that quote that you have given from Tomasz Kapitaniak, kinda resembles me.
Here is another view. You roll an infinite times the die. You don't know the next roll. But you can predict the chance on what you might roll. And thus you can predict what you roll, when you roll all the time. Perhaps, I should say, you predict the outcome of a group of rolls.
The same goes for radioactive material. Each instable atom is unpredictable. There is a slight chance that it might radiate. I don't know, a chance of 0,000000000456% per second or something? But 1 mole (an unit of Avogadro’s constant=60.221.417.900.000.000.000.000.000.000.000 atoms/mol) is the dice pool.
You can calculate the average radiation. And let me tell you, the number is so big, that the accuracy is so high. That you can't beat the odds in the lifetime of the universe.
It would be funny though if a block of uranium (asuming one mole), suddenly completely changes. The chance on that would be the chance to the power of Avogadro's constant.
PS. Radiation is a form of dwindling. :)
Estimate vs. Predict
If you replace your some words like "predict" with "precisely estimate", (and change things a little) this is the best explanation for what we are talking about I have ever heard!
Predict: "Say or estimate that (a specified thing) will happen in the future or will be a consequence of something."
Precisely: "in exact terms; without vagueness."
Estimate: "roughly calculate or judge the value, number, quantity, or extent of."
What do you think of this "joint definition"?
http://itsadeliverything.com/accuracy-vs-precision-in-estimation
-Eamon
And what of the requirements of players in the role of random?
Eamon, you raise a considerable number of interesting points regarding the elements of randomness in board game design. I've been working with this very element in a strategic-like war game myself, where I've been tinkering heavily with the element of random input and output.
A point to consider, which I don't believe has been discussed thus far, is WHAT happens in a game after the random element effects play. While I personally enjoy a bit of randomness to a game, I don't always enjoy a random game factor which requires a multitude of steps to adjust and make changes to play. I would argue that the role of randomness works exceptionally well when its effect is able to be quickly integrated into play. Some games might be too 'heavy' or complicated to support this, I understand that, but where it can be done I think that a random input or output element can elicit an exciting "aha!" consistently throughout a game.
For example, in the card game Slap Jack, when players play the same card on top of each other, they can slap the deck and win all of the cards out in play. Likewise, a random element which causes players to quickly rethink their strategy, or to react mentally or physically to the sudden situation at hand, can turn a predictable play choice into an unpredictable shift for strategy's sake.
Great point
You bring up a great point here.
Dealing with random forces is for sure either a player's strongest or weakest points in a game when it comes to their overall strategy. Part of the issue is the way random forces used, and part of it is "how hard" the random element is inserted into gameplay.
This is the difference between needed to make a course correction because of unusually high winds and having your starship hit by a random asteroid that disables half your ships systems.
Just like in real life the more "impact" a random event has on things, the harder it is to build a strategy to deal with it before it happens and also how long it takes to build a new plan after it happens.
This also deals with whether or not a player "enters" the game fully understanding what random force strengths are like in the game, so they can plan for them. There is an element of "expectation management" to be done. A less experienced player might not "know" about the random asteroid card... so they don't have a plan for when they are hit by it.
As designers, we do need to understand the role we want randomness to play and also understand how that will affect the player's ability to make and stick to their plans. The more we use randomness at the right times (input & output) and with the right "scope of influence" (limited or subtle influence rather than outright surprising jolts) the easier it will be for the players to plan for and adapt quickly to those factors.
Thanks to all for taking the time to read and comment, this is how ideas become more real to us!
I need to work some of this thread back into the main article at some point!
-Eamon
I agree
That joint definition sounds perfect in my ears. Replacing a word is a good idea.
And I love the link that you have given. Those dart boards make me think of various games. Perhaps these dart board can be used to explain the differences between games as well. As using the whole dart board for a maximum in randomness. While closing in on the centre would mean this precise estimation.
I guess, more dart boards have to be made if more subjects are explained. Like a maximum in randomness (the whole board is used) and the absolute absence of it (All darts fly in the same spot).
Even balance and practical balance can be explained by some extra boards.
***
O man, I completely forgot about the size of an impact of an event. I have a rare find on this effect. Where imbalance can be achieved between several options.
It is resolved by now with several solutions, all used at once. But here is the link to the problem, if you are interested in reading:
http://www.bgdf.com/forum/game-creation/design-theory/new-weapon-propert...
But I think, that goes a bit off topic at the moment.