Week 9 - A Step Up

This week I had an epiphany of sorts regarding tile efficiency. I also had a pretty amazing streak of results, but not as amazing as I'd like.

Before we get into that, though, a quick update regarding the Efficiency Trainer. Recently, I added localization support, so now it can be translated into whatever language people want to translate it to. If you're bilingual, and would like to expand its audience into a new language, please see the GitHub repository. I'd greatly appreciate any translations, even partial ones! Check the list of forks or ask on Discord to see what translations are currently being worked on, if any.

Now, onto the Mahjong. I mentioned having an efficiency epiphany. A lot of times, people will say things like, "I get rid of bad shapes early," or "I value shape over efficiency," or "there are times you want to sacrifice short-term efficiency to gain long-term efficiency." They're kind of unspecific, huh? Surely there's a way to measure that, other than just spamming HMS and learning over time?

Well, something clicked with me a few days ago. You may have heard the advice to riichi anything other than a bad-wait non-dealer riichi nomi (unless it has enough value to dama for win rate). I followed this advice, and if I got to a bad-wait non-dealer riichi nomi tenpai, I'd break it and try to rebuild. But, it's better to just not get there in the first place, right? Look at the following hand.

If we cut the 4m, we get to the widest iishanten, accepting 78m6p3z, 13 tiles. Or do we? If we get tenpai on any of those tiles, we get a bad-wait non-dealer riichi nomi. If we'd just break the tenpai, we should view that as not being tenpai. So, that iishanten actually has a worthwhile ukeire of zero!

So, what upgrades does that iishanten have, that will get us a hand that won't be bad-wait riichi nomi? 6m for the 67m ryanmen, 4p for the 45p ryanmen, 0m for the dora, 0p for the dora, 8p for the 78p ryanmen. Thirteen upgrades.

What if we didn't discard the 4m, and instead discarded, say, 7m? (I've been told 7m, 9m, and West are all good discards, but 7m makes the most sense to me at my level of understanding) We'd lose the 7m3z ukeire, so it goes from 11 to 8 ukeire. However, the worthwhile ukeire is still zero, so we don't actually lose anything!

What we gain is upgrade potential. Since 4m is dora, anything that connects to it will give us a worthwhile hand. So, we can add 2345m to the upgrades, going from 13 to 24 upgrades. This makes it much easier to get to a hand we'll actually be able to riichi with!

This idea of "worthwhile ukeire" is what made things click for me. I don't know if the idea has been mentioned before and I just didn't understand it at the time. I'd be kind of surprised if I was the first person to put it into writing. You can extend it further. For example, if you need a mangan in S4, then any ukeire that leads to a non-mangan hand isn't worthwhile.

It also gives a new explanation for "reinforcing weak shapes." For example, if your last two shapes are 446m and 556p, the first shape has an ukeire of 6, while the second has an ukeire of 10. 4->6 is more significant than 8->10, so you'd want to cut 5p. Another way to think of it is that, if the second shape completes, you get a bad wait, while if the first one completes, you get a good wait. So, it's 4->6 worthwhile ukeire vs 0->0 worthwhile ukeire.

Does this stuff make sense to anyone that's not me? Is it common knowledge? Is it super obvious stuff that people figured didn't need to be said? Is it helpful? Should I write a full article with more detail and examples? I don't know.

As for my recent results, the focus of efficiency has helped me out a lot. I might also be getting acclimated to the joukyuu life, and there's probably a good amount of luck involved.

Still eleven people luckier than me, though. With the revelation described earlier, I've been putting a lot of focus on that. Perhaps too much focus, since I've been doing it at times that I don't need to, which is hurting the "pure efficiency" playstyle I adopted in the previous post. Let's look at some examples, from this replay: https://tenhou.net/0/?log=2019071108gm-0089-0000-f757cf45&tw=1

 We can sort of apply the same idea to open hands. In this case, I have the pon of red dragons, and the dora 6p, so I should just cut the east instead of holding it for the 3 possible value increases. We have a hand that will be "worthwhile" so we can just go for efficiency. The 3p is too good of a tile to cut at this point.

 Here, cutting the 12p is going back in shanten in order to remove a bad shape. However, we already have the 0s incorporated into our hand, so we won't end up as a bad wait riichi nomi. We can just cut 7s and then cut one of the penchans if the 0p or dora 8p evolve into a shape. Though it's probably fine to cut the 12p this early to hope for the 58p7s to become a better shape, that's another step up that I'm not at yet, so I'll ignore it until I understand it better. Or maybe it's not fine. Who knows. Not me.

Now, this was just silly. I think I got so caught up in looking at my hand's blocks that I forgot 8p was dora. Careless. Just cut the honors, silly Amber.

So much for a short post, huh? Well, if I keep this pace, I should be able to get Tokujou by the end of the month, I think. That'll be pretty exciting. Let's see where I'm at next week!