【30,000 Words Long Article】The game design is left-handed, complex systems and emergence

Author: Yu Tian, producer of "The Wandering Ark", WeChat account yutianTerean

Welcome to the author's personal public account "Fish Pond Game Production Workshop"

Zero, Introduction

This is an article about the application of complex systems in game design, and the whole article is divided into three parts, the first part introduces complex systems and emergence, the second part proposes a new way to measure games through complex systems, and the third part will delve into complex systems and the ways in which they emerge in design through actual prototyping.

Warm reminder, the number of words is 3w+, it is expected to take a long time to read, welcome to pay attention to the collection and look slowly, the words of a family, throwing bricks and jade, and also welcome friends to exchange axes.

1. Introduction to complex systems

First of all, what is a complex system?

There are many definitions of a complex system, but the more concise definition is that it is a system composed of a large number of relatively simple self-organizing individuals, and through simple rules of interaction between individuals, complex macroscopic behaviors and phenomena can emerge.

Ant colonies, brains, economic systems, immune systems, ecosystems, the internet, cellular automata, human society, and so on, are all complex systems.

There are two key words for the definition of complex systems here, self-organizing and emergent.

"Self-organization" means that there is no commander inside or outside the system who tells the individual what to do, but the individual himself controls himself through rules.

"Emergence" refers to the simple interaction between the components of the system, but the emergence of new characteristics or behaviors at the macro level, these new characteristics or behaviors are called "emergence". This is characteristic of nonlinear systems, i.e., 1+1>2, and another way of saying that the whole is greater than the sum of its parts.

Complex systems research is an interdisciplinary scientific research that covers many branches, and the following figure shows the distribution of complex systems science, which is https://www.art-sciencefactory.com/complexity-map_feb09.html from the source website:

【30,000 Words Long Article】The game design is left-handed, complex systems and emergence

The emphasis on the study of complex systems is relatively late and is currently being explored, and there is no accepted definition, unified theory, or general principle. The exploration of the commonalities of different complex systems has always been a top priority, and scientists in different fields have put forward their own theories, but there is no complete consensus.

We only focus on the application of complex systems in games, and generally speaking, the application of complex systems in games can be divided into two general directions:

Complex systems with agents (humans, animals, AI, etc.) at their core, such as ant colonies, corporations, social, economic systems, etc. Sandbox, story generator games are used more, such as Kenshi, The Sims series, Ring World, Dwarf Fortress, etc.
Complex systems with objective rules as the core, such as turbulence, weather, N-body motion based on physical and chemical laws, etc. Physics simulation games, games using active puppets, programming games, some construction and assembly games, etc

Of course, the two can also be combined, but generally due to performance and complexity considerations, there will be some emphasis.

The author's two recent games also make in-depth use of complex systems and emerging design elements, the most recent game "Wandering Ark" uses the physical system as a complex system to generate favorable collision displacement emergences for us, and is a complex system utilization with objective rules as the core.

In the last game, "Followers Lianmeng", the AI that combines GOAP and behavior tree was realized, so that the behavior interaction of the entire NPC forms a complex system, which is a complex system with agents as the core. For an introduction to GOAP, please refer to an introductory article by the author:

Hatps://mp. Vexin.K.com/S/U9CMJ2utev91Kaposhdfzark

All in all, complex systems provide me with a lot of different perspectives in game development, both in terms of technology and design, and are a very useful tool. Therefore, this article is to combine some of my practices and thoughts to share with you briefly.

In the next three chapters, we will unveil the veil of complex systems through the very representative chaotic systems in complex systems.

2. The Origin, Dynamics and Prediction of Chaos

People hate fatalism, but they always want to predict the future.

- I said

First, let's talk a little bit about dynamics.

Dynamics began with Aristotle's naïve but erroneous theory of motion, and later Galileo, Copernicus, and Kepler overturned Aristotle's popular theory for 1500 years with experimental observations, and then Newton came out of nowhere to invent calculus and formally created classical mechanics. The famous Newton's three laws were proposed:

In any case, all objects always remain at rest or in a constant linear motion when they are not acted upon by external forces.
The acceleration of an object is inversely proportional to the mass of the object.
The action and reaction forces between two objects, on the same straight line, are equal in magnitude and opposite in direction.

After Newton further developed the law of gravitation, Newtonian mechanics perfectly explained the motion of all objects at the time, and it applied to objects of any size, whether it was a falling apple or a burning sun. Theoretically, if people know the initial position and velocity of an object, they can calculate the subsequent motion, and naturally some smart children think: if they know the current state of all particles in the universe at a certain moment, then they can predict the state of the universe at any subsequent moment.

The above hypothesis of accurate prediction is the famous "Laplace demon", the mathematician Laplace said in 1814:

We can think of the present state of the universe as its past effect and as its future cause. If a wise man could know the forces of all natural motion and the position of all the objects of natural composition at a given moment, and if he could also analyze these data, then the motion of the largest object in the universe to the smallest particle would be contained in a simple formula. Nothing will be ambiguous for this wise man, and the future will only appear to him as it has been. ”

Nowadays, we have powerful computers, and we don't need a smart man who is very intelligent and close to demons, maybe in the future there will be computers that are powerful enough to do this job. If this is the case, isn't it true that everything in the universe is already predestined, like a clockwork that has been wound up, following the three laws and going on in a regular manner, what a desperate "deterministic universe" it is.

Fortunately, in 1927, Werner Heisenberg proposed the "uncertainty principle" in quantum mechanics, proving that it is impossible to accurately measure the position of a particle and its momentum at the same time. In this way, the accurate prediction of dream shattering in the microcosm also comforts those who do not want to believe in fate.

Statistical Mechanics and Prediction

Heisenberg's uncertainty principle is only for microscopic situations, while the statistical mechanics created by Ludwig Boltzmann tells us that even if the motion of each molecule at the microscopic scale cannot be calculated, it is possible to statistically predict the average position and velocity of a large number of molecules as a whole, and when the number of particles is sufficient, his method is "almost always right". Statistical mechanics is a bridge between thermodynamics and classical mechanics, and the second law of air pressure and thermodynamics can be well explained by statistical mechanics. So even if it is impossible to predict at the micro scale, is it possible to statistically predict the overall change of the system at the macro scale?

The answer is no, and now is where the famous chaos system comes in.

3. Chaotic systems

Sometimes a small difference in the initial conditions can make a big difference in the final phenomenon. A small error in the former will cause a huge error in the latter. Predictions will be impossible, and we are dealing with contingencies

—Poincaré, Science and Method, 1908

In 1959, Edward Lorenz (1917-2008), a 42-year-old mathematician and meteorologist, tried to predict the weather using a primitive computer, only to find that small differences in the input parameters could make the results completely different twice, even though the computer used the same method for each step.

后来，罗伦兹在一场学术会议上发表了论文来探讨这种对初始条件十分敏感的天气系统，论文题目叫——《巴西的一只蝴蝶舞动翅膀，可以引发得州的龙卷风吗？（Does the Flap of a Butterfly’s Wings in Brazil Set off a Tornado in Texas?）》，这就是著名的“蝴蝶效应”（The Butterfly Effect）。

The weather system is a typical chaotic system, and it may be a bit counterintuitive here, why are some systems so sensitive to initial conditions? The answer lies in the "nonlinear" and "self-referential" nature of chaos.

The following is a classic mathematical abstraction that can embody the essence of chaos - logistic map, which is very simple, and its characteristics are very shocking and fascinating. Here is a reference to the logical mapping chapter of Melanie Michel's book "Complex", which is highly recommended for everyone to read this popular science book on complex systems.

Here is the equation for the logistic mapping:

X(t) is the current value and X(t+1) is the next value. R is a parameter, we can adjust it ourselves, leave it alone. What we need to do is very simple, set an R, then start with an X(0) between 0 and 1, substitute this formula, get X(1), then substitute X(1) into the formula, and so on to get new values. This is a very simple "nonlinear" equation.

Let's try what happens when R=2 happens, and we find it interesting that no matter what value X(0) inputs, X(t) will eventually stay at 0.5, which is the so-called fixed point: the time it takes to reach this point depends on the starting point, but once it is reached, it stays still. When R = 2.5, you will also find that the system has reached a fixed point, but this time it is 0.6.

And when R=3.1, interesting things start to happen:

Regardless of the value entered by X0, eventually Xt oscillates between two values (0.5580141 and 0.7645665). If you substitute the former into the equation, you get the latter, and vice versa, so the oscillation goes on forever. This eventual change in position (whether a fixed point or an oscillation) is called an "attractor" because any initial position will eventually be "attracted into it".

Going up until R equals about 3.4, the logistic map will have a similar change: after iterating a few steps, the system will oscillate periodically between two different values (the final oscillation point is determined by R). Because it oscillates between two values, the period of the system is 2.

But if R is between 3.4 and 3.5, the situation suddenly changes again. Regardless of the value of x0, the system will eventually form periodic oscillations between four values instead of two. For example, if R = 3.49 and x0 = 0.2, the value of x will soon start oscillating periodically between four different values (they are approximately 0.872, 0.389, 0.829, and 0.494, respectively). That is, at a certain R value between 3.4 and 3.5, the final oscillation period suddenly increases from 2 to 4. The end result looks like the picture below.

Then, something even more interesting, at a certain R value between 3.54 and 3.55, the period suddenly doubled again, jumping to 8 at once. A certain value period between 3.564 and 3.565 jumped to 16. Between 3.5687 and 3.5688, the period jumped to 32. The cycle multiplied again and again, and the interval between the front and back R became smaller and smaller, and soon, at the time R was approximately equal to 3.569946, the period tended to infinity. When R is equal to about 3.569946, the values of x no longer enter oscillations and they become chaotic. Explain it below. Replace x0, x1, x2...... The sequence of values is called the orbital of x. In the creation of a chaotic R-value, so that the two orbits start from a very close x0 value, the result will not converge to the same fixed point or periodic oscillation, but will gradually diverge. At R=3.569946, the divergence is slow, but if we set R to 4.0, we see that the orbit is extremely sensitive to x0. Let's start by setting x0 to 0.2 and iterate on the logic sti map to get a track. Then make a slight change to x0 so that x0=0.20000000001, and then iterate on the logic map to get the second track. The solid line of the solid circles in Figure 2.14 is the first track, and the dotted line of the open circles is the second track.

The two tracks are close at first, but later on, they begin to separate and are unrelated. This is where the "sensitive dependence on initial conditions" comes from.

This is one of the simplest manifestations of chaos, a formula that tells us that such a simple logical mapping is completely deterministic: every xt value has and only one mapped value, xt+1. However, the resulting Chaos orbit looks very random. In addition, if there is any uncertainty in the initial condition x0 for the chaotic R-value, the orbit after a certain amount of time can no longer be predicted.

Therefore, if for an actual chaotic system, even if it is as simple as only one particle, if there is an irrational number in its initial state (the vast majority of numbers are actually irrational numbers, and rational numbers are only a few), such as the most familiar π, then we have to build a machine that can input an infinite number of decimal places, otherwise no matter how many decimal places we are accurate to, the error behind that digit will cause the entire chaotic system to be unpredictable after a certain period of time. This is "a small miss, a thousand miles".

At this point, it is impossible to predict the future from both a macro and a micro level.

4. Chaos systems and games

As mentioned earlier, our game "Wandering Ark" is a complex rule-based system, a simple description of the gameplay, is that players on both sides control five sphere heroes, collide with each other to fight, and their collisions and rebounds are simulated physics rules, and there is a PVP gameplay. And the mechanics of our game have actually formed a typical N-body system (yes, N equals three is the famous three-body system), and therefore encountered a chaos effect.

We were using a frame synchronization solution, and on different phones, we had the problem of two players being out of sync. After ruling out various causes, including network factors, it was found that the cause was a small (really very small) difference in floating-point arithmetic on different machines. As a result, a small difference in a certain round will make the battlefield situation of the two sides completely different after a few rounds, and it will directly become their own fight. Later, we had to abandon Unity's built-in physics system and re-implement a fixed-point accurate physics system by ourselves, which was very tragic. So, the game makes use of complex systems, and it also has to endure the problems that complex systems can bring.

Conversely, of course, we are also taking advantage of the emergence of complex systems. For games, emergences are not an end in themselves, but what emergences can bring to the game.

First of all, emergences can bring unexpected experiences to the game, parts of the experience that are beyond the player's control. If the developer can have some control over this part, they can do more. Of course, unexpected experiences are also new experiences, and emergences create new experiences in themselves.

At the same time, the beauty of emergences is that not all processes are unpredictable, and often a small part is predictable, which brings a certain degree of adaptability to players of different levels or willingness. However, the mutation of complex systems also brings about a critical point of phase transition, which in turn ensures that the control of emergences by different players does not differ too much.

For example, the physical collision system of our game, the complexity comes from multi-body motion, and players cannot predict the situation after N collisions, but most players can predict the outcome after 1 collision, which is the initial predictable part, which will enter the player's strategic considerations, and for some more strategic players, the second collision after the first collision rebound can also predict the part, which is adaptability, but when the rebound is three or more, basically no player can predict the situation, which can be compared to the critical point of chaotic phase transition, to control the strategic gap of different players, and even theoretically, even if there is a supercomputer similar to Deep Blue, and cannot fully control the situation after N becomes larger.

Chaos systems are just a branch of complex systems, but they are indeed a stepping stone to understanding complex systems, and after a general understanding of what complex systems are, let's take a look at the deeper relationship between games and them.

When designing complex systems, you will find that the boundaries between games and complex systems are very blurred, and a complex system, often with some reasonable goals and challenges, is a game with a certain degree of playability. And many games, in their own right, are complex systems. Therefore, some research on complex systems can also be applied to game design, for example, the study of "complexity and information" in complex systems is very valuable for game design. In the next chapters, I'll explore a way to measure and optimize your game, but if you're not interested, you can skip to Chapter 8.

5. Measuring "Complexity"

First, let's take a look at how scientists measure the "complexity" of complex systems:

1. Scale measurement

The simplest measure is size, but this is obviously not true, since single-celled amoeba has 225 times more base pairs than humans.

2. Shannon entropy

Another straightforward measure of complexity is Shannon entropy, which is defined as the average amount of information or "amazement" of an information source relative to the receiver of the information. That is, if the information is highly ordered, e.g. "A A A A A...... A", the entropy is zero. Completely random sequences have the maximum possible entropy. This measure is also obviously insufficient, and completely random sequences do not make sense to the recipient of the information. So the most complex objects are neither the most ordered nor the most random, but somewhere in between.

3. Computer Description

The third measure was independently proposed by Andrey Kolmogorov, Gregory Chaitin, and Ray Solomonoff, who defined the complexity of a thing as the length of the shortest computer program capable of producing a complete description of something. This is known as the algorithmic amount of information of things. It should be well understood for those who are learning computers, that is, how much code is needed to express something.

For example, if a game A can be written with at least 10 lines of code, and another game B needs to use 100 lines of code, then game B is more complex than game A in terms of algorithmic information. This seems reasonable at first glance, but in fact, there are limitations, is 100 lines of code necessarily more complex than 10 lines, and is the complexity of the code itself simple linear. Essentially, this is an abstraction of the complexity of the system to the code, and it does not completely solve the problem, but only turns the complexity problem of the system into the complexity of the code.

4. Statistical complexity

The fourth measure is a measure called statistical complexity, defined by physicists Crouchfield and Karl Young, which measures the minimum amount of information needed to predict the system's past behavior in the future. Statistical complexity is related to Shannon entropy, in which a system is defined as a "message source" whose behavior is somehow quantified as discrete "messages". Prediction of statistical behavior requires observing the information generated by the system, and then constructing a model of the system based on the information, so that the behavior of the model is statistically consistent with the behavior of the system itself.

For example, the information source model for the sequence ACACACAC can be simple: "repeat A C"; However, unlike entropy or algorithmic information, there can also be a very simple model for the information source that generates the sequence ACGTGGTAGC: "A, C, G, or T are randomly selected." "This is because the statistical complexity model allows for the inclusion of random selection. A measure of statistical complexity is the amount of information for the simplest model that predicts the behavior of a system. Thus, the values of statistical complexity are low for both highly ordered and random systems, and high complexity for systems in between.

The fourth measurement method is not only intuitive, but also worth thinking about is that it does not fully measure the system itself, but also measures the complexity of the system itself by predicting the evolution of the system.

There are many other ways to measure complexity, and I won't go into them all. All kinds of metrics have some feasibility, but they all have limitations and are far from being able to effectively characterize the complexity of the actual system. The diversity of measures also suggests that complexity thinking has many dimensions that may not be characterised by a single metric scale.

For most systems, such as games, it is not static, but constantly updated and iterated, and it is interesting to predict future possibilities to extrapolate the current complexity.

6. Complex systems and effective information of games

I believe that many designers have wanted to "measure" their own games when designing games, "playability", "complexity", "difficulty to play", etc., which are very difficult to define and quantify in a unified manner. In this article, we will not discuss the overly large topic of "playability", but only focus on the discussion of the amount of effective content of the game, which can also be measured from many angles, such as gameplay time, system complexity, system depth, etc. From our point of view, we measure and optimize a game from the information level, which I call "the effective information degree of the game".

There are many ways to abstract games, and one of them is that the player receives information, interacts with the game, influences the game, and the game responds to the information. Information is the ammunition for interaction, the basic material of strategy, known information as a source of security and input for decision-making, unknown information as a hook to attract players to explore, as an uncontrollable factor to bring about change, and so on.

"Game effective information" refers to the amount of information provided by the entire game system that is meaningful to players' expectations after being displayed through certain rules. The greater the amount of information, the richer the content of the game, the higher the playability and playability.

Of course, the information richness of the game is not exactly equal to the playability, but for all players or humans, there is a desire to constantly obtain new information, and at the same time, humans are efficient pattern recognition machines, when players are familiar with specific patterns and rules, and have not ingested new information, they will quickly get bored. Information stimulation itself is also an important part of playability.

Let's take a concrete example to explore the concept of effective information:

For example, now we've made a linear level A, where you defeat a different enemy in each level and go to the next level until you complete it, and there's no direct connection between levels.

Let's say we make a series of levels, such as 10 levels, and we expect players to play these levels once and no longer want to play them after completing them, let's assume that the game has a valid information level of 10.

With this in mind, there are a couple of interesting questions:

Question 1

At this point, add a setting that will disrupt all levels after clearing the level, assuming we expect the player to clear the level again because of this setting, which takes the player twice as long.

Will the effective information level of this game become 20?

If not, is it more skewed towards 10 or skewed towards 20?

Question 1: Explore

Obviously, it's not 20, the point is that the effective information degree is not equal to the game duration, but the player's perception of the content, the amount of meaningful information that can be received, and if the level is just disrupted randomly, for the player to play, the level information has not changed, in other words, there is no new information to the player, and there is no new experience. Therefore, the effective information degree of the game in question 1 should be close to and slightly greater than a value of 10. Note that we don't care about the absolute precision of the effective information (and we can't be completely accurate), we just need to try to quantify it and approximate the actual value by means of comparative analysis, etc. This problem also echoes the measure of complexity of complex systems, which are both pure random and pure order, with low complexity. (So the replayability of a good roguelike doesn't just come from randomness)

Question 2

As we already know in the discussion of question 1, the length and number of games are the result, not the effective information. On this basis, after all levels are scrambled, players can repeat the game all the time, and the order of the levels will change each time, so what is the change in the value of playing?

Question 2 Discussion

Essentially, what is consumed is the experiential value brought by different sequences

Echoing the previous information, pure randomness and pure order are both states with a low amount of information, that is, the game value will not be very high

At the same time, there is additional significance in discussing linear and nonlinear systems, if the correlation between levels is small, then there is no essential difference between A and then B or B and then A, and if the levels are non-linear and affect each other, then the order itself is given meaning, or additional information and complexity, it may produce more effective information. Therefore, the focus is not on the randomness of the level disruption, but on the design of the relationship between the levels, whether the randomness is given enough meaning.

Question 3

On the basis of the original plan, 1 more level was done, and it was set to clear the customs once, and then clear the customs again to play this additional level.

Assuming that each player reclears the game once because of this setting, what is the effective information degree of the game at this time?

What is the difference between playing 11 levels and giving them to players?

Question 3 Discussion

Similar to the discussion in question 1, perhaps the measurement of the game's playtime is close to double, but the effective information is only close to about 11, which means that the quality of the player's experience has actually decreased in the nearly doubled experience time. In fact, there is no essential difference between directly releasing 11 levels to players. So why do many other games use this technique to extend the play time? In fact, it comes from the replayability brought by the above game mechanics themselves, similar to the non-linear levels discussed in question 2, these additional factors make the game more effective information in other dimensions, so it is not within the scope of question 3. What this example wants to illustrate is that the analysis of effective information must be objectively stripped of other factors, otherwise it will lose its reference value.

Question 4

Or is it a game with linear levels, where nothing changes, but what about the effective informativeness of the game given an extra objective, a beat time, and then allowing the player to replay?

Question 4 Explore

In this problem, theoretically, the effective information degree has not changed much compared with the original scheme, but we intuitively think that the scheme is more playable than the original scheme. The point of this question is that intrinsic drivers such as skill challenges can bring different levels of playability to games with similar effective information. It's also an example of how effective information doesn't exactly equal playability.

Question 5

Based on the design of the additional clearance time objective in question 4, it is further explored that if there is an item in the game, the strength is average, but it is helpful for speedrunning. Therefore, when the player does not have a clearance time objective, and when there is a clearance target time, the effective information changes?

Question 5 Discussion

We can see that there is no difference except for the change in the goal, and theoretically, there is no difference at all in the amount of information. But in fact, it is clear that the latter must be more effective than the former. This is the meaning of "valid" in the definition of "effective information": a piece of information can only be called valid information after it is transformed into information that is meaningful to playability through the mechanism. See this in the next chapter.

After the above questions, you should be able to understand the significance of the effective information degree of the game, although it is not equal to playability, but it is a measure of the richness and effectiveness of the player's experience in a certain period of time. In particular, it helps us to think about the pitfalls of ineffective playtime, and to some extent helps us to compare and contrast game design.

7. Effective information and game design

How does effective information really help with design, and how should we think about and manipulate information in the game?

The process of playing a game can be seen as a process of receiving information, processing information, and obtaining feedback. The designer provides a series of information for the player to process, and the player can get different feedback through their own operations and strategies when processing this information. For a gameplay, it is a good way to play from an information perspective if the player is constantly and rhythmically given the right amount of meaningful new information for the player to process comfortably.

The information of a game is not all released to the player at one time, and the information of a game has many dimensions, such as information about the rules, information about the game world environment, information about levels and challenges, feedback and harvest information, and so on. In the feedback loop of the player's goal challenge, new information is constantly being released at each stage. And what is the meaning of new information? is to produce new changes.

However, as mentioned above, a new piece of information often produces changes that are not a new gameplay experience, and our goal is to make one piece of information produce ten changes, and finally transform it into two parts of playability.

With limited content, how to plan this information will be key. Let's talk about a few ways to do it:

1. Rules assist opponents in generating new information

Take the board game - chess as an example, there is not much information about all the rules of chess. The battlefield environment information (the current chessboard situation) is ever-changing, and we can easily know that chess does not rely on the gradual release of rule information to produce changes, and at the same time, chess does not have any randomness, so it is a good example to help us focus on this method - rules help people generate information.

The dynamic game between people brings continuous playability, which is not a mysterious thing, but from the perspective of information, there are actually requirements, people produce meaningful content, and they need rules to guide, what is guidance? For example, the restrictions on the way all the pieces move in chess are a kind of guidance, which allows the player to partially anticipate how the opponent will act next, so as to combine the scene, and finally give the player in the current round, a constantly changing but decision-making reference information, this information will be combined with the opposing player's playing style, level and other factors, resulting in changes between multiple games.

A good mechanism or rule that tells the player what he can do is basic, and more importantly, it restricts what the player can't do, compresses the infinite choice space to a reasonable range without strategy convergence, and allows the opponent to feedback new information for each choice, for example, there is no obvious optimal solution in a single step, whether it is a defense or an aggressive changer, in fact, there is hidden information about the opponent's style, this information, is based on the rules but beyond the rules of additional information benefits.

2. Assist players to generate new information by themselves

We can abstract the game into a three-way loop of goals, challenges, and feedback, and the overall game is made up of an infinite number of large and small loops nested. From an information perspective, we can look at the cycle from a new perspective, and each part can be designed to assist the player in generating new and useful information.

For goals, it is often the most explicit and simple information, and for a single loop, the information about the target is generally not deliberately hidden, because for the design, the clearer and clearer the goal of a single loop, the better. But on the whole, there can be multiple goals, which can be released gradually, and can guide the player to give the goal himself, and there is a very large space for manipulation, and the goal is often the starting point of everything, which will multiply the effect of other information.

The best way to do this is to motivate players to give themselves goals, such as a sandbox game, by packaging information such as world view, so that players want to climb to the top of the mountain that has nothing to do with the mission, and want to save an insignificant villager. The information provided by the designer is used as an external driver as a guide, and the player's spontaneously generated goals are used as a strong internal driver.

For challenges, there is a division into the challenge itself and the process by which the player overcomes the challenge, such as strategy and action. For the spontaneous generation of this part of the information, the challenge itself needs to provide a high enough adaptability so that the player's own situation can be reflected in the process of challenging and overcoming the challenge. At the same time, it is necessary to be sensitive enough, and it is best to reflect the subtle changes of the player.

For example, there are multiple ways to complete the challenge, and players can choose one of them according to their own preferences and abilities.

For feedback, the focus here is not on giving feedback to the player, but on the feedback given to the system after completing the challenge, and the focus is on the relationship between the different challenges before and after, that is, the subtle differences made by the players need to cooperate with the above challenge parts, and the feedback needs to make the system have enough influence on the next challenge, and there will be exponential changes. Such a combination is also an emerging design method. (Details in the following chapters)

3. Promote interaction between multiple people

As with the essence of the first point, there is a slight difference, leave it for thinking questions haha, and analyze the difference between mahjong and two-player chess. Interested friends can refer to the author's mahjong-related articles: https://mp.weixin.qq.com/s/RUFNdkrFy8e2oqEhsYp_Xg

4. Gradually release and reorganize information

Players discover information spontaneously, rather than designers stuffing it into the player, and the process of discovery itself is a meaningful act of acquisition that will enrich the entire game. This is a good way to release information gradually.

Reuse information by subtly reorganizing it. This is easier to understand, so I won't go into details.

8. Complex Systems and Emerging Applications

The above chapter briefly introduces complex systems and some of their relationships with games, and is inspired by complex systems to propose an immature idea for game metrics. The following chapters will take a closer look at the use of complex systems and emergent design in game design, using actual prototyping as an example.

Thought experiments

The first step is to do a thought experiment, as mentioned above, there are many complex systems in the real world, we might as well try to pick a complex system in reality, assuming that we are the almighty computer god (dog head), has implemented this system 100%, at the same time, as a designer, we know that simulation is not equal to fun, then now, we need to think about what is not fun in this system, remove, what is the fun thing, strengthen. Add a little spices, add a little goal, set some challenges, give enough feedback, and at this time, I believe we already have a game in our minds, which may be simple, but it can be established.

The process of abstraction and tailoring of reality is particularly important, and we will gradually peel back the cocoon, remove the unimportant externality, and see the core that emerges, and give random examples.

Suppose we take the real economic system as our complex system, and we first give a simple goal, which is how much money to make. Then we remove some of the redundant stuff and end up with a system like this:

There are a number of economically rational people in the system, including players, who will find ways to get the maximum benefit at the lowest cost, and these decision-making entities are collectively referred to as "agents" - agent is a term in the field of computers, artificial intelligence and economics, which refers to the subject who receives information, makes decisions, and performs behaviors, which can be people, AI, organizations, a program, etc. In this process, the agents will interact with other agents and the environment, and continue to exchange time, information, resources, and revenue. Imagine what is the source of the fun we get in this system when this system is in operation, and what is the difference between it and a normal trading simulation game?

This simple system puts aside redundant appearances such as acoustic and optoelectronic feedback, we can find that the player's core strategy is to collect information, make their own decisions to maximize profits based on the information, and their own decisions will affect the whole system, and the system changes the behavior of other agents, the behavior of each agent is quite simple, but there will be many interesting phenomena as a whole, such as robbery, monopoly, inflation and deflation, etc., which make the challenge/problem/puzzle dynamically change, like multiple spiders playing the same spider web.

Above, this complex system is providing meaningful variation and generating new playable content, which is fundamental to replayability. At the same time, the feedback will be stronger because the player's impact on the environment is profound enough.

Let's add a little more material to this game, if the game adds another player, and both sides have the same goal, then it is easy to find that the uncontrollable part of the complex system is balancing the advantages and disadvantages, narrowing the distance between players of different levels, and partially lowering the threshold of operation and strategy.

Further, we will find that, similar to the real economic system, our rudimentary complex system is itself an adaptive system, sometimes like a sponge, absorbing the major actions of players and agents to reduce the impact, and sometimes like a trumpet, amplifying some small noises. At this time, if we add some guiding designs to this simplified version of the economic system, such as the regulation of prices and resource consumption (similar to the central bank and the Federal Reserve), we can manipulate its sponge or horn to achieve the purpose of changing and controlling the pace of the game.

Taking it a step further, we are not satisfied with a single objective, if we add multiple victory conditions, we will find that the multi-objective based on the complex system itself is better designed to meet the needs of different types of players.

In the last step, let's remove the opposing player, compare some of the actions that the player will perform with the behavior of the agent, and assign some simple behaviors that the player will do aggressively, or irrationally, to the agent. And by changing the goal of the confrontation with the player to the confrontation with the agent, we will find that compared with other games, the agent will be closer to the real player in the player's perception.

When a player is confronted with a complex system as a whole and fights against some of its agents, it is more likely to rate the intelligence of the system than against a single NPC. Because the player's evaluation of the intelligence of an object, in addition to the intelligence itself, is also composed of many factors that are not related to intelligence, such as appearance, strength, quantity, environment, etc. For example, the Halo Designer shared a point in GDC that if you simply increase the strength of the enemy without changing the AI, the player's intelligence rating of the enemy will be higher. For example, because Palu is far away from the image of people, players have low expectations of his intelligence, but instead think that his intelligence performance is good. The uncontrollable and explainable of complex systems is a better "difficulty amplifier", and at the same time, the feedback of amplified behavior of complex systems also applies to agents, which is to explain and mask some non-intelligent behaviors.

Through the above thought experiment, we can roughly summarize the advantages of complex systems and emergence, which is what we designed them for:

Generate meaningful change to deliver new experiences
Enhance player impact and feedback on the environment
Balance the advantages and disadvantages, shorten the distance between players of different levels, and lower the threshold of operation and strategy
Change and control the pace
Carry the play needs of different types of players
Partially substitute for other players

In the above thought experiment, each friend may choose different complex systems, but when we think about several more complex systems and abstract the real world, we will find that complex systems and emergence can be divided into two biases:

Complex systems with agents (humans, animals, AI, etc.) at their core, such as ant colonies, corporations, social, economic systems, etc.
Complex systems with objective rules as the core, such as turbulence, weather, N-body motion based on physical and chemical laws, etc.

Of course, the two are also a mixture, which will not be discussed here, but only these two categories, and the core difference between these two categories is, of course, the existence or absence of the agent subject. For games, the design approach of these two biases is very different, the first type of agent core games are Sims, SLG, etc., while Minecraft, noita, and our own game Wandering Ark, are more biased towards the second category.

Next, we'll take a look at some of the best of the emerging designs through seven gameplay archetypes. These are paper prototypes, only need a simple pen and paper, or some simple props such as dice can be realized, interested friends can follow along to play, will have a deeper understanding. If you are not interested in design, you can skip directly to the conclusion and discussion part of Chapter 16.

9. Prototype 1 emerges in random succession

1.1 Design Purpose

Prototype requirements

Try to keep the rules as simple as possible, using only simple numbers, addition and subtraction
The basic game that can be established will not have a strategy convergence
It can provide a certain depth of strategy and widen the gap between different players
There is room for randomness, and randomness is used to simulate emergence
It can show replayability, balance of advantages and disadvantages, and other characteristics

Random and emergent similarities and differences:

Since one derives chaos from determinism and the other derives chaos from randomness, the former is much more controllable than the latter's perception of the player, which means that the player will feel that the emergent outcome is more related to his or her own decisions or inputs. At the same time, the player's attribution of the two outcomes will be different, the random result will be more attributed to the system, and if the emergence, if the design is good, the player will be more attributed to himself. This feature is a very important point for emergence over randomness, and it is also very important for game design.
The first point is the degree of perceptual control, and in fact, for the player, the emergence part is theoretically controllable, and the player will be more adaptable to different situations.
The two are completely different from the deterministic to the uncertain starting node, random * random, a certain layer of random, often the starting point of chaos, relatively discrete, and the emergence is much continuous, which is also very meaningful to the design.
Emergence is deterministic, just like the logical Sidi equation shown above, each iteration is deterministic and calculable, but the part that exceeds the computational power is chaotic and unpredictable for the calculator.
For random, even if the probability of different random results is known, for the calculator, it is still an uncontrollable result, if a lot of heavy randomness is superimposed, that is, we often say random * random, at this time it is more likely to have unclear expectations of probability, so completely abandon this part of the calculation, become chaotic unpredictable, for this part, random and emergent in the results are similar.

However, there are still several very important differences between the chaotic parts of the two:

There are also some similarities and differences, summarized in the table below:

In summary, because randomness and emergence are in the uncontrollable part, there are certain similar characteristics, and the difficulty of random implementation is much less than that of emergence, so the prototype is generally used to replace the emergence, so the feasibility is verified by the prototype first.

1.2 Gameplay Rules

Overview of gameplay

Double play, there are three battlefields, each player has four cards, a total of four rounds, each round both sides put a hand card to the battlefield, the final settlement points, the three battlefields win and occupy more than two players win, and the draw is won by points.

Round Phase:

Each side draws 4 cards at random from 6 cards of 1-6
3 field effects are randomly selected from the 5 field effects, and they will not be revealed yet
Reveal the first field effect, where each player chooses to put a card in their hand on that battlefield, face up, and then reveal
Repeat until 3 cards are placed on each of the 3 battlefields
On the 4th turn, both teams place the last remaining card on any battlefield, and both sides do not know the opponent's placement (cover it with two useless cards)
The last card is revealed, and the result is settled

Settlement stage

If the number of winning fields is the same, the total number of points is compared, and the player with the highest total number wins

If the total is the same, it is considered a tie

For example, if there are 1 and 2 in the field, and the field effect is [number less than or equal to 3 + 1], it will eventually become 2 and 3

Each number is corrected individually according to the effect of the field
Compare the sum of card points of both teams in each arena
The side with the higher sum wins the field, and the same is even
The player with the most winning venues wins

5 field effects

Odd-numbered cards result +4
Even-numbered cards result +3
Cards greater than or equal to 4 get +1
Cards with less than or equal to 3 get +4
Roll a six-sided die, and the number of points you hit is considered 7 on that battlefield

1.3 Design ideas

foundation

Create 1 archetypal gameplay with simple numbers
The minimum game is 1 number for each side to match the size, and the bigger wins

Join strategy confrontation and strategy depth

Tier 1: Introduce a best-of-3 Colonel game to test players' ability to allocate resources
Tier 2: Add a 4th card, which can be used as a key card to turn the tide or bluff to test the player's gaming ability
The two layers of strategy are multiplied to create sufficient strategy depth

Random and anti-random

Fight randomness through resource allocation
Fight randomness by reasoning
Fight randomness through games
Consider the subsequent possibility of the site effect to allocate resources and counter the randomness of the site effect
Reason the number of the opponent's last 1 card based on the opponent's first 3 cards to counter the randomness of the opponent's hand
In the case of a disadvantaged hand, develop a high-risk, high-reward strategy to bluff and steal chickens to counter the randomness of our hand
In the second round of random, the randomness range is small, and the two-layer small randomness only carries a little replay and strategy sliding, and does not add too much strategy burden
Equalize the value of each number through the field effect, fight against the first round of random, reduce the impact of random hands on winning and losing, and prevent strategy convergence
In the same way as the first round of randomness, after revealing the first effect, there are only 6 subsequent combinations, and after revealing the second effect, there are only 3 options in the last 1 field, and players only need to decide their own resource allocation strategy among the single-digit options
[Even+3], [Odd+4], Odd Extra +1, Odd and even Values are guaranteed to be the same
[Large number + 1], [Decimal number + 4], the decimal number is 3 points more than the large number, and the value of the large number is closer
Random 1 number becomes 7, the third round of small random, increases the value of waste cards (cards that cannot eat other field effects), and provides players with the possibility of turning the tables or bluffing
The first round is random, the randomness range is small, and the uncontrollable parts are limited
In the final battle stage, there are only 3 options for the opponent's last card, and players can reason according to the 3 options
Guaranteed that for any hand, one side must have at least one odd and even number, at least one small card and a big card
Card 6 with 4
Field effect 5 with 3
Player-versus-random confrontations

1.4 Comparison Optimization with Poker

Prototype 1 has implemented a random alternative to emergences in a more elegant and concise way, but there are still shortcomings, let's compare them with poker:

Information exposure density

Optimization: In the turning stage, both sides take turns to play cards, reduce the exposure density, and the back-hand players can play cards according to the cards played by the first-hand players and the effect of the field, which reduces the amount of reasoning of the back-hand players, and the back-hand players have achieved an information advantage in the ground-turning stage, and the advantages and disadvantages of both sides need to be balanced
However, the above optimizations introduce a priority problem.
Poker information exposure frequency is high, the amount of information is small, every time a card is flopped, every time a bet is added, a layer of information is exposed, and the relationship between the front and back is close, which is convenient for players to reason and the burden is low.
Prototype 1 has a low frequency of information exposure, but a large amount of information, with 2 cards turning each time, doubling the amount of analysis, and it needs to be reasoned through global cards and land cards, which is a high burden.

The degree of information controllability and the scope of information

Optimization: After randomizing the field effect, directly reveal all the field effects, or reveal the first 2 first, reducing uncontrollable information.
However, it is better to add a raise setting that does not reveal the field.
The river is known in poker, and the uncontrollable information is the opponent's 2-card hand, and the number of players will be smaller and smaller, and the uncontrollable information will gradually decrease
In prototype 1, the player has 4 cards, and the opponent's card and card distribution strategy is uncontrollable information for the player, doubling the number of pokers, but because there is only 1-6, the information range will be relatively reduced
There are also 3 field effects in Prototype 1, and the field effects are being revealed one after another, which is also part of the uncontrollable information before it is revealed

Victory conditions

Poker win conditions are simple, with card suits and numbers for size comparison, which is easy for players to handle
The victory conditions of Prototype 1 involve the calculation of the winning area and the calculation of points, which are not easy for players to deal with

1.5 Prototype 1 Summary

Prototype 1 already has a certain playability and depth through very simple rules, as well as replayability. By simulating the emergence with two very controlled randomness at the same time, the two characteristics mentioned above can already be felt:

Generate meaningful change to deliver new experiences.
Balance the advantages and disadvantages, shorten the distance between players of different levels, and lower the threshold of operation and strategy.

This is a very simple prototype that shows how to use random to replace emergence, and the general feeling of emergence, and we can also use this prototype as a basis to expand the rules, such as adding poker betting:

Just add a rule, for the first three rounds, players can reveal cards on the battlefield after each round, the higher player bets, and the opponent must call to continue to the next round. At the same time, one side can also give up at any time, and the other team takes all the chips.

In this way, a psychological game is added on the basis of the original strategy, and the strategic fun of information exposure in each round is brought into play, and the upper limit of the strategy is raised.

Prototype 1 is a simple rule-based emergence-based design presentation, which will start with an abstract real-time combat prototype and gradually explore agent-based emergence.

10. Prototype 2 real-time combat abstraction

2.1 Design Purpose

Let's try to abstract a typical real-time battle, simplifying the battle to a 1v1 between two players, each controlling a unit, and the goal of victory is to kill the opposing unit. Various other spin-offs can be considered as variations of the smallest battle in a single round. To further abstract it, the player is allocating all the resources in the battlefield to achieve the goal of defeating the enemy, and divides all the resources of both the enemy and the enemy into the following types:

Survivability that is directly related to victory or defeat, such as HP, endurance, etc
The output ability that directly correlates with victory or defeat, direct and indirect damage, limits the enemy's endurance, and so on
Variants of survivability that directly correlate with victory or defeat, such as mobility, the ability to limit enemy mobility (both survival and output)
An indirect version of the first three abilities, such as the abilities of summons or environments, or the abilities they give to key units

Like the previous prototype, we need to use the simplest rules, only the addition and subtraction of numbers, to simulate this combat model, and this prototype also needs to achieve the following purposes:

The basic strategy game that can be established
There is enough accurate abstraction for real-time 1v1 combat, especially for attributes closely related to combat strategy, such as mobility
Robust enough for expansion to serve as the basis for subsequent emergent designs

2.2 Rules of Gameplay

Overview of gameplay

In two-player battles, players invest resources in output/survival/mobility attributes, and mobility attributes can be allocated again, allocated to [attack] and [dodge], both sides attack each other, and the damage caused is determined by the resources allocated by the player, and the first to return to 0 HP fails

How to play

For every 1 point higher than the attacker's [Dodge Points] and the attacker's [Attack Points], the damage dealt by the attacker is reduced by 20% or up to 100%, and if the attack is not high, the damage dealt by the attacker remains unchanged

Allocate 15 base points to the output/survival/mobility stats of the respective combat units, and then announce the distribution
Output attribute*10 as the combat unit's attack value, and survival attribute*100 as the combat unit's health (not necessary, for the sake of simplicity, reduce the decimal calculation)
At the beginning of each round of battle, the base points of the mobility attribute are distributed once (the team with the higher mobility attribute can get an additional 1 point), which are allocated to [Attack] and [Dodge] behavior, and then the distribution status is announced
The side with high mobility attributes will attack first, dealing 1 damage to the opponent according to the attack value, and the damage needs to be corrected 1 time according to the behavior points of both sides
After the attack of both sides ends, if no unit HP is 0, start again from step 3, and the HP of any unit will be 0, ending the battle

2.3 Design ideas

The smallest unit is used first, and the two combat units attack each other

Simulate the instant with discrete rounds and get to the essence more directly

Abstract the strengths and weaknesses of combat units into 3 dimensions, output/survival/mobility

Introducing points, players need to allocate points into 3 dimensions to simulate battles with simple numbers

If the mobility is high, the output/survivability will be reduced
The first round of allocation, by limiting the sum of points, simulates the effect of maneuverability on output/survivability

After the first round of distribution, the output and survival points are fixed, and the mobility points can be redistributed

The mobility ability can be used to chase down opponents and increase output
Maneuverability can be used to dodge attacks and increase survivability
If the number of maneuver points is slightly higher than that of the opposing side, you will have an advantage in each round of attacks, but if you exceed a certain value, you will no longer receive benefits
The redistribution of maneuver points affects the damage of combat units, again simulating the effect of maneuverability on survivability/output
The redistribution of maneuver points, which is allowed to be different for each round of battle, simulates the ability to transform maneuver ability

2.4 prototype 2 multi-V multi-expansion

It is very simple to extend Prototype 2 to the rule of many VS many:

There is an additional target selection phase, where units with low mobility start selecting first (random when equal), and cannot be changed this turn after selecting a target

After the target selection is completed, the stage of mobility allocation begins, and the rules are supplemented:

Mobility attack points can only be assigned to the unit's previously selected target, while evasion points can be assigned to all enemies targeting the unit

However, for each additional target assigned, mobility costs an additional 1

Once all units have been assigned mobility, the battle begins

The order of battle settlement is that all of them can be settled at the same time

The rules at settlement are the same as the basic battle rules

2.5 Prototype 2 Summary

By allowing players to allocate limited resources in battle, two characteristics of mobility are revealed

It can be freely converted into output/survivability, which has an effect on output/survivability
When the maneuverability is too much higher than that of the enemy, it will become a disadvantage

In Prototype 2, it can be easily expanded to 1-to-many and many-to-many gameplay, further simulating the situation on a complex battlefield, which preliminarily proves that Prototype 2 has strong scalability

The allocation of resources can also reflect some of the design ideas in the battle game

Action Commands in MOBA Games (Move, Attack, Release Skills)
Action Commands (Attack, Defend, Release Skills) in turn-based games
The cost of actions in card games
And when designing the style of combat units, resources are also allocated to these three aspects
There can be some space for players to make allocations, such as adding a talent system and a development system
All the strengths and weaknesses of combat units can be abstracted into three aspects: output/survival/mobility
The mobility attribute that can be redistributed in battle creates a strategic space for players in the game, allowing players to make changes and create different game content

The prototype 2 strategy game is divided into two layers, one is the allocation of three initial combat resources at the beginning, and the other is the use of mobility in the battle.

11. Prototype 3 preempts resources

3.1 Design Purpose

At this point, we have an abstraction of a basic battle, and before archetype 3, we first integrate the ideas of archetype 1 into archetype 2, and again analyze the emergence from another angle

The changes between the player's existing units, which are unified with the player's changes and affect each other, are abstracted as:

The player is weakening the enemy player's survival, output, and mobility abilities, with the ultimate goal of zero survivability

At the same time, the player defends the enemy against the weakening of these three abilities

In the above process, there is an impact on the environment and feedback from the environment

Feedback from the environment can be beneficial to us or to the enemy, and is divided into controllable and emergent parts

1 controllable part

Players consciously pursue controllable parts that benefit them, and need to be as simple and quantity as possible, as the environment itself needs to be as fair as possible to both parties
The real purpose of this part is to stimulate the player's interaction with the environment as a trigger for emergence
For example, if a player cuts down a tree to pick apples, the tree falls and hits the player or the enemy player
It is worth noting that the controllable part is best designed to be consistent for both players, and it is natural to distinguish whether it is beneficial or not due to the various differences of players, so as to reduce the complexity of information

2 emerging parts

The uncontrollable part is beneficial to the enemy or to us, but in terms of design, the emergence of the controllable part that tends to benefit us is equally beneficial to us
At the same time, the emerging part, the player is not completely uncontrollable, but has a low degree of control, and the controllable part is a gradual change process, which is also used as the upper limit of gameplay
Finally, the emergent part needs to be as controllable as possible, such as tending to help vulnerable players, etc., but it needs to be as reasonable as possible

Ignoring prototype 2, we abstract a minimal model from scratch that fits our gameplay to implement the above theory:

The simplest uncontrollable parts are expressed in random numbers, and the higher the number, the greater the advantage
The simplest controllable part is consistent with the core mechanics
Abstract the player's abilities into three parts, output ability, survivability, and maneuverability, expressed in numbers
The side with high mobility has an advantage, but the marginal benefit is rapidly decreasing
The mobility ability has its own role, as well as bonuses to survival and output

At the beginning of the battle, the two sides freely allocate 9 points to the three abilities

At the beginning of the battle, a random 1-M is made based on the number of points M allocated to mobility, with an additional 1 for the larger side of both sides

Both sides can freely allocate the final points M to output and survivability as temporary output and temporary survival

Both sides use their own output to damage each other's survival, and the temporary survival is deducted first, and the round is over, and the temporary output and survival are cleared

In the round-robin round, the side whose survivability reaches zero first loses.

Then we'll emerge abstracted as:

Temporary bonus, permanent bonus, temporary cost, permanent cost

Choose your own risk, high risk and high return

Test class, when the output or survival or mobility is greater than or equal to N, a bonus X is obtained

Temporary cost and temporary bonus class, output or survival or mobility temporarily minus X, temporary bonus X

Permanent cost permanent bonus class, output or survival or mobility permanently minus X, permanent bonus X

Finally, we break down the emergence:

The player-controllable interaction part includes player-controllable interactions, and the feedback given by the system emerges. The player can give the emergence system a resource information, current/maximum/base/temporary output, survival, numerical information of mobility, b resource itself, consumption or change output, survival, mobility itself, and c. player's choice or operation interaction.
The influence of controllable interaction on emergent systems, and the self-organization of emergent systems. The logic within the emergent system itself needs to support the acquisition of information from the outside world and the exchange of information internally.
Emerge the system makes changes spontaneously, and the feedback generated by the player.

Based on the above analysis, let's take a look at the actual rules of Prototype 3:

3.2 Gameplay Overview

Based on the base combat of Prototype 2, it is changed to generate three temporary attributes, temporary attack, temporary life, and temporary mobility every turn

Three random resources appear in turn in each round, and the two sides invest their three temporary attributes to compete for the bidding mode

Once you've fought for three resources, use the remaining temporary stats for a basic battle

3.3 Specific Rules

At the beginning of the battle, both sides freely allocate N attributes to attack, health, and mobility, which are permanent basic attributes, and the results are revealed
Three resources are randomly selected from several types of resources and placed on the field, without being revealed
At the beginning of each round, a resource is revealed, and three temporary resources corresponding to the number of points are generated according to the current base stats
The two sides start bidding for this resource, one side can bid for N points of temporary attributes, the other side can choose a temporary attribute greater than the number of points of the opponent, or give up, and so on
The side that obtains the resource will trigger the effect of the resource immediately, and after the resource is triggered and settled, the next round will be carried out.
When all three resources are contested, both sides use the current temporary resources for a round of battle, and the battle rules are the same as in Prototype 2, and the additional damage that exceeds the temporary life is deducted from the base life attribute.

Specific effects of resources

6 Resource Consumption Cards, Temporary Attack/Life/Mobility, Permanent Attack/Health/Mobility

6 Resource Gain Cards, Temporary Attack/Life/Mobility, Permanent Attack/Health/Mobility

Each time a random resource effect is effected, one card is randomly selected from the consumption card and the obtained card to form the effect of the resource

Each combination has a different settlement effect, as follows:

1. Consume Temporary XX and get Temporary XX

The cost of the player's choice of N points from 1 to 10 cannot be greater than the number of points he has

Roll a 10-sided dice, the number of points needs to be greater than or equal to N to succeed, and the harvest after success is 2 times the price, and the failure price will not be returned

Example:

Draw a temporary attack to gain a temporary maneuver, at which point the player has 6 temporary attacks

Players can choose the consumption of 1-6, for example, if 5 points are chosen, then 10 dice will be rolled, and if it is greater than or equal to 5, it will be judged successful

The player deducts 5 points for temporary attack and gains 10 points for temporary maneuver, otherwise, 5 points for temporary attack are deducted and nothing is gained.

2. Temporary replacement

The player chooses the permanent cost of N points from 1 to 10 (minus the base attributes), and N cannot be greater than the number of points they have

10-sided dice need to be greater than N to succeed, and the harvest after success is 5 times the price, and the failure price is not returned

3. Temporary replacement for permanent

The player changes 1 point of permanent resources, the player has N points of the temporary resource, and the result of 10 dice is less than or equal to N to succeed

For example, if the player draws a temporary life for a permanent life, and the player has 7 temporary life, then a 10-sided dice throw of 7 or less will give 1 fixed life

This temporary resource is only used for judgment and will not be consumed

4. Perpetual for permanent

Fixed 1 resource for 2 permanent resources

No judgment is required, and 1 point of the corresponding resource is directly consumed in exchange for 2 points of the target resource

3.4 Design ideas

The actual scenario simulated by this prototype is that if the emergence factor of the environment is introduced, the player will actually invest temporary resources during this period of time to gain environmental advantages, such as classic mobs, river buffs, or occupying an advantageous terrain. For example, when the player moves to an advantageous terrain, it is actually to exchange a part of the mobility for the terrain advantage, and if the terrain advantage is to add defense, it is actually our temporary mobility for temporary life. In the same way, all behaviors that do not directly exchange battle losses with the enemy in battle can be abstracted with this archetype.
At the same time, one of the goals of the prototype is to achieve emergent adaptability in a minimalist way, especially to changes in the strengths and weaknesses of the battle situation.
Introduce players against uncontrollable experiences, by giving players the freedom to choose risk-reward strategies
At the same time, the player's decision-making indirectly affects the outcome of the system and the subsequent battle situation through randomness, partly simulating the process of emergence
A simple normal distribution is achieved with a 10-sided die, which ensures that the return is maximized in the middle number, with low value, low risk and low return, and high value and high risk

3.5 Prototype 3 Summary

Just talking about this prototype, there is some room for optimization, such as:

1. The random acquisition of survivability resources will slow down the pace, we can remove the random acquisition of permanent survivability in resource acquisition, and go further, we can directly remove the temporary vitality, and only retain the setting of temporary mobility and temporary attack.

2. The mobility supplement formula needs to be optimized, and it is not actually the output or survival to drop a little, and the mobility to increase a little. After the calculation, the dynamic design can be maintained, but it is necessary to ensure the distribution of about 15 points and the balanced distribution of 555 as much as possible, so the error is small

Finally, it is highly recommended that you get started and play the prototype 3, in which the design has a deeper experience of playing with your own hands, the prototype 3 already has a certain degree of playability, replayability, emerging balance advantages and disadvantages, the player's influence and feedback on the battle environment is also reflected in a part, at the same time, even if there is no central control system, the prototype also reflects a part of the rhythm control ability, which is quite elegant.

At the same time, let's recall the complex system and the purpose of emergence, which is also what we achieve:

Generate meaningful change to deliver new experiences
Enhance player impact and feedback on the environment
Balance the advantages and disadvantages, shorten the distance between players of different levels, and lower the threshold of operation and strategy
Change and control the pace
Carry the play needs of different types of players
Partially substitute for other players

This prototype, to a certain extent, has already embodied the first 4 advantages. However, this is just a relatively simple prototype, and if we think of a resource as a unit, the current solution has only one input, the input of a single player. There is also only one output, the output to a single player. This is a typical example of regular class emergence, so next, we'll try agent-based emergence-based prototyping.

12. Prototype 4 nine-square grid emerges

4.1 Design Purpose

Let's assume that there is a third-party unit as an agent in a single round of battle, and this kind of unit needs to achieve the six purposes of the complex system, so how can they achieve it in the simplest and most direct way? How can the requirements for the complex system be deduced from the purpose? The specific analysis is as follows:

1. Generate meaningful change to deliver new experiences

Provide a new validation scenario that puts the points we want to test the player on these units. New verification scenarios, such as if you want to test the player's shape, or other things that can't be tested in a simple enemy unit, you should first consider whether you can put it on the enemy unit, and if you can't put it on the enemy unit, then consider putting it on a complex system.

Thinking about it the other way, what the enemy unit can't do is what it should be done by a complex system. For example, the enemy unit is determined by the opposing player, and some controllable and stable needs cannot be realized, and the system is determined by the designer, which can be more controllable. For example, enemy units will try to favor the enemy's advantage, and the system can carry a neutral and equal design for both sides.

2. Enhance the player's impact and feedback on the environment

As part of the feedback, it is to give feedback to the player as clearly and magnified as possible about any actions done to the player, and it will be meaningful to the battle situation. Feedback requires complex systems to be sensitive to information such as player behavior, and needs to respond in a way that is consistent with the player's perception

3. Balance the advantages and disadvantages, shorten the distance between players of different levels, and lower the threshold of operation and strategy

Closing the gap between players and helping the weaker side requires a complex system that can identify and react to the strengths and weaknesses of both sides

4. Change and control the pace

Agents need to change and control the tempo at the right time, such as speeding up the pace during the endgame, etc

5. Carry the play needs of different types of players

Players with different purposes are designed according to their goals, such as killing AI units to get more money, breaking the ring to attack the audience, etc.

For example, the goal of dividing players and shaping a variety of single-game pursuits outside the game can be reflected in the game.

For example, in addition to defeating the enemy, similar to coc to engage in resources, pave the way for the next game, make connections between games, and give weak players a compensation, but it is more suitable for asynchronous PVP.

For example, the correlation between individual games can be to give the agent the ability to remember, and the impact goes beyond the single game based on the player's past behavior.

For example, if there is an NPC faction that has a favorable opinion of the player, and encounters it randomly in a single game, it will behave differently towards the players on both sides depending on the favorability of both sides

Therefore, players can have a third goal in the game, brush the faction favorability, or change the opponent's faction favorability, or the faction can issue a mission and complete the task as the third goal.

6. Partial substitution of other players

Complex systems need to be challenging enough for the player as a whole, and the agent needs to have the same underlying rules as the player.

The above is our goal, step by step, next, we will introduce the simplest agent, the agent's behavior is also very simple, pick a suitable target, just fight. At the same time, we need to use simple rules to make multiple agents generate emergent behavior in the whole.

4.2 Gameplay Rules

The battlefield is a nine-square grid map. In addition to the players on both sides, there are 7 neutral units on the map, and the attack attributes and life attributes of neutral units are randomly randomized with 10 dice at the beginning of the game, and the mobility is fixed at 0 and ignored, and the player units are the same as the aforementioned prototypes, and 15 attributes are freely assigned. As follows:

At the beginning of each turn, traverse all neutral units in the nine-square, attacking all nearby non-neutral units

The logic of neutral target selection is that the total output of the cell is the highest, the survival is the highest, and the random

When an offensive unit has a player unit, the player can choose whether to use a mercenary unit or the main body to defend first

The rules of the battle are the same as the base battle of Prototype 2, and the maneuver of neutral units and mercenary units is considered 0

During the player's action phase, the player with high mobility moves first, and can spend a little mobility to move to nearby tiles or not move

After the movement phase is completed, if the cell is an allied unit and nothing happens, the enemy unit or neutral unit will fight

If it is a neutral unit, it will fight directly, and if there is an enemy unit, it will fight with the enemy unit, and the battle rules are the same

If there are both enemy units and enemy mercenary units, the enemy can choose who will fight them, and after the battle is over, the enemy can choose to continue attacking the next turn

When a player defeats a neutral unit or an enemy mercenary in the slot, if there are no other friendly mercenaries in the slot, they can choose to consume their own attributes to generate a friendly mercenary unit, which is twice the cost of the mercenary unit and has a fixed mobility of 0

For example, if the player's attack is 5 and health is 8, they can consume N attack and M health to generate a friendly mercenary unit with 2N attack, 2M health, and 0 mobility

When the player finishes the action, the round enters the end phase, and if there is a vacant seat on the map during the end phase, a neutral unit is randomly generated

4.3 Prototype 4 Summary

The above is the most basic proxy simulation, in which it is reasonable to use the consumption of movement power to prevent snowballing, and the neutral units of prototype 4 have had a certain meaningful impact on the game, and can even "emerge" the simplest pinch, but don't underestimate this simple pinch, it is similar to the cellular automata behavior in Conway's life game, which is caused by the simplest rules (the neutral unit in the prototype selects the target rule), some phenomena that appear on their own. The emergence in this archetype may be so simple that it can be seen at a glance, but it still reveals the essence of the emergence.

Although Prototype 4 is basically established, there are still the following problems:

It is unreasonable to consume mobility for map displacement, it is not a mobile resource at the same level, and it is necessary to design a large map movement ability, which is related to mobility, but different levels need to be independent.
Need a strategy to prevent passive escape, add home base, add movement ability changes, add ranged damage to the map?
High risk and high return require clearer, concise appeal and feedback.
After there is terrain, there is still a lack of local creation to test the strategic ability to fight more and less, as well as the maintenance ability to fight less and less, which is used to provide the space and timing control possibility of scheduling resources.

13. Prototype 5 main base emerged

5.1 Design Purpose

Absorbing the advantages and disadvantages summarized in Prototype 4, we further optimized and shifted more focus to emerging how to participate in a reasonable goal challenge feedback structure. A battle is made up of multiple coexisting, nested goals, challenges, and feedback:

1 goal

Reasonable, difficult but reachable goals, goals that can be broken down
Different types of goals at different levels, big goals are always hopeful but difficult to achieve, as the ultimate pursuit of hooks
The medium goal is challenging but can be achieved within the limits of the acceptable delayed feedback, as stage feedback
Small goals serve as filler feedback for quick learning to achieve after easy to achieve or easy to fail

2 challenges

Once you have a goal, you have to set a challenge, which is the core of everything and the hardest part
There are different challenges for different sizes, and a good challenge is one that needs to be difficult but can be completed
The best challenge is to use all your skills to complete the challenge to the limit, and the methods generally include adapting to the challenge difficulty, adapting to the goal, taking failure as part of the challenge, blurring the concept of challenge, and so on

3 Feedback

Unlike goals and challenges, no matter what the size of the goal or the difficulty of the challenge, feedback has a common requirement
First of all, feedback can be used to identify progress, so that players know where they are in this challenge and how close they are to their goal
Human perception is uneven, magnifying the distance when it is far from the end point, or when it is not knowing the distance from the end point, and shrinking the distance when it is close to the end point. Therefore, timely and clear phased feedback can help players better carry out a challenge and get closer to a goal
Secondly, positive feedback gives the player motivation and confirms the player's operation
Finally, negative feedback is also important, teaching the player, punishing the player, and motivating the player in reverse

We've talked about the emerging advantages before, and the same applies to this goal challenge feedback structure, so let's actually try it out:

5.2 Rules of Gameplay

It is still a nine-square grid map, each player has a main base, and the main base fails when it is captured, and the map design is as follows:

The sub-base is the same as the Prototype 4 mercenary unit, except that it was initially fixed to our unit

The attributes of the main base can be 9 attacks and 9 blood, and the sub-base can be 5 attacks and 5 blood, and the attributes can be adjusted appropriately to try, which will determine the rhythm of the game

If the player's unit or main base is destroyed, the game fails

Units controlled by the player can move one square per turn

The flow of the round is that the team with the highest mobility moves first

If the slot where the player controls the unit is located has no enemy or other friendly units, a mercenary unit can be generated, and the spawn rules are the same as those of Prototype 4

After both sides have moved, all buildings can assign points to adjacent slots to generate mercenary units, up to a maximum of 2 non-player units controlled by allies in each slot

After the two sides take turns to operate, the battle link will be carried out, and the rules of the battle link will remain unchanged

In addition, the value of a single unit attribute is not greater than 9

5.3 Prototype 5 Summary

Compared with Prototype 4, Prototype 5 solves some problems, such as passive war avoidance, unreasonable maneuver resources, etc. At the same time, for the optimization of the target challenge, reflected in the results, we will find that the player's strategy freedom will be much higher, and some more complex strategies can be carried out based on the simple nine-square grid, such as creating a local to play more and play less. At the same time, the emergent part that we are most concerned about is still close to the prototype 4, and there is nothing to lose, on the contrary, because the player's decision-making freedom increases, in terms of perception, the emergent part will be more prominent, which is also an interesting phenomenon embodied in this prototype. From this, we can know that for emergent exploitation, at least a certain degree of policy freedom is required, and with policy freedom, it is possible to make more inputs to the entire complex system.

Next, we do the opposite, do subtraction, and implement the same game in a more abstract way.

XIV. Prototype 6 Abstract Phased Battlefield

6.1 Design Purpose

According to the verification results of prototype 5, we can further abstract to feel the emergence:

It is possible to focus on resource allocation, reduce the factors related to movement and battlefield location, and be more abstract
The mobility of the mercenary unit is critical, and if you want to bring mobility and get closer to the wingman than the building, the neutral unit can still be a building
By directly assigning points, the base can exist in another form, and the existence of points also lays the foundation for the next prototype
Join the design of units in the middle of the battle to decentralize decision-making

6.2 Gameplay Rules

The battlefield consists of Battlefield Effects and Battlefield Benefits, as well as neutral defending units

Battlefield effects are designed to assist decision-making and balance strengths and weaknesses, and there are three types:

Test, if the total number of units invested is greater than XX, the temporary bonus is XX
Support, if the total number of units invested is less than XX, the temporary bonus XX
Consumed, strengthens neutral units by X

Battlefield effects are not revealed on the first turn, and are not revealed until after the end of the round

Battlefield gains, where the player pursues objectives and are controllable, are revealed at the beginning

Increases the total free point limit by X per turn, X is random, and determines the number of points for neutral defense
Enhance the player itself, not for the time being

At the beginning of the game, 15 points are allocated to their own units, as well as X free points

When you start the game, you put three battlefields to reveal the battlefield benefits, but not the battlefield effects

Players can double-blind do two operations per turn, distributing free points to one of the three battlefields to become a mercenary unit, which is different from the previous prototype in that the mercenary unit can be mobile, which is basically the same as the player's unit

The player moves their units to or from a battlefield back to the base, and the player unit needs to move back to the base to another battlefield, as shown below:

Battles are resolved at the end of the round, as before

Each battlefield can have a maximum of one mercenary unit in addition to the player unit, and if there is a mercenary unit, the base cannot send new one

Free Points are full at the start of the turn, and if there are no neutral units on each battlefield, a random neutral unit is regenerated

Neutral units spawn 6 dice, three rolls represent output, survival and maneuver, and no spawning if they are cast to survive 0

6.3 Prototype 6 Summary

The experience of prototype 6 is different from the previous two prototypes, although it is more concise, but the playability is relatively higher, this is because after the spatial position game is more abstract, it can instead strengthen the effect of a single area, and the performance of the emergence is greater, which is also one of the inspirations that this prototype gives us:

The existence of agents and environments is a very important part of the emergence, they are both components of complex systems, they complement each other, or they can be directly considered as one.

When playing prototype 6, I also found some parts that could be better, such as the first step to reveal the battlefield in turn can reduce the cost of decision-making, and it also reflects the uncontrollable part to reduce the cost of decision-making. For example, the player will control the unit a little bit dominated by neutral units, and the battlefield income can be a little higher. On top of that, the overall number of decisions is small, and the burden of each step is too high.

15. Prototype 7 subdivides the advantages and disadvantages of decision-making

7.1 Design Purpose

According to the verification results of prototype 6, in addition to optimizing the details, we try to split the decision-making to reduce the information uncertainty and burden of each step of the decision, but each step of the decision is interrelated, testing the depth of the player's thinking, and shifting from a precise strategy to a fuzzy strategy, which is one of the key points of prototype 7.

On the other hand, we once again focus on the balance of advantages and disadvantages, and explore ways to use external resources to bring in and emerge combined:

For a regular game with a single-game battle as the core, the ultimate goal of a single-game battle is to obtain the happiness and reward of victory, and the reward is to serve the development of the outside, and finally to better carry out the single-game battle. At the same time, according to the development, experience, skill and other factors, for a player, there are 3 states when he first enters the game in battle: 1 is obviously stronger than the opponent, 2 is close to parity, and 3 is obviously weaker than the opponent.

Slightly stronger and slightly weaker and evenly matched situations are the most common and the ones we expect the most (through elo matchmaking, etc.)

In this case, the expected player goal is to get a victory, and the reward for the victory is greater than the other paths.

The difference is that in this case, the player can have two biases, the first is to magnify the advantages and narrow the disadvantages and then fight, and the second is to directly fight to divide the winner and loser, the weaker side will favor the former, and the stronger side will tend to the latter.

For players who are significantly stronger than the opponent, we need to give them a path that is more difficult to achieve, has higher rewards, and weakens the player in the process, such as destroying enemy bases, destroying all enemy buildings, attacking third-party units, and so on.

Conversely, for players who are weaker than the enemy, they can help third-party units defend, obtain defense resources, etc. Of course, it is also necessary to have a reasonable winning condition, and it is necessary to consider whether the weak side will take the initiative to take the damage, play negatively, surrender, and end the game quickly.

It is also not possible to use simple combat power for the system to determine the advantages and disadvantages, which will promote the emergence of suppressive combat power behavior, and the combination of complex systems, we need players to judge the advantages and disadvantages of both sides according to the battlefield situation, and make different decisions to determine the advantages and disadvantages.

So how can external resources achieve the goal of balancing advantages and disadvantages?

We divide the player's goals in battle into two categories, victory and resource acquisition, which do not conflict but have a focus;

When players achieve these two goals, they have two choices: first, reduce the resources obtained outside the game to obtain stronger combat power in the game, and second, sacrifice the combat power in the game to obtain more resources outside the game;

Among them, there is a gradual change, so that regardless of the advantages and disadvantages, you can take two paths, risk and return coexist and are controlled by the player himself;

For example, let's say we can get 10 points of loot for a battle win and 2 points for a defeat. There is a complex system in the game (such as a third-party faction), players can interact with it differently, attacking will get more outside loot, but also damage their own combat power, or they can do a second type of interaction, using the consumption or reduction of outside loot to enhance their own combat power.

In this way, we achieve a goal, theoretically the optimal solution for players is to exchange as much combat power as possible for off-game resources under the premise that the combat power has just outperformed the opponent's ability to win, so that players will play a very dynamic game with complex systems to adaptively balance the advantages and disadvantages.

As a result, Prototype 7 will introduce a new resource that simulates a disposable resource other than combat units that are brought into the game from outside the game.

7.2 Gameplay Rules

There are three battlefields, connected in a straight line, with one end close to player A and the other end near player B, as follows

The player's own flagship starts at the main base, and the flagship cancels mobility, and the initial distribution points are 10, but the flagship's HP is multiplied by 10

Players can allocate N points from their free points each turn to become a mercenary unit, which can have 0 points, double-blind, and must spawn at the main base

The next step of the round is a double-blind decision on whether the flagship moves forward or backward or does not move

Next, if there are no enemy or neutral units in the slot of all mercenary units, they will automatically move one block to the enemy's home base

The next battle will settle the damage, and the rules will be the same as the previous prototype

At the end of each turn, neutral units spawn on the battlefield

Neutral units still spawn by rolling 6 dice twice, as output and survivability, and mobility as 0

Each turn generates battlefield gains, and the size of the gains is randomly rolled with 10 dice

Added interest design, adding 1 point at the end of every N-point round, reducing the difficulty of predicting the enemy and increasing the game space

Players are encouraged to be more aggressive by adding a one-time reward for successful capture, such as +1+1 for capturing units, or free distribution of loot to capture units or repossession

The distribution of loot is completed after the next round of monsters and proceeds are generated, and before the mercenaries of both sides

You can cast up to two mercenaries per battlefield

7.3 Prototype 7 Summary

Prototype 7 embodies the dismantling of the strategy, which is more in line with the design of emergence, because emergence, which requires a certain step of decision depth to pull the difference. It's not just about emergence, but when we do design, we tend to blur the precise strategy as a whole, not just for the sake of emergence, but more in line with human decision-making instinct.

Especially in real-time games, precise strategies are generally provided as an additional depth cap to test the player's ability to master the game, collect information, react, remember, and make quick decisions. Operationally speaking, the bias is towards high-frequency operation to test the response, and the benefit cost is smaller, and the low-frequency operation test strategy has a larger benefit cost. The speed of operation, the accuracy of operation, and the depth of operation strategy are impossible triangles. So for us, we prefer fuzzy strategies over exact strategies, and the focus is on using emergences to give players the illusion that fuzzy strategies are exact strategies, and then using mechanics to amplify the feedback of fuzzy and precise strategies.

At the same time, the more ambiguous strategy can only be established with the advantages and disadvantages of dynamic equilibrium mentioned above. (With a completely accurate strategy, the optimal solution of the superior player is very clear, and on the premise of being better than the enemy, all combat power will be exchanged for an out-of-game resource, and the inferior player will be able to do nothing)

16. Summary of emergent design

Thank you for your patience to see the friends here, we have designed seven prototypes above, step by step to show some key points of complex systems and emergent design, the prototypes are quickly designed by myself, relatively simple, if you have a better idea in the process of playing, you can modify the prototype at any time, and achieve the design purpose. With the above prototyping, I would like to summarize some of the experiences of emergent design:

1. The rules should be as low-level, abstract and universal as possible

For complex systems, the complexity comes from the interaction between agents, not the complexity of the rules themselves. Conversely, for design, the simpler the rules are used to achieve the design purpose, the more elegant it is, and the easier it is to debug, anticipate, and exploit emergent phenomena. Multiple iterations of input and output, each iteration is as simple and easy to understand as possible, and the rules are clear. For example, if we design a complex system of elemental reactions, we should not design the rules that wood burns when it encounters fire, and fire extinguishes when it encounters water, but we should design more basic general rules, such as a temperature system, all of which change the temperature of the surroundings, burning when the temperature exceeds the ignition point, and freezing below the freezing point. In this way, all the elemental reactions that can be described by a single metric involving temperature change are unified, which is not only concise, but also highly scalable. This is an example, the lower the rules, the more abstract, the more concise, and the more general they are, the more suitable a design for complex systems.

2. The balance between game space, decision-making complexity, learning cost, and player control

The information should not be too complete, otherwise the threshold for decision-making is high, and it cannot be too missing, otherwise the strategic significance is too small. Generally, the control of information density is achieved through gradual release. At the same time, in addition to the impact of information on decision-making, it often also affects the player's controllability of the battle situation, which is often closely related to emergence, and it is especially necessary to pay attention to a phenomenon: a certain information is related to high returns, forcing players to have to go deep into the emergence, bringing huge decision-making costs. In the same way, information, including information about the rules within a complex system, does not need to be taught in a hurry, and in a game with a complex system, there is no need to pursue too much control of the player, exploring the complex system itself can also be used as one of the fun, and the familiarity with the complex system can also be used as a sliding difference between different players.

3. Splitting reduces the cost of decision-making at each step, provides timely feedback, and reduces long exams

As mentioned in Prototype 7, the strategy is more suitable for the design of emergence, because the emergence-in-itself requires a certain step of decision-making depth to pull the difference. What we pursue is that players try to use fuzzy strategies to deal with complex systems and emergence, but through design methods such as emergent results, UI hints, etc., players feel like they are doing precise strategies and get high feedback.

4. Watch out for the abuse of randomness in emergence

Most of the time, the uncontrollable part is to reduce the cost of the strategy, because it does not need to be taken into account in the strategy, and controlling randomness, fighting randomness, and using randomness can also bring considerable fun. Emergence is similar, they are all uncontrollable, then some friends may think, what will happen if the emergence is added randomly?

Let's take a simple example, the Sims have their own ideas, but at this point we let the rules of some ideas appear randomly. At this time, what will some players think about the emergence of these ideas? The conclusion is that they can't see it, and if they abuse randomness, randomness will erode the meaning of emergence, and finally, it will become random and overshadow the fun of emergence, or the two will affect each other.

Let's go a little deeper, if we divide the combination of emergent and random into three categories:

We will find that the pre-emergence randomness has little effect on the emergence, but it also does not have much significance
Random after emergence, which destroys the meaning of emergence
The doping is random in the process of emergence, which greatly increases the degree of uncontrollability, and actually destroys the emergence

Therefore, in the same game, it is best to separate the random part and the emergent part, and the random result can be related to the emergence, but do not mix the two, or the emergent result is randomized together

5. Balance the strengths and weaknesses

It is better to balance the advantages and disadvantages of the system with a self-consistent rule, and the best is to let the superior party adaptively pursue high risk and high return to balance the advantages and disadvantages. For example, the way in which outside resources are brought in.

6. Use early cognition to reduce the cost of comprehension learning

Complex systems generally have a relatively high cost of understanding and learning, and there are many existing complex systems in real life, so it is necessary to use the advance cognition of real life to lower the threshold. For example, if you have designed some rules originally, you can also find the closest complex system in life and wrap it up in the past.

7. Similarities and differences between agents and AI

Proxy is not equal to AI, which is a common misunderstanding, because many times the two of them have overlapping parts, but from the design point of view, it must be noted that the agent is a component that executes input and output with certain rules, and it does not need to pursue intelligence, complexity, flexibility, etc., if it is designed in this way, it will be deviated. It can be seen that the above prototypes will design the agent very simply, the focus is not on how complex and intelligent the agent is, but on whether the direct interaction rules between the agent and the agent can produce the emergence we want as a whole. A good agent should enjoy the same underlying logic as the player's control unit, but take advantage of the agent's unique advantages. Similar to the design of AI, most realistic AI is not the best experience AI, and the best is to use the fact that AI can cheat to maximize the role of AI in player experience. For example, the director system similar to the path of survival, the aura enemy AI, and so on.

Agent is not equal to AI, but you can refer to the strategy that AI and players can use to design, and here is my incomplete summary of the strategy game of 1v1 confrontation:

The simplest rock-paper-scissors, discrete loop restraint, the most basic strategy
Colonel game, resource allocation strategy, defeat the strong with the weak, and game resource allocation ability on the basis of total resources
Continuous cycle restraint, for example, a resource is a little higher advantage, too much is a disadvantage, most of them cooperate with the colonel's game or resource planning
In the bidding game, both sides spend resources to bid for a goal, a continuous psychological game
Raise games, typically heads-up poker, are also continuous psychological games
Deep game, competition for thinking step length, most chess
Racing game, the speed of the two sides to achieve the goal, itself is difficult to establish, generally supplemented by the design of the interaction between the two sides, mahjong is typical, or combined with other strategy design, such as combined with resource planning

Most of the actual strategies are compounded by the above multiple designs, and these strategies can be applied to the design of the agent.

8. Go beyond complex systems

Finally, this article explores complex systems and how they can be used in games, and we never seek to simulate a complex system, or how much to pursue an emergence that is consistent with reality. We pursue the use of complex systems and emergences to serve the purpose of our design.

So we have to learn about complex systems, but to go beyond complex systems, we can reconstruct real complex systems, such as the design physics implemented in the Wandering Ark, which is different from real physics. At the same time, we can also have external forces to give the system, do not need to pursue complete internal system to achieve self-adaptation, but also can actively amplify the emergence, etc., of course, when the complex system and emergence do not help the core gameplay, do not hesitate, cut off, everything for the purpose of design.

XVII. Conclusion: Emergence, Life, Wisdom

The above 8 chapters are all from the perspective of a designer, more rationally to analyze complex systems and emergence, and the purpose is very strong to explore the way to use it in the game, I hope it will be helpful to everyone. In the last chapter, casually talk about some less rational feelings, and casually communicate and laugh.

Fascinated by complex systems and emerging phenomena is inseparable from awe and curiosity about one's own life and wisdom. Many friends don't know that there is no accurate definition of life until now, what is life? Similar to how to define complex systems, this is a topic that has not been fully agreed upon, scientists can only put forward some more general characteristics of life, such as:

Structure, the various chemical components of an organism are not randomly stacked together in the body, but a tightly ordered structure. Only when macromolecules form a certain structure, or form an orderly system such as cells, can life phenomena be manifested.
Metabolism, life and the surrounding environment will constantly exchange matter and energy flow. After some substances are absorbed by living organisms, a series of changes occur in them, and they are excreted as the final product, which is called metabolism.
Stress, in which organisms respond to external stimuli. It consists of two processes: feeling the stimulus and reacting to it.
Self-replicating.
Evolution and adaptation, life can adapt to different environments through heredity and evolution.

The above are just some of the characteristics, and there are many definitions of life, and there is no precise definition so far. From the perspective of complex systems, life is nothing more than a whole phenomenon that emerges from the aggregation of simple components. As the American physicist Murray Gell-Mann once put it, the complex behaviors we see in the world around us – even in the living world – are only "complex appearances emerging from esoteric simplicity."

One might also ask whether emergences are just an illusion, meanings that we subjectively impose on the appearance of the system? This question, in fact, is not fundamentally different from whether wisdom exists, whether it is an illusion of the brain, everyone may have a different answer, but as a carrier of wisdom, it is difficult for me to admit that we can build our brilliant civilization only by illusion and leave behind famous poems that have been passed down through the ages, but perhaps in the eyes of higher intelligent beings, the spark of our civilization is just the stupid cluster wisdom shown by ants huddling together, and it cannot be defined as a cluster effect, and it cannot be worn the crown of wisdom.

For the game, the last part of the game, which was not explored in depth in the previous chapters, is the role of complex systems to partially replace other players, which comes from the inspiration for the connection between the three emerging life intelligences. Part of this comes from our latest thinking on long-term games – what direction is best for a small team with a focus on creative design. This is a very big and heavy topic, I will not expand it here, and I will have the opportunity to write a separate article in the future, and I will directly talk about the three feasible directions that I think:

Multiplayer competition, relying on the interactive game between multiple people, constantly produces new content.
After taking advantage of the emerging part of the function of replacing players, the 1V1 competition makes up for the part of the 1V1 competitive game that is not as good as the content of the multiplayer.
Use emergences or other methods, high-replayability and high-design space of single-game PVE, and efficiently update levels to do long-term.

The second point is the direction we have been exploring, how to use the emergence of simulated players, so that the system can replace the experience that some players give to each other. As I said before, this is not to make a realistic AI to simulate human intelligence, but to use the emergence system to simulate the kind of feeling that human players bring to other players, with chaos in reason, uncontrollable and uncontrollable, impulsive madness and rational decision-making intertwined.

Well, writing and thinking about the role of game design, occupational disease attack, this article took several months before and after, it may be a little verbose and incoherent, thank you for your patience to see the friends here, welcome to communicate more, and welcome to pay attention to my public account "Fish Pond Game Production Workshop".

Let's end this article with a piece of gibberish that I deleted from the previous article!

"Life and wisdom are not accidental, but inevitable.

The second law of thermodynamics, the law of entropy increase, states that in a closed system, the overall chaos increases over time until everything is at its most disordered. So Victorian thermomechanics imagined that the universe is a closed system, and all energy will be slowly converted into waste heat, stars will go out, black holes will exhaust themselves, and eventually everything will be in a bland and consistent silence, which is called "heat death". However, in this irreversible process, the universe seems to be a little unwilling, and gradually calms down the cosmic ocean, making waves in some corners, and some orderly constructs emerge from disordered matter. These constructs tumble in the primordial amniotic fluid of the planets, absorbing the light of the stars with all their might, using simple carbon as skeletons and other things to build complex adaptive systems, some of which can even produce beautiful flowers. Although, the creation of these tiny orders requires energy from outside the system, at the cost of more chaos in other parts of the universe.

However, in the process of the universe moving towards disorder and chaos, it still quietly gave birth to orderly life in some corners, and they are so precious, they are the cosmic whispers against disorder, and they are the laurels that the universe crowns itself. And the wisdom that emerges from life is more like the whispers of the universe just sung into a ballad, a brilliant jewel in the crown of laurels, a miracle among miracles. Everyone's soul is unique, a flower that struggles to grow out of chaos, and everyone's soul is a part of the soul of the universe.

So, live hard, think hard, and live up to this miracle. ”

Thank you for reading.

【30,000 Words Long Article】The game design is left-handed, complex systems and emergence

Read on

How to choose 520 romantic gifts, a must-have ROG white gaming mouse with high value!

I'm a PC! 4th, 4th, 4th, 4th

Netflix Korean dramas are really daring to shoot, and the setting is more cruel than "Squid Game", and the excitement is slightly inferior

Avoidance and lies: Rohingya incognito, identity play in Myanmar's jade market

Tencent's copied shooting game was launched on overseas PC, and it was praised to the sky without being scolded, foreigner: good copy

"Top Internet Celebrity" Guo Youcai: The signing fee reaches 50 million, and the game of Internet celebrities is to make money nakedly

OPPO Reno12全系搭载星速旗舰芯,游戏体验更畅快

The Glory of Kings was suddenly severely criticized by the state, and the alarm bell of the game industry has sounded, and the answer is revealed in the article

The Apple iPhone 16 Pro real machine is suspected to be exposed, and the size and design have been significantly changed

The Faker Hall of Fame skin was announced, and it was complained that it was over-designed and looked like Ultraman

Song Yi's outfit, did the designer forget to consider going to the toilet? Is there a mystery hidden in the bottom design?

HUAWEI Enjoy 70S is coming with a large eye-friendly screen, once again breaking through the aesthetic design

Just know to give him love! Inventory of 20 R-star boutique games! There's always something for you

Hongmeng + Pura70 with the same processor, Huawei's new machine will be launched soon, and the configuration and design will be exposed

The new Toyota Crown entered the store, with a wheelbase of 3 meters, and the design became business, and 299,000 concessions

Honor 200/Pro Picture Reward: Art Nouveau Mirror Design, Elegant Light and Shadow Photo Master

Netflix restricted Korean drama! The scale explosion is comparable to "Squid Game", and there are all domestic dramas that dare not be filmed!

Can you earn 500,000 yuan in two years of playing games? Uncover the hidden corner of professional leveling and playing

U.S. stocks closed: the Dow fell back to a new high for the Nasdaq, Nvidia rose more than 7%, and the game station once soared 40%

Can you earn 500,000 yuan in two years of playing games? Every time the reporter reveals the "hidden corner" of professional training and playing with him

The child starts skipping classes and indulging in games! When psychological problems first appear, it is the best time to intervene effectively

The fourth-generation "BMW X3" real car was exposed, using a new design style and adding a plug-in hybrid!

The NVIDIA RTX 5090 graphics card still has a dual-fan design: the volume is almost the same

The woman's request for a refund for recharging millions of games was denied, and the game company had been cancelled when the invoice was requested

Plaza del Torico. At the center of the old town of Teruel is a square called Torrico, in the center of which there is a fountain with columns, columns

Eastern Seawall Reconstruction Project (Phase III)· Yangmei pit section / Hanbo design

Farmers' Market Renovation Design - Exploring the Innovative Road of "Multi-functional Integration".

How to combine intelligence and ecology in the planning and design of farmers' markets?

Hongmeng 4.2 + Kirin 5G, Huawei nova13 series configuration and design exposure, as soon as this month to launch

Strong signal, fast internet speed, stable game! Rui Jai Snow Leopard BE50 is a Wi-Fi 7 router that is more worth buying

Blue and pink are matched to create a different dream. Go and see the exhibition of design artworks #Zhiyoutide play ##潮玩节玩开心#

Fast update and iteration of software and hardware to meet consumer needs On May 31, 2024, the new M7Ultra was newly launched, becoming a luxury SUV with five seats and six seats