Introduction

To those who have followed me, it is no secret that I love Zung Jung Mahjong. I have famously, or perhaps infamously, at times critiqued other Mahjong variants such as Riichi. What sometimes gets lost in these critiques is that I am not saying that these variants are unfun, bad games, or that people should not play them. Rather, I am saying that the design logic behind them often feels lacking to me, especially when compared to Zung Jung.

Some people do not care much about game design and only care whether a game is fun to play. This is a perfectly valid perspective. I, however, care deeply about system design and, by extension, the game design of the games I play. For me, the quality of a game’s design and how enjoyable I personally find it to play are separate questions.

For example, from a systems-design perspective, many of Riichi Mahjong’s mechanics feel structurally inelegant to me, even though I still find the game enjoyable. Conversely, Chess is generally regarded as an extremely well-designed game, and I would agree with that assessment. Yet personally, I do not find Chess particularly compelling to play beyond a game or two; I’d much rather find myself at a Riichi Mahjong table than looking at someone else from across a Chess board.

You may be asking yourself at this point, what do I mean when I say something is structurally elegant or inelegant? That is the point of this article. This article will discuss the design features of Zung Jung Mahjong that I find structurally compelling and that are a major reason I love the system so much. I will compare it to Riichi Mahjong, Mahjong Competition Rules, and Hong Kong Mahjong to contrast how other systems address similar issues in the game and to highlight why I think the approach Alan Kwan pursued with Zung Jung is more elegant from a game design perspective. This is not primarily a strategy document, for the strategy of Zung Jung, several Riichi strategy concepts actually transfer over as well as strategic concepts from variants like Hong Kong or MCR. Additionally, you can purchase my zine, Mahjong for Everyone, for a bit of Zung Jung strategy if you are interested (at the time of writing, May 19 2026, it is on pre-order, and I expect to begin shipping them out next week).

Zung Jung’s elegance, for me, comes from five major areas: non-redundant modular design, iterative design, symmetrical design, mathematical balance, and lack of luck-based scoring elements. When I say Zung Jung is more elegant than another Mahjong variant, I am fundamentally saying that it excels more in these five areas. Others may have different criteria that they consider when evaluating the elegance of a variant and that is fine, but this is the criteria I am using.

It may be helpful for some to follow along with this article with the Zung Jung, Riichi (Japanese version), MCR, and Hong Kong cheat sheets from the Google Drive in Game Catalog.

Before I get into the main five areas where I find Zung Jung particularly elegant, I’d like to give a special mention to how easy Zung Jung is to learn and teach to a new player, and to how its pattern list was laid out by its designer, Alan Kwan. First, Zung Jung is very easy to learn. On its surface is appears to be quite a simple game and I think that can betray how deep it is. Over time, I’ve come to see simplicity as a form of elegance and a game that remains simple and easy for new players to learn while also having deep strategic choices and a high skill ceiling gains a lot of points on elegance from me. I think simple but deep is the highest form of elegance.

Regarding accessibility for newer players, I also greatly appreciate how Alan Kwan designed Zung Jung’s pattern list by category and series. This is an elegant way of designing a scoring pattern list that makes it easy for new players to understand. Instead of having to remember a bunch of individual patterns, they can memorize patterns by families and characteristics. This significantly reduces the cognitive load for new players learning the game and its something I think other Mahjong variants could learn from. What makes Kwan’s pattern layout even more elegant from my perspective, however, is that the organizational presentation also tells you which patterns may and may not be scored together, as two patterns organized into the same series may not be scored together, but, if they are not in the same series you can score them together (with the exception of 10.0 Irregular Pattern which I think Alan Kwan dealt with in a rather inelegant manner).

Non-redundant modular design

Non-redundant modular design occupies two axes. First, by modular design I mean a design philosophy toward the scoring patterns where each scoring pattern evaluates the winning hand on a single metric rather than multiple ones. Second, by non-redundancy, I mean that you do not have multiple scoring patterns that evaluate the same criteria for winning hands. In this section I will first look at modular design then look at non-redundancy.

Modular design

Modular design is something that I have come to strongly appreciate about Zung Jung’s game design. It seems very trivial but in my opinion, it really improves the quality of the game and the speed at which someone can learn the same since each pattern only has one condition that a player needs to remember. To illustrate a modular design I will look at two analogous scoring patterns in Riichi and Zung Jung: Iipeikou and Two Identical Sequences.

Structurally Iipeikou and Two Identical Sequences are the same pattern, they both evaluate whether or not you have two sequences of equal rank and suit in your hand.

However, Two Identical Sequences exhibits modular design, whereas Iipeikou does not, since Iipeikou evaluates not only if your hand contains two equally suited and ranked sequences, but also whether or not your hand is closed. If you hand is open, your hand does not qualify for Iipeikou even if the structural requirement is met. Zung Jung has two scoring patterns that accomplish what Riichi compresses into a single pattern: Two Identical Sequences and Concealed Hand. Two Identical Sequences evaluates whether or not you have two identically suited and ranked sequences, and Concealed Hand evaluates whether or not your hand is closed. For me, this is a more elegant way of approaching scoring. Hand state and hand structure are separate, and it gives a feeling of less arbitrariness.

Zung Jung’s approach also addresses another design feature of Riichi that I think is inelegant: unequal bonuses for the concealed-hand state. Yaku such as San Shoku Doujun, Chanta, Honitsu, etc., receive extra points if their hand is concealed. Pinfu, Iipeikou, and Ryanpeikou must be concealed. Tanyao, San Shoku Doukou, Toi-Toi, and Shou Sangen receive no such bonus (unless, of course, you qualify for San Ankou, or Suu Ankou, which is a completely separate Yaku and the former of which doesn’t require the entire and to be concealed and latter of which gives you a Yakuman). Zung Jung just assigns every hand Concealed Hand if it meets the requirement. This is more structurally elegant to me, as it is consistent, whereas Riichi deals with hand state inconsistently (I also know that Riichi has Concealed Hand in it, but this works at the level of Fu, not Yaku).

Non-redunancy

In the interest of fairness, I am going to pick on MCR for this section. I can be a bit hard on Riichi at times and besides, MCR exhibits the property I want to talk about more strongly.

MCR was the variant I learned first. I taught myself how to play it. Many of its design principles may be why I love Zung Jung so much; namely, MCR has a very modular structure. However, it also has a highly redundant structure (with some non-modular features as well). There are two cases I would like to look at here: First, how MCR deals with Concealed Hands because I think it identifies the issue clearly. Second, how MCR deals with Concealed Pung and Kong, since Zung Jung’s implementation is more elegant.

MCR has 81 scoring patterns in its system. Of these you have Concealed Hand (which we have discussed), Self-Drawn (which evaluates whether or not you drew the winning tile and Riichi players will know as Tsumo), and Fully Concealed Hand (which evaluates both whether or not your hand is closed and you drew the winning tile, Riichi players know it as Menzen Tsumo). Concealed Hand in MCR scores you 2 points and Self-Drawn scores you 1. Fully Concealed Hand scores you 4. First, If you have Concealed Hand and Self-Drawn, why is Fully Concealed Hand needed? And why does it score an extra point? This gives me the feeling of arbitrariness. You don’t need a non-modular pattern like Fully Concealed Hand if Concealed Hand and Self-Drawn can stack. Again, where is the extra point coming from, and why? This creates confusion when scoring as players need to consciously remember which patterns they qualify for and which ones they cannot combine with how they are scoring their hand which increases cognitive load and likelihood of scoring errors.

Zung Jung does not give you extra points for self-drawn wins so I’d like to look at a different set of patterns from MCR to show Zung Jung’s implementation of the same idea. In MCR, you have an entire series that evaluates the state of Pungs/Kongs in your hand: One Melded Kong, One Concealed Kong, Two Concealed Pungs, Two Concealed Kongs, Three Concealed Pungs (which applies to Kongs this time as well), Three Kongs, Four Kongs. This is an inconsistent and inelegant expression of this idea. Zung Jung takes an approach of considering the state of concealed Pungs/Kongs in your hand: Two Concealed Triplets, Three Concealed Triplets, Four Concealed Triplets (these apply to both Pung and Kong) and then a different series of patterns that evaluates whether you have Kongs in your hand: One Kong, Two Kong, Three Kong, Four Kong. This is structurally cleaner and more elegant than MCR’s implementation of this idea.

Iterative design

Another aspect I find very beautiful in Zung Jung’s game design is the iterative, escalatory nature of the patterns in its system. The idea that if a structural relationship is considered meaningful at one magnitude, then the system should consistently represent adjacent magnitudes where logically appropriate. Alan Kwan directly talks about this design aspect in his design notes on the pattern selection criteria he applied for Zung Jung:

Each category is sorted into one or more series. For the sake of logical consistency, once lower patterns in the series are adopted, the higher patterns must always be adopted (such as “Three/Four Identical Sequences”). Conversely, the lower patterns in the same series are considered, and “Big/Small Three Winds”, “Small Three Similar Triplets” etc. are adopted. Two-set patterns which are too easy and would increase the compexity of the exclusion rules are rejected, such as “Two Similar Sequences”, “Two Similar Triplets”, “Old and Young” etc. Among this type of two-set patterns, only “Two Identical Sequences” (which demonstrate the strongest consistency) is adopted. “Original Call” would increase rules complexity, and is thus rejected.

This is something that other variants do very imperfectly. One of the things that has drawn my ire from Riichi Mahjong is the presence of Iipeikou (Two Identical Sequences) and Ryanpeikou (Pair of Two Identical Sequences), but the absence of Three and Four Identical Sequences in mainstream Riichi Mahjong. Conversely, MCR also has Two Identical Sequences (Pure Double Chow), Three Identical Sequences (Pure Triple Chow), and Four Identical Sequences (Quadruple Chow), but does not have a Pair of Two Identical Sequences (Ryanpeikou). This is less egregious, but it is worth noting.

This partial-escalatory aspect we see in Riichi and MCR is inelegant design. It’s unintuitive. Intuitively, if I score points for achieving a pattern, doesn’t it make logical sense that I be rewarded even more strongly for increasing the magnitude of this structure? The response from Riichi and MCR is “sometimes,” the response from Zung Jung is an unequivocal “yes.”

Symmetrical game design

I am going to pick on Hong Kong Mahjong for this section in the interest of fairness, but what I am saying here also applies to Riichi, MCR, and many other variants. It is worth noting that Hong Kong is a highly decentralized Mahjong variant and as such how it is played can vary widely across different tables and communities. I am using Hong Kong Mahjong as presented in the cheat sheet I designed, which I believe includes the canonical patterns most Hong Kong Mahjong players recognize.

Symmetry, in the context I am using it here, refers to the idea that if a meta-pattern is present in one domain, it is mirrored and universalized rather than being merely a quirk of a specific pattern. That probably sounds very technical and hard to understand, so let me illustrate it with these two patterns: Small Three Dragons and Small Four Winds.

The structure underlying these patterns is that all honor tiles within a given class, either dragons or winds, are represented exactly once through a combination of triplets and a pair. Small Three Dragons consists of two dragon triplets and a pair of the remaining dragon, while Small Four Winds consists of three wind triplets and a pair of the remaining wind. Structurally, these patterns express the same underlying relationship across two different honor-tile domains but at different magnitudes. If we consider iterative design we can universalize this structure and produce two patterns which are exclusive to Zung Jung, as far as I know: Small Three Winds (Two Wind Triplets and a pair of a third Wind), and Small Three Similar Tripets (Two triplets of equally ranked number tiles and a pair of the third)

This symmetrical design also encourages intuitive pattern building: if this meta-pattern is valid for Small Three Dragons, I can use the same concept everywhere, its portable. This greatly helps you understand which patterns you can build and why. Zung Jung takes this aspect of design far more seriously than most Mahjong variants. When you consider the iterative design principles of the variant something really beautiful happens that you can see clearly with these Honor Tile patterns and is the reason I chose them to illustrate this design principle: Small Three Dragons, Big Three Dragons, Small Three Winds, Big Three Winds, Small Four Winds, Big Four Winds, Small Three Similar Sequences, Three Similar Sequences. Of these patterns, only four are in Hong Kong Mahjong, five are in Riichi, and six are in MCR, but Zung Jung fleshes out the entire logic of the meta-pattern through symmetry and iteration to include all eight.

Mathematical Balance

I have a background in mathematics, this aspect of Zung Jung’s design is particularly compelling to me. One of the things I most appreciate about the system is the degree to which its scoring structure attempts to maintain mathematical proportionality and internal balance.

I will not go into the nitty gritty details of the probability of getting various Mahjong hands (nor do I want to do these calculations myself at this time). Others have done various calculations that you can read about but important here is Alan Kwan since he is the designer of Zung Jung. Alan Kwan is a statistician and has several articles, which are now only accessible via the Wayback Machine, in which he discusses his calculations of the probabilities of different Mahjong hands. I will link them below for any curious readers:

Thirteen Orphans vs. Nine Gates

Two Identical Sequences vs. Two Similar Sequences

Seven Terminal and Honor Pairs vs. Thirteen Orphans

Three Similar Triplets (San Shoku Doukou) vs Half Flush

This mathematical approach to assigning values to scoring patterns I find compelling and its an aspect of Zung Jung that I greatly appreciate. Other variants attempt to assign values based on the hand’s rarity, but do so inconsistently. For example, Itsu (Nine-Tile Straight) and San Shoku Doujun (Three Similar Sequences) are both significantly more difficult to complete than Yakuhai (Value Honor) or Tanyao (All Simples), but Riichi assigns them the same value. In some cases, Tanyao can actually net you more points than San Shoku Doujun because of Fu interactions, despite the fact that Tanyao is a lot easier to construct!

We can see a similar situation in MCR, where Seven Pairs is worth 24 points, whereas All Pungs is worth 6 points. Seven Pairs is not that much more significantly harder than All Pungs to complete, to put it in perspective, MCR also scores Three Similar Sequences, Three Consecutive Triplets, and Half Flush all as 24 points, all of which are more difficult to construct than Seven Pairs. Seven Pairs is a relatively common hand; these other hands are not.

Lastly, most scoring systems will give you the same payout for Nine Gates and Thirteen Orphans. However, Nine Gates is significantly more difficult to construct, and you can see this reflected in Zung Jung’s system, where Nine Gates is worth 480 points and Thirteen Orphans is worth a third of that at 160.

This mathematical approach makes scoring in Zung Jung feel earned and directly reflective of player skill. You score what you are able to construct, and what you are able to construct determines what you are able to score. The value assigned to those constructions is then grounded in their statistical rarity and difficulty. That, to me, is elegant.

Lack of luck-based scoring elements

This section is the most subjective part of this article. I know that many Mahjong players enjoy luck-based scoring elements like Flower or Dora tiles in their game, but for myself, I’d prefer not to have them. These scoring elements are often easy to achieve, distorting the signal-to-noise ratio of skill. One example that particularly draws my ire is how Riichi Mahjong is presented as a skill-based game, and in many ways it is, but it is possible to get a Counted Yakuman purely based on luck alone just through drawing Dora and Akadora tiles and getting lucky with the Kandora/Uradora. This is exciting for many yes, but frustrating when someone can take the game purely on a fluke.

That being said, Mahjong is and will always be a stochastic game, luck will always play some role in it. However, I prefer a game design that does not magnify this aspect of it as I think fast convergence on skill expression is elegant game design. That being said, a little luck still makes the game a little exciting and Alan Kwan does acknowledge this with the inclusion of Category 9, Incidental Bonuses. The difference between Zung Jung’s implementation of luck-based scoring mechanics and other variants is that they are rare: Final Draw, Final Discard, Win on Kong, Robbing a Kong, Blessing of Heaven, and Blessing of Earth. With the exception of the Blessings, they also do not score very much, just 10 points. So this is like icing on the cake, rather than the cake itself, which I think is a happy compromise.

Conclusion

I think Zung Jung Mahjong is the most elegant Mahjong variant from a game design perspective. This is because it performs best when evaluated on non-redundant modular design, iterative design, symmetrical design, mathematical balance, and the absence of luck-based scoring elements. Not only that but Zung Jung is also easy to learn and easy to play. It can be a fun social and family game that you don’t have to think very much about, or a deep and serious strategic competitive game that can, and has, hosted international competitions with the best Mahjong players in the world. Its design features and flexibility appeal to me, and maybe after reading this, they appeal to you too.

You may have a different opinion about Mahjong than I, and that’s okay. Ultimately, what matters is if you are having fun while playing. Zung Jung is a beautifully designed system, and I wanted others to see what I see in it, so I wrote this article. Thank you for reading!