This is an old revision of the document!
Determine a unified means of output so that all submissions have an identical format.
Derive a set of tests that all submissions should perform to ascertain correctness (state the tests, the inputs, and the expected outputs). In conjunction with conforming output specifications, all submissions should match (this is the basis for writing a verification script that can automate the process).
Which, being said: once output specifications and verification tests have been established, anyone writing a verification script to automate this can be eligible to receive bonus points.