Yeah, I had an underlying assumption in my mental model that they would have split the questions so that half were yes and half were no. But you are right that that assumption may not hold. A random programmer with a random program will be a 'no' far more than 50% of the time.