In many previous compression tournaments, the aim has been to compress
some fixed body of text - and the metric used to measure the compression
ratio measures the total size of the compressed data plus the
size of the compressor.
This results in a pressure to squeeze the compression code.
With a challenge based around a fixed body of text, this is necessary -
since otherwise the compressor could be made so that it contained the
target text, and simply produced it on request.
This problem is usually dealt with by compressing a large volume of data -
so that the size of the compressor is of low significance.
The Matching Pennies Tournament is a bit different in that
There is no fixed text to compress, but rather the output of competing
programs. Not all the details of those other programs are known to
contestants in advance.
There are still size constraints - but they are some distance
out - and competing programs may grow to be quite large,
without being penalised for doing so.
This should hopefully reduce obfuscation of the code - or other
measures that decrease size at the expense of clarity or readability.
Prize for Compressing Human Knowledge (Hutter Prize)
The Calgary Corpus Compression & SHA-1 Crack Challenge
Tim Tyler |