You have used a gas hob. You turn a twisty knob to control the rate that the gas is released, and you press a button to light the gas.
A problem is that when the gas is turned down too low, the flame can go out. If this happens, gas leaks silently into the house. Light it, and the explosion destroys the house and everyone in it.
A design solution
A designer once decided to solve this problem. They hoped to solve it by preventing users from being turning the gas down too low —that is, the hob must either be off, or it must be above a minimum safe gas level. The designer also hoped to make it easy to distinguish between being set to off and being set to low, so that users could easily identify a dangerous leaking hob.
Their solution was this user interface, now used on millions of hobs:
As before, the orientation of the knob controls the gas level. But now, the maximum gas level is taken by the middle position that the knob can be in. From there, the gas level decreases towards both ends. One end is the ‘Off’ position, and the opposite end is the ‘Low’ position, which is the minimal safe gas level.
This allows users to ensure that they don’t turn the gas down too low. The user is supposed to put the knob either in the ‘Off’ position or somewhere between ‘High’ and ‘Low’. The design also makes it very clear whether the gas is ‘Off’ or ‘Low’.
A design failure
There’s just one problem. Users don’t do what they were supposed to. In my unscientific sample, everyone wanting to put the hob on a low gas level turned the knob not to the ‘Low’ position, but instead to an orientation near the ‘Off’ position. They never use the positions between the ‘High’ and ‘Low’ positions. Here are the possible orientations of the knob, showing which ones are valid, and which ones are used:
Why don’t users use the knob correctly? There are many answers. Perhaps because they didn’t read the manual. Perhaps because the knob is an affordance which tells the user that it works like their bathroom tap and the volume control on their loudspeaker. Perhaps because the knob seems to work even when being used incorrectly.
A few people are dead today because they didn’t use the knob in the way they were supposed to. We could blame the users and chalk up the deaths to natural selection. But a better response would be to say that the design is a failure.
Quantifying design success
Abstracting away from the specifics, the knob is a state machine. There are two big problems with the design: (1) users can put the device in invalid states; and (2) using invalid states is more intuitive than using valid states.
I illustrate this in the following diagram, which divides up states into ‘valid’ and ‘invalid’, and shows which of these states are used. States which are invalid and yet used are marked in red; states which are valid and used are marked in green; all others states are marked in gray.
This diagram lets us quantify the success of a design: green minus red. By this metric, the designer’s proposed solution is terrible.
Better design solutions
The diagram points us to several ways we can improve on the design:
- Make the invalid states impossible. We could do this by restricting the knob to move between ‘High’ and ‘Low’, and introduce an additional ‘On/Off’ switch. All combinations of these two controls are valid states.
- Make the invalid states obvious. We could make the hob stay off in all the states between ‘Off’ and ‘High’, instead of being a gradual change. Or we could introduce a notification light to tell the user which states are supposed to be invalid.
- Make the invalid states transient. We could make the knob ‘flick’ between ‘Off’ and ‘High’, so that it is not able to stay in an invalid state without being held there.