Country : United States
IHNN posted @ 2022-03-14 9:27 PM
Thanks for a great contest with great puzzles. This time, I looked a bit more into the scoring system. First, I could not find any description how it works.
If we start at the "Leaderboards" page, there is a column "B" in the table with my personal points.
If I look at the "Statistics" page, there are various values I don't understand. I can't find the values in the "50th player" and "100th player" column anywhere else, especially not in the detailed ranking of the corresponding puzzle. (Also the marking HH:MM for these columns is probably wrong.).
Scoring is pretty simple - the "A" column is 101 - your rank (1st = 100 point, and 1 less until 100th gets 1 point). The "B" column is a flat reward for solving the puzzle at all, scaled based on difficulty with that difficulty calculated from the top results. The exact mechanism isn't important, but it rewards persevering to solve a harder puzzle over just being fast on the easy ones and giving up on the hard ones.
The "50th and 100th player columns are correct with HH:MM as that's how long the puzzle was available until 50 and 100 people solved them. Personally, I think the fact that most of the the 50th player times are in the multiple hours means that the early bird stuff is a bit of a miss, but some people clearly like trying to be in the first few and it barely harms anything to be there so whatever.
With that sum being 1600, it's save to assume that the main contributor to the ranking (beside solving all puzzles) are the "Type A" points. The contest started with 350 puzzlers and as expected, there is a slow decline. There were a stable 200-250 successful solvers per puzzle. Of these, up to 60% got nothing. That's just not a good idea in my opinion. Of course it doesn't matter if you can constantly get Top100 results, but a lot of people (even experienced puzzlers) can not. So for all but the maybe 50 people who can constantly get Top100 results, the ranking correlates very badly with the performance.
I do agree that the scoring staying at 100-1 points for 1st to 100th doesn't really work with the scale, as it didn't for Tapa Train either. But while Tapa Train had no idea of the expected level of participation, Rassi Silai Race should have known and I think should have adjusted slightly. I think the best proposed solution was to award 100-51 points for 1st to 50th, but then both 51st and 52nd would get 50 points, 53rd and 54th would get 49 points, and so on - extending the "tail" of the scoring.
Country : Turkey
bskbri posted @ 2022-03-14 10:44 PM
Maybe the multiplication of A and B would work better. Being in the top place for a harder puzzle should worth more points than an easier one.
Country : India
prasanna16391 posted @ 2022-03-14 11:00 PM
Country : India
prasanna16391 posted @ 2022-03-14 11:16 PM
I think IHNN has responded to most of RealShaggy's concerns and thoughts, I'll add this much:
- I agree that the scoring systems could be more prominently displayed. We will do that for future series, perhaps as a link by itself next to "Rules & Info" so that it is standardized and remains there. I think it was all mentioned in detail during the Tapa Train introductory posts but we failed/forgot to carry it forward here. That'll be rectified.
- We were a bit cautious with Rassi Silai about going beyond the "top 100 get points" system because we still had some concerns (maybe Tapa is popular so it had that many players, maybe the shading interface is better, etc.) and wanted to wait just one more contest before we changed that. We addressed the penalty system this time, and think that is good to stay, we will address the scoring now that we know the format itself has been successful over two very different genres. Since we are looking at this long term, we wanted that bit more information before acting.
- Regarding early bird rankings, as well as the scoring system, I want to remind people again that this is supposed to be a format where the priority is to encourage more participation. So we largely want to keep things light and easy to understand. That is mainly why I agreed with the changes in points above, but I also want to caution that we will be against any complicated solutions and formulae. This is not meant as a response to the posts so far, I like IHNN's suggestion and that'll definitely be something we keep in mind when discussing tweaking the scoring. I'm just pre-emptively mentioning this so that people know what we are going for.
- I welcome more feedback and discussion, hopefully from newer solvers as well.