> But then I don't get why the logs need to be in a undocumented binary format, inaccessible to anything else.
Because with this you can achieve a lot higher compression ratio if you store metadata separately. Even higher than using bzip2 on the xml or json. Not to mention lookup and search performance on very large datasets.
It would be nice to have different back-ends for this though. Since you only get an API to access it anyway, the storage format could be made transparent.