OpenTable’s Violation of the Principle of Least Astonishment

I was taking an OpenTable survey about a recent dining experience, when I noticed a poor piece of user interface design.

The last question asks about the noise of the restaurant.

Quiet – reasonable first choice.

Moderate – yup, still with you.

Energetic! Yes. What could be louder than energetic?

Do not recall.

This is a rather strange (and bad) choice for an option, as it doesn’t follow the convention of the other controls on the page (or that I’ve ever seen on any survey ever). I would wager that more than half of the entries in their database they get for the “Do not recall” entry were intended to be “extremely loud” but the survey participant didn’t even stop to see the text change but instead assumed that the largest value would correspond with the loudest entry.

This design is a clear violation of the Principle of Least Astonishment, since it violates the strong convention that on a survey, lower is to the left and higher is to the right. Furthermore, a ‘do not recall’ answer should either be inferred from not answering the question, or by an alternate control (e.g. a checkbox) that would disable the control with the noise level readings.