MAXIMUM POINT!
YAKUMAN!

Whenever I wrote a blog, I will try to tell my riichi mahjong story, from normal win to super lucky win, I put it at the start of the paragraph before starting the actual blog.
But for this case, the whole blog will be about riichi mahjong so this is the first time I write a full passage about my map xc_yakuman
As always, I will try to make this blog not that short and not that long, it will be somewhere in the middle.
Also I am trying to format this blog not like before so it will be more readable I think? With that said I am also going to write the content of the blog in document first
before implementing it to html.

Before this, I wrote all of the blog content straight on Visual Studio Code. It is not a fun experience and it is very time consuming, thinking about it makes me not want to write a blog.
That is the reason why the Polusi blog split into two parts, since writing without a draft is not fun at all.

With all of that said, let us go to the main blog!

What is xc_yakuman? And why did I make this map?

xc_yakuman is an experimental xc map that does not play out like your usual xc maps. With the nature of the map being a “rat” type of map that is when the platform is bigger than the player, designing your usual xc map is quite difficult.
This map is actually an old project from last year that I tried to make, but without vscript knowledge, I shelved it for later when I had time and ideas to make it work.

Here are the screenshot of the whole folder containing the map:

As you can see it was around December last year when I tried to make the first draft of it. The idea of the gameplay is somewhat similar to this current build, that is to collect yakuman while navigating the map, but the current gameplay is a later idea when I revisit it this August.

Before this, I wanted to do your usual xc map gameplay with the tiles floating around the map so that the player can shoot to store a tile and build a yakuman hand!

Now we move on to why I made a riichi mahjong themed map? It was around late last year when I started playing riichi mahjong. I knew the game before but I did not want to play it until one of my friend that is Green Tea, played it. I asked if it is a good game? Is it hard or not? And then I just started playing. Played with bots first and then real players after. The conclusion? I had fun playing it and I think it is the best game to play casually or competitively.

Riichi mahjong as a game is somewhat similar to chess in my opinion, people back then portrayed chess as this game inspired by warfare, since it is a 1v1 game, it looked like a game where side A vs side B. You still see it nowadays in the media when the writer wants to show that this character is smart, or something…

The difference with chess is that riichi mahjong is a 1v1v1v1 (4 players) or can be played 1v1v1 (3 players), people usually play the former. Riichi mahjong is more relevant in this day and age since you feel like you are fighting from all sides, not you versus your opponent like chess, but you versus yourself, your opponent to the left of you, your opponent across from you, your opponent to the right of you.

While chess is in my opinion 100% skill
Riichi mahjong is 70% skill and 30% luck

Learning to play riichi mahjong is a better skill to learn nowadays, since it is somewhat can be applied to your life!
Rather than chess that is 100% skill and little to no luck involved.

You can learn both if you want, and I am not stopping you to learn only one of it or both, it is just in my opinion, learning and playing riichi mahjong is better!

That is why I made this map! I want to show you that riichi mahjong is a fun game to learn and it is somewhat relevant to your everyday life so it is not a waste of time learning how to play it.
Let us be honest with ourselves, even if you do everything right in real life, sometimes it will not play out like we think it will. Even with the most optimal discard you do in riichi mahjong, waiting for the perfect 3 sided wait, a confirmed baiman hand. It can somehow lose to someone with open tanyao tanki wait.

So with all of that said, let us move on to the making of xc_yakuman!
How do you make riichi mahjong interesting to people that have never tried it before? Explaining the complicated rules of it? This will make people lose interest super fast.

We use the easier method instead, that is introducing the YAKUMAN hand or the BIG HAND or RARE HAND or the DOPAMINE HIT hand.

With a simple tutorial on how to build a hand that I implement it onto the map and the list of the YAKUMAN hand that you can get, now it is more straight to the point.

The last map I made was xc_polusi and I tried AI to help with the development of it. Super useful tool to speed up my work with the input/output for the system I used on that map.

After releasing that map, I got a promo on my data provider that gives me a free 6 months of AI plan. I obviously claimed it and tried to experiment on the AI.

And there is the idea of revisiting last year's project that is this map to try and make the system work with vscript, also this vscript is 100% vibecoded since I did not learn the programming language Squirrel at all. I started the project in early August and it took me 1 week to vibecode the whole system with vscript and make the tiles in blender using AI as an assistant to give me a step by step tutorial on how to make the tiles.

Took a week break and started to continue on working the map on and off.

Doing this gives us a challenge that is the AI is sometimes hallucinated or too confident on fixing stuff that is not a problem before thus making it a new problem to fix, and then breaking the whole system all together. That is why I started to do frequent backups and make the prompt more specific on what to do and what not so the AI will be more predictable on making the code.

If i were to give you an analogy, an AI agent is like your fresh graduate, top of the line student that has all the internship experience in all the big companies, they are super smart, probably have 300+iq, super rational, etc. But without giving the training on what to do on this job, the super smart fresh graduate can only guess on which is the most optimal way to do the job or solve the problem that will satisfy the company. That is why you need to be specific on telling the AI agent on what to do or not to do so it will give you the best interest in mind!

I do not recommend vibecoding without any knowledge at all, since I know how to use the hammer editor in and out (probably) and stress test this, I think I did a good job on this one!

Think it will still have bugs on it, but I think this is enough for it to be playable.

If I were to learn Squirrel, I think I would not release this in a year let alone around a month like this project. AI is a good tool moving on that speeds up some task that is sometimes a hurdle to do or maybe a task that needs automation. A good mix of AI use is with human supervision, since you need to double check everything before releasing something with the help of AI.

Ah, we are nearing the end of this blog!
Before ending this I think I want to share my JLPT test result and what is my next plan after this.

I passed the JLPT N5 with satisfying results. My weakest part is grammar, which is expected since my mother tongue is Indonesian and the Japanese word order is different from it so I still have a hard time on grammar, like the particle placement or conjugation.

I still learn everyday, and I think it is getting better day by day. I could read and listen to some simple Japanese sentences most of the time, but making a whole sentence is still a skill that I need to learn next.

Since I could only read or listen and could not make a whole sentence myself, the translation of the map was AI assisted with me proofreading it by opening the dictionary. I tried my best on the proofread so if it is still wrong, please forgive me.

I do not plan on taking N4 since I just wanted to know the experience of taking the JLPT test and I do not plan to work in Japan, this is just a hobby of mine. This could change in the future if I were to change my mind.

Here are the screenshots of the result:

My next plan is probably to try and livestream a map making process with a theme similar to xc_kanji.
I already did some test runs of livestreaming it and I think it is a good enough setup that I can stream comfortably.
Since I just freestyled on making xc_kanji, I think it would be interesting to livestream the whole map making process from start to finish.

Also, it has been one year since I made this website in late July last year and the website is only made using HTML and CSS without AI vibecoding. I wrote all of this by hand and while this is an important skill to learn and practice, it is very time consuming. But it is worth it!

Moving on, I think I will try to let AI assist in cleaning this website code, since I learned HTML and CSS this past year and I can proofread the code enough so that the AI does not mess up my website. I will try to make it HTML and CSS only since it is good and fast enough to make my personal website work.