BUILDING A SHORT HIGH CEILING TEST
(and having some fun doing it)
Chris Cole

Our Editor is on record decrying the lack of a short, high ceiling test. This test could be used as a means of recruiting people of high intellect who are not willing to put in the grueling effort required by tests such as the tests of Ron Hoeflin and Kevin Langdon. Ron and Kevin formulate their tests by publishing long trial tests and selecting the best questions from them. Can we use this same technique to build a short high ceiling test (and have some fun doing it)?

What constitutes a good question for such a test? It is easy to formulate questions that are too hard -- pick your favorite unsolved problem of elementary arithmetic (e.g., Fermat's Last Theorem). It is also easy to formulate questions that are too easy -- pick up your favorite Mensa puzzle book and open it to a random page. We want a question that is just right. An idea: a question is just right if exactly one member of the Mega Society can solve it within a reasonable time.

It is hard to formulate such questions. Some people, such as Ron and Kevin, are very good at it. It goes without saying that answering such questions is challenging. It seems reasonable that such Herculean effort should be rewarded. But how?

Suppose we charge a scoring fee for people who wish to take our hypothetical test. Ilis is certainly reasonable and customary. Suppose that we charge \$10 to score the test. Suppose half of this figure (\$5) goes to the person actually doing the work of scoring the test (to cover expenses, mailing costs, labor, etc.). This leaves \$5 per question [read "testee"  --KL]. Suppose 1000 people take the test. This seems reasonable, since several thousands have taken previous tests that were much longer. 1000 takers times \$5 per equals \$5000.

I propose that we distribute this \$5000 evenly over the twenty questions on this hypothetical test, paying equally to the author and the solver of each question. That would be \$125 for either creating an ingenious question, or coming up with an ingenious answer. I propose that authors send their submissions (limited to one question per issue-don't forget to include your solution) to the Editor. Each issue, the Editor will publish our best questions. Solvers have until the next issue to submit answers. If no one can answer a question, it is disqualified; if two or more people can answer a question, it is disqualified. As soon as twenty questions qualify, the contest is over.

The Editor will then attempt to get the resulting test advertised and to get someone to serve as the scorer.

So, fellow members, rev up the little grey cells and send your best questions to the Editor. By the way, if any of you previous test publishers wish to submit questions from tests that have already been published in Noesis, that is fine. You of course run the risk that people have been working on them for awhile.