It may be a sanity check; often when writing parsers for "in the wild" data you have to pick a cutoff for where "doesn't make sense" is so that you can tell if you're looking at junk data and stop trying to make sense of it (and the RAM / CPU entailed in such).