Codidact Fractional Byte Consensus
Over on the Code Golf StackExchange site, there is a meta consensus that answers can be scored in fractional bytes if there exists such an encoding method.
So, to quote the linked consensus question:
For languages with fractional byte code pages, such as half-byte code pages or huffman coding, should functions be allowed to be submitted without padding to a whole number of bytes?
This came up when I went to write this answer and then realised "hang on there's no consensus about fracbytes here yet"
1 answer
I'm going to reiterate and build off ideas I laid out on this post on PPCG.
First let's point out that at current the leader board does not support fractional scores. In this answer I am going to assume this is unchanged. That may prove to be false, in fact I think it is likely a false assumption, but I think it's a useful assumption for the sake of this answer.
What does it mean to allow?
The question asks if fractional bytes should be allowed, but I want to dig into what that means.
Answerers are allowed to pursue other goals than just byte count. They are allowed for example to try golfing a problem while at the same time implementing the optimal algorithm. People are allowed to try golfing using restricted methods. People are allowed to golf in Java.
If people pursue a lower fractional byte score is anyone going to stop them? It seems like disallowing fractional bytes can only mean forcing users to round their score up to the nearest whole in the title of the post. This means users are still perfectly capable of noting their fractional score in the answer. And if one answer is 4.3 bytes and another is 4.97 bytes then everyone is going to know that the 4.3 byte answer is the one that's really shorter, even though the "official" score says they're the same.
Allowing fractional bytes isn't that much different. The leader board doesn't interpret fractional bytes correctly so comparison of fractional bytes still has to be done by individuals. It seems at most you could say that the "disallow" solution standardizes the way scores are presented to the machine for ranking, while allowing leaves anything of that nature unofficial.
What should we do?
With respect to the above, I think we should try to update the leaderboard to support fractional bytes. If people can and will score their answers in terms of fractional bytes regardless, then updating the leaderboard to reflect the reality of how people are scoring is probably for the best. This is a descriptive approach to the leaderboard.
0 comment threads