It’s just something I wrote for myself, mainly for use in the commentary booth (as the URL suggests), where quick and correct math weirdly seems to be more elusive than in most other places.
It’s meant to give a quick idea of the possible outcomes at a glance by color coding, and to provide more details in actual English that, in a pinch, could just be read off, in case the magic of the booth causes all other mental capacities to fail.
During the last ball, players can be ordered by their current scores in order to get live updates on the possible outcomes.
Very nice @umbilico. I really like your conclusions. Quite different to mine and very useful. Its a hard thing to condense to make useful and readable in the heat of the moment. I’ve got similar color coding coming soon too.
Thanks, I’m sure you also realized, working on yours, that that’s the bulk of the problem – just doing numbers would make it the quick weekend project @gdd mentioned.
With mine, most work went into condensing the numbers into the least amount of information that’s necessary to understand what’s going on, e.g. making it figure out that it’s better to write “unless Bob comes in 2nd” instead of “if Bob comes in 1st, 3rd or 4th”, or preferring to state the conditions based on players that can still influence the result.
What was the thing that needed the most amount of work in your project? Also, out of curiosity, is there actually any sort of AI involved in yours (apart from whatever generated the logo )?
Yes, pinwin has 1500 lines of code just devoted working out the conclusions. Which is more than half the total code base. The UI is 800 lines for comparison. The conclusions code got a lot more complicated when I added the predictions over the top of it which both changes the permutations but also changes the ‘if player x comes nth’ logic because its now set where they actually come.
I started this project a year ago, and was building out a mobile app with this feature and 2 other modules in it, but that started to get too complicated (I got a bit ambitious), so I abandoned that, ripped out the conclusions code and built a very targeted web app around it.
There is no AI in the this module. When I did start again, ChatGPT 3 was out and I did try to get it to draw the conclusions on its own - but it couldn’t do it well enough. I’ve just tried it again now with ChatGPT 4 and it does a much better job.
One of the other modules I have in the works does however use AI so I thought I’d use the .ai domain for that - and also because its trendy for anything with an ‘if’ statement in it!