This is the final exam for The No Free Lunch Theorem. It pulls together the whole course: what the theorem claims, why the tie is forced when you average over all problems, why methods nonetheless differ hugely on the structured problems you actually face, the idea’s many everyday faces, and the paid lunch you can have. Take your time and reason each one through.
How this exam works
Read carefully — this exam is final. Each question appears one at a time. Once you submit an answer it is locked for good: there’s no going back, no retry, and no restart. Your score is hidden until the end, where you’ll see a pass/fail verdict. The pass mark is 70%. A few questions ask you to select all correct answers.
Which sentence best states the No Free Lunch theorem?
Select an answer to continue.
Course Recap
Big picture
The No Free Lunch theorem, in one picture
- No Free Lunch
- The claim
- No method wins everywhere for free
- Why it's true
- Over all problems, wins and losses cancel
- The catch
- Real problems have structure — methods differ there
- The paid lunch
- Buy an edge with knowledge; match prior to structure
- The claim
Key takeaways
No method beats all the others on every problem — averaged over everything, they tie, because each method’s wins are paid back by its losses. But you never face “everything”: real problems have structure, and there a method whose assumptions (its prior) match that structure wins big. So the free lunch is a myth, but the paid lunch — an edge bought with genuine knowledge of the problem — is a feast. Don’t shrug that nothing works (nihilism) and don’t crown one favourite (the universal hammer). Read the structure, match the tool, and know the edge of your own competence. Carry that, and “what’s the best way?” becomes a question you can finally answer: best for which problem?