Page 2 of 5

Re:History of Mahjong: Cheating

Posted: Sat Jan 24, 2009 1:14 pm
by pringle
If you also want to take out cheat, why keep ippatsu-tsumo?

Get ippatsu tsumo too often is even more like cheating than non ippatsu (do I need to explain how it can be done?).

And if Menzen is luck, ippatsu-tsumo not even higher luck?

Since you say high yaku menzen happen too often, could you provide its statistic? so we can have a look.


By the way, it seem we mixing 2 topic (luck and cheat) into one isn,t it?

Re:History of Mahjong: Cheating

Posted: Sat Jan 24, 2009 2:14 pm
by HotelFSR
I think everyone agrees that Ippatsu Menzen is the most lucky.

I assume that\'s the main reason Ippatsu is not used in competition rules. If, for instance, Ippatsu were instead only allowed with Ron, then I don\'t think it\'s as bad as Menzen Tsumo. However, that is not the case.


Pringle:

You\'re right, this has become two topics in one!

As for the percentage of Menzen Tsumo, it tends to be around 20-30% of all hands which is an extremely high reward for something that is pure luck!

Bear in mind that this means it\'s just as common as Pinfu, which tends to have a very similar percentage and is also a hand that must be closed. Why should a luck bonus be just as common as a valid hand pattern bonus like this?

So, as you can see, that is pretty out of control.

Re:History of Mahjong: Cheating

Posted: Sat Jan 24, 2009 11:19 pm
by sir_seagill
well, reaching is luck, why not take that out too.. -_-
(so are the hands you\'re dealt and the tiles you draw..)

Re:History of Mahjong: Cheating

Posted: Sat Jan 24, 2009 11:34 pm
by HotelFSR
Obviously!

I\'m assuming that\'s sarcasm...

Mahjong is, and always will be- indeed should be- a game that combines luck and skill.

The difference we are talking about here is the distinction between pure luck, so to speak, and what you might call partial luck.

Many probabilites can be influenced by the strategic choices you make as you play the game. Others are completely beyond your control and have nothing to do with your decisions.

You can choose how to play the tiles you are dealt, but Menzen Tsumo is completely random, unlike patterns you can aim for and choices you make based on the actions of other players.

Reach is also a gameplay decision that you make which has its own effects.

That\'s the difference.

Re:History of Mahjong: Cheating

Posted: Sun Jan 25, 2009 2:44 am
by Benjamin
According to the statistic book I have, menzen tsumo occurs in winning hands 18.8% of the time. That seems pretty reasonable when you consider that most hands are menzen and around a quarter are going to be won by self-draw.

You\'re right that you get a bonus for self-draw, but I think you\'re forgetting two things. One is that the bonus isn\'t actually that big. The most tsumo is going to help you by is 1.5x, not 2x. (Unless it bumps you over on fu).

The other thing is that not all self-drawn hands qualify for menzentsumo, it has to be closed. There are quite a few reasons why it\'s good to have this element. Someone else brought up that it makes it possible to win with any hidden hand and I think that\'s important. I also think its important to be able to wait to improve your hand into a better wait or better yaku without risking making yourself furiten. If the scoring system essentially forced players to reach *any* hand would be kind of ridiculous. You\'d see much less defense, for one thing.

Re:History of Mahjong: Cheating

Posted: Sun Jan 25, 2009 2:50 am
by pringle
To think that reward for Menzen is too high or not, I would need answers for following questions first.

1) What is the percentage of the hands going for Menzen (success or not) in Pro games and in Amateur games?
2) What is average number of tiles they been waiting?
3) What is percentage of hands winning by tsumo?
4) What is percentage of hands winning by concealed ron and not concealed ron?
5) What is average yaku winning by Menzen and non Menzen (only yaku not han)?

I still don\'t think Menzen is a pure luck. It can be aimed for by choices we make based on the actions of other players. And usually it was forced too because no one deal into our hand.

It requires more skill than getting a pair of your seat wind frrom the start. You may cast it away because it too easy but not because of luck involve.

Also, I think 20-30% is not high.

Re:History of Mahjong: Cheating

Posted: Sun Jan 25, 2009 2:52 am
by Benjamin
Forgot to mention this--

Tournament rules DIFFER GREATLY! :P
Pro-renmei (JMPL) has a set of \"A-rules\" that ban ippatsu/ura but not all groups use these rules. Also, large tournaments tend to not use these strict rules, some even use red-5s.

Re:History of Mahjong: Cheating

Posted: Sun Jan 25, 2009 11:49 am
by HotelFSR
>> According to the statistic book I have, menzen tsumo occurs in winning hands 18.8% of the time. That seems pretty reasonable when you consider that most hands are menzen and around a quarter are going to be won by self-draw.

Looking at current pro stats online it tends to be around 20-30%. Compared to even the most common hands that is huge. I think that\'s precisely why it\'s not reasonable- it means that you are getting this excessive added randomness very often. It creates unncessary variance in results for competition- all it means (and statistically this is can\'t be denied) is that it\'s more likely than normal for good players to lose against less good players (than it would be with Menzen taken out). This is particularly bad in tournaments, because not all that many hands are played to iron out the nubers. The extra random variance from Menzen is thus not balanced out in tournament play. This is not a good thing, surely?


>> You\'re right that you get a bonus for self-draw, but I think you\'re forgetting two things. One is that the bonus isn\'t actually that big. The most tsumo is going to help you by is 1.5x, not 2x. (Unless it bumps you over on fu).

The fu thing is totally fine. It\'s the yaku for Menzen Tsumo that is the problem, particularly when it pushes you over a limit, like 3->4 or 5-> yaku. That is a massive reward for a bonus that it both pure luck and very common.


>>The other thing is that not all self-drawn hands qualify for menzentsumo, it has to be closed. There are quite a few reasons why it\'s good to have this element. Someone else brought up that it makes it possible to win with any hidden hand and I think that\'s important.

I\'m well aware than Menzen is only for closed hands, as mentioned in this thread. You can still win with a hidden hand, it just needs to have one yaku. It\'s easy enough to win with a low scoring hand as it is, and I the factors mentioned above outweigh the small defensive benefit. If Menzen could only work when you have no other yaku, then fine, but the problem arises when it combines with bigger hands.


>> I also think its important to be able to wait to improve your hand into a better wait or better yaku without risking making yourself furiten. If the scoring system essentially forced players to reach *any* hand would be kind of ridiculous. You\'d see much less defense, for one thing.

Again, this problem only occurs with a zero yaku hand. The situation you describe is fairly uncommon. It\'s certainly less common than Menzen randomly pushing hand scores over a limit, which is a much bigger issue. It would definitely benefit pros, statistically speaking, to have the Menzen Tsumo bonus removed. This is especially true in tournaments where every hand counts. When you look at the numbers it seems hard to argue with this.




EDIT: to answer Pringle\'s Questions

1) What is the percentage of the hands going for Menzen (success or not) in Pro games and in Amateur games?

66% (averaged out between players of all levels)

NOTE: every closed hand by definition is going for menzen tsumo, so to calculate your own just look at your percentage of closed/open hands attempted.


2) What is average number of tiles they been waiting?

10-11 Tiles Until Ready

3) What is percentage of hands winning by tsumo?

20-30% on average

4) What is percentage of hands winning by concealed ron and not concealed ron?

Roughly 60-70% Ron and 60-70% of that is concealed. Simple math inferred from the previous numbers.

5) What is average yaku winning by Menzen and non Menzen (only yaku not han)?

Both the same, except Menzen +1 yaku on average (over non Menzen average). Logical. It makes no other difference.



>> Also, I think 20-30% is not high.

That is very high. Consider that 20%-30% is the same frequency as the most common hands such as Pinfu and Tanyao. More even. It\'s also almost double the frequency of Reach.

Re:History of Mahjong: Cheating

Posted: Sun Jan 25, 2009 12:17 pm
by WorTeX
Oh yeah pros would certainly like that tsumo would be removed, hey, why not remove riichi and add joker tiles also, possibly a fourth suit of tiles, also you\'d have to dance around the table every time you claim a tile, also you\'d have to sing out loud every time a honor tile hits the discards you also have to show every fifth tile you draw to others while playing guitar with your toes..... I think these improvements will clear everything in this game

seriously, tsumo doesn\'t give any edge to any player because you can not predict it. and as stated, if you have 0 points without tsumo with a _ready_ hand and you just drawn the tile you need, what are you going to do? furiten riichi?

Also, when you reach and you play with PROS, they probably figure out shortly what you might be waiting on, and the only way to get the tile is by yourself, and if that happens, i\'m always happy to take the reward from that..

Re:History of Mahjong: Cheating

Posted: Sun Jan 25, 2009 12:26 pm
by HotelFSR
I thought this was a forum for discussion. :dry:

Please consider this little more thoughtfully, I\'m saying all this not to be difficult but as a serious player and fan of the game.


Don\'t take my word for it. Look at the numbers for yourself.

If you look at Pro stats you will find that people get Ronned quite often. The statistical difference between Pros and average players is not as great as you seem to think. Tsumo is not the only way to win.

Winning is its own reward anyway without an extra bonus!


You are absolutely right that the Menzen issue affects everyone randomly and equally. That is precisely the problem. It adds variance, which means that it adds to the possibility that the more skilled player can lose out in a tournament.

Over many hands this is ironed out and becomes less of an issue, but in tournaments you don\'t play all that many hands. Hence it is a problem in competition.



If it did not effect everyone equally, then that would be fine because it would be something you could influence more and that would be a skill in itself. At the moment it stands out as the one single thing in competition rules which is pretty much entirely random.

They took out Ippatsu, they can take this out too. Its not taboo and I am sure this line of thought has occurred to plenty of pros as well. After all, someone in Japan had to come up with the A-Rules, didn\'t they?

Re:History of Mahjong: Cheating

Posted: Sun Jan 25, 2009 12:50 pm
by Benjamin
Look, I\'m really at a loss as to what to say to you. I\'ve been involved with the Japanese Mahjong world for years now and this is the first time I\'ve ever heard someone recommend getting rid of the tsumo yaku. So the very pros you claim are gypped out of winning by this \"extraneous\" bonus probably do not agree with you. Of course, that doesn\'t make you wrong.

There are, however, some things which you are saying which I think are flat out wrong.

--You\'re getting your stats from pro versus amateur games, right? Or are they pro-versus-pro only? Also, when you say 20-30%, are you talking about winning by self-draw or the frequency of menzen-tsumo. The 18.% figure I gave is for the yaku, not the way you win.

--Having a no-yaku hand that you want to win on but don\'t want to riichi on is pretty common. Heres an example:

(111) 23457 NN WWW (N is north, W is west. You are neither seat)

If (1) were dora here, most good players would not riichi this hand because the wait is so bad. Since you want a good chance of winning with this, it\'s best to not riichi and hope to draw to improve the wait. Changing the strategy so that its best to riichi this type of hand would frustrate many many people.

[Drawing 1, 3, 4, or 8 makes it a ryanmen (double-edged) wait. Drawing N makes it a nobe-tan, which is almost as good. Drawing 5 or 7 makes it a fairly good shanpon (2-pair wait) since you\'re waiting on north.]

Again, situations where its best to not riichi and wait to improve your hand are pretty common.

--This is barely worth mentioning, but this isn\'t the single-biggest thing in competition (you mean A-rules, right?) rules thats \"entirely random.\" How about your initial starting hand, the people you play against (in the bottom leagues, this matters), tenhou and chihou. How about the fact that the tiles you draw are random?

Okok, so I\'m being a little bit facetious, but the 1.5x bonus (at best) given by tsumo is NOT the biggest random factor that swings games. It just isn\'t.

--Also, *its not entirely random!* I really don\'t understand why you think it is. This goes double for high-level games. The level of defense in \"tournament rule\" competitions is incredibly high, such that the tsumo rate is probably even higher there than it is in normal games. Pros who understand how to construct their hands the best are going to tsumo more. It\'s not because they\'re lucky, it\'s because they\'re good.

--They didn\'t \"take out\" ippatsu or ura-dora. Those yakus are relatively new. Tsumo\'s one of the oldest ones. Yes, that doesn\'t necessarily mean its a good yaku.

In closing I\'d just like to remind that the A-rules are neither the only \"pro-level\" rules nor the most popular ones. Most Mahjong enthusiasts in Japan like to play with ura-dora, ippatsu and even red-fives!

Ok, I think thats enough of this for me. I\'m going to let Wortex continue my end of the conversation.

Re:History of Mahjong: Cheating

Posted: Sun Jan 25, 2009 12:52 pm
by Benjamin
Sorry, made an error in the hand I constructed, it shouldn\'t have a yaku. Change the Chuns to West.

Re:History of Mahjong: Cheating

Posted: Sun Jan 25, 2009 12:56 pm
by Benjamin
WorTeX wrote:seriously, tsumo doesn\'t give any edge to any player because you can not predict it.
But it does given an advantage! It gives an advantage to the player who makes hands that are improved by the greatest number of tiles. In other words, it helps people who are better.

Re:History of Mahjong: Cheating

Posted: Sun Jan 25, 2009 1:19 pm
by sir_seagill
HotelFSR wrote: Mahjong is, and always will be- indeed should be- a game that combines luck and skill.

The difference we are talking about here is the distinction between pure luck, so to speak, and what you might call partial luck.

Many probabilites can be influenced by the strategic choices you make as you play the game. Others are completely beyond your control and have nothing to do with your decisions.

You can choose how to play the tiles you are dealt, but Menzen Tsumo is completely random, unlike patterns you can aim for and choices you make based on the actions of other players.

Reach is also a gameplay decision that you make which has its own effects.

That\'s the difference.
Well, tsumo is also a gameplay decision. You have to decide to not steal anything. All the way until you draw that last tile.

Re:History of Mahjong: Cheating

Posted: Sun Jan 25, 2009 5:04 pm
by HotelFSR
Thanks for your responses guys, most of all to Benjamin.

I think we\'ve maybe gone far enough for now. I\'ll keep playing the game and learning and see how my thinking evolves.


The last I\'ll currently say on the topic, though, is this:


* Menzen Tsumo is generally worth more than 1.5x because it is a yaku.

* There are already very many rewards for a closed hand. They are also all not so random- besides Menzen, which also favoritizes hands like Pinfu at the expense of rarer hands.

* Having a good wait and winning by Tsumo is simply a reward itself, so why should it have a bonus? And why should you be allowed to win at all on a hand with no yaku? Is allowing this to happen worth the downside it causes in statistical variance? People have simply gotten used to it. Nobody complains that you can\'t Menzen an open hand with no yaku! These are just conventions that players blindly accept. Yes, a better wait is more likely to Tsumo, but also more likely to win in general. It\'s not specific, so in that sense it is totally random to reward the closed self draw in particular.



Finally- and most importantly- I submit to you:


* I completely accept that you have spent a lot of time in the Mahjong world and that nobody has brought this up with you. On this matter your blog, Benjamin, comes to mind. In particular, your entry about the general attitude in Japan towards statistics and an alarming belief in superstition that persists in questions of \"analog\" vs. \"digital\" play. Even a few legendary pros, as you mention, take a supersitious approach. Coming at the game from this mind set, it is not at all surprising that nobody will have brought up issues like the problem of statistical variance introduced by things like Menzen Tsumo and the Dealer\'s score boost. It just doesn\'t enter into the mindset of many in Japan, one might imagine. Most pros probably don\'t realize that they are being statistically screwed out of certain wins and just chalk it up to issues of \'nagare\'. Perhaps this will evolve over time, and with the exposure of Reach Mahjong to a wider international audience with a more scientific approach (e.g. the high level Poker crowd).


I know I am not the only one with this stance. Alan Kwan, the inventor of WSoM Rules (based on computational analysis) has gone on record many times stating he agrees with this.

Time will tell.