About Psyho (twitter: @fakepsyho)

Entries

 
Ludum Dare 32

Psyho's Trophies

Psyho's Archive

My thoughts on Ludum Dare (and how to make it better)

Posted by (twitter: @fakepsyho)
Sunday, May 10th, 2015 4:24 pm

I highly doubt that my thoughts are going to be original, considering the long history of LD and the amount of participants it has. For the record, this was my first LD, I loved every part of the LD throughout the first 48h – even though it meant that I spent my almost entire weekend alone and working :) But that excitement quickly left me as soon as the voting period has started.

Since no one likes long posts, I’ll directly dive into the problems:

1) I thought I’ll leave people valuable feedback on their games and (hopefully) this will create some interesting discussions about the games itself. Ideally, no one would feel bad about constructive criticism and we would end up in a win-win scenario. But it’s not possible right now, since: (a) there’s no threaded view of the comments (b) there’s no way of tagging people. It’s one of the (smaller) reasons why most of the comments are completely shallow, since the discussion is impossible and has to move towards the personal blogs.

2) If I understand correctly the final result is just an average of votes. If I’m wrong please correct me, but I guess at least some of this still applies. This is really bad idea and easily fixable. (a) First of all, bayesian average is a trivial improvement over it, which in general decreases the variance of the result (no more taking top spots by submissions that had votes barely over minimum). (b) The other easy thing that should be much better is to use percentile rank. You can also drop top and bottom ~10% to reduce the effect of the outliers. And yes, I really think that if someone’s giving only 1s or only 5s, then those votes should transcribe to the exactly same score. (c) Ideally, the proper way would be to toy around with the machine learning, considering the prediction of the “true” ranking should be a rather standard problem in recommendation system. It’s actually absolutely the same problem Netflix prize used if you assume that we’re predicting ratings only for the “fresh” users.

3) If I stick only to games recommended by “Play and Rate Games”, I’m going to play mostly unfinished or just very weak games. It’s because the fun and interesting games have a lot of votes already due to the self-promotion. After 5-10 games that may have no gameplay, are completely bugged or someone was clearly just starting his/her adventure with gamedev, I’m beginning to feel frustrated. This is a pretty sad situation since this forces me to hand-select the games, which in turn completely skews the rating system and it makes even worse situation for everyone else. I guess in the past there was absolutely no system at all (only hand-selection), so nowadays people look at the coolness system as something that is an improvement over what LD had before. There’s very little you can tell about the games that are completely unfinished (or sometimes even not started). In those games the feedback is either the same, or is purely motivational. This means that weak game won’t suffer too much from having fewer votes/comments. This makes since at the same time, we want for good games to have as many votes as they can, in order to reduce the variance of the top places. (a) You could use the current rankings to adjust the priority in the “play and rate” queue. If people would get more fun games in the queue, they would use it more, so in the end the weak games would still end up with decent number of votes. The main difference is that we would reduce down the huge spam of “PLEAAASSSEEE RATEEE MY GAMMMEEEE”, which honestly is not the best thing in the world. (b) Somewhat risky idea would be to stop counting the votes of hand-selected titles altogether. Or alternatively, reduce their weight (and added coolness).

Edit: Also, please read my reply to tompudding. I think I did a better job there explaining the main idea behind (3).

Anyway those were my 2 cents. If this stuff was brought up already then I hope someone can give me links to those topics.

Also, no link to my game since this would defeat the purpose of the whole post :)

[cache: storing page]