Thanks to tails (for both creating a better disassembler and submitting many solutions). We can have some nice statistics on op usage. This information could be useful for modifying Nibbles or anyone creating future languages.

Also thanks to shinh for hosting a well organized golf server. It would be more difficult to do statistics like this from the stackexchange problems because there isn't a standard way to document input and many problems will use command line args of which their type needs to be known at compile time (one would have to manually curate inputs).

Code to generate data is on github.

Analsysis on March 4, 2022

It is difficult to draw many sound conclusions because even though the data set contains 200 solutions, many are short, and when you break binary ops down by both argument types, the counts aren't that high. They could easily be biased by which types of problems have been solved. None-the-less, a few things to note:




Action items:

Feel free to let me know any other conlcusions you draw!

TODO stats on implicit ops

Nibbles versus other languages on

As of March 1, 2022:

Comparing to GS2 (leading language at and Jelly (leading language at stackexchange)

For problems where both Nibbles and Jelly were used

For problems where both Nibbles and GS2 were used

One possible bias is that Jelly was created mostly to use command args instead of stdin (although stdin is supported too, but I'm not sure how well) (all input is via stdin on this server). From random problems I have done on nibbles won 7, tied 2, and lost 2 (versus Jelly) so it isn't too far fetched. One note about jelly on golf.shinh is that it is converted to code page indicies to be represented as bytes, so there are no issues there with its unicode chars taking more than 1 byte.

Reading too much into this data would a mistake as it could be gamed, but at face value it is promising.