IMHO, the expected order of the answers and the expected duplicate solutions for 1 query, both greatly restrict the possible algorithms and approaches to tackle this problem.

Only the exact algorithm used to produce the expected results would be validated.

For such a complex task, why not put more effort in a more relaxed answer validation ?

Any administrators ?

How often do you guys scan for duplicate solutions? :)

Is there any arrow parsers installed for Haskell?