Michael's Workbench

Reading Why yesterday, I ran into a reference to interesting research experiment. Shank, Pearson and Dickinson in 1989 would display a triangle on the screen in response to user pressing a key on the keyboard. Users would then answer a question as to whether the triangle appeared in response to the keyboard press or not. The finding was that if the triangle appeared within 2 seconds of keyboard press (or faster), more users were confident that the triangle was the result of their key press. If the triangle took longer to appear, high percentage of users would not identify the key press as the cause of the triangle appearing.

In 2003, Buehner and May ran a similar experiment (with a light bulb instead of triangle), but they gave the users prior instruction, telling them to expect a delay. When the delay was always the same, users identified the cause with delays of up to 4 seconds. When delays were inconsistent, the same 2 second boundary was the point where users stopped linking the light bulb turning on with flipping the switch. Interestingly, when users were told that the light bulb that takes longer to turn on was green and energy efficient (making them feel good about the delay), cause identification improved.

Why am I boring you with this psych 101 stuff?

Well, think about this experiment differently:

A user chooses an action in the UI. If your application takes longer than 2 seconds to do something in response to that action, the user thinks applicaiton hangs – pressing the key does nothing. Anything your UI does with a delay of 2+ seconds is likely to be perceived as a problem – application hanging, poor performance, etc. The user’s natural reaction is to see anything you did at 2+ seconds as unlinked from the previous action by the user, and even though they understand it not to be the case, they are far less happy with your app.

So, keep in mind: you got 2 seconds!