Boxer shorts have a very simple interface. One hole at the front with some buttons, two larger holes for legs and one large opening at the top.
Simple as pie. And it isn't that I'm inexperienced at putting on underwear. I've done it way more than 10,000 times so far.
Maybe three times a year* [*Of course this isn't the result of a properly controlled study.] I put my boxers on the wrong way round. Whatever is causing my usability problem it isn't a lack of experience with them.
The interface is simple and I like to think I'm a smart guy. And yet… Even so… Getting my boxers on the right way round each and every day defeats me. Not for the 99% of the days I get it right, but for the 3 days a year I get it wrong.
What does this tell us? Maybe that boxer shorts should be even easier to use. Those tricky devils!
What we need are better boxer shorts with the front a completely different colour to the back!
If we standardise our shorts with blue fronts and green backs we can all be completely “boxer proper™” every day of the year. Any company wishing to advertise their “Standards compliant boxers” have to use those two colours. Blue front, green back.
I'm sure that studies will show that the error rate will fall. Boxer industry testing will prove that 99.9% of boxer application events are completed successfully—ten times as many!
Somebody will then realise that we need to have a solution that caters for the colour blind who make up most of the 0.1% remaining error rate.
“Standard Boxers 2” is a new standard which has green and white stripes at the back and solid blue at the front. Testing people who have adopted the new standard will prove without a shadow of a doubt that the error rate is 0.01% — that's four nines reliability now! Something to be proud of for sure.
But, there's probably more than a billion incidents of boxer application a day. Even at four nines that means more than 100,000 incorrect boxers a day. Maybe a new standard can get us five nines?
So what does all of this have to do with software? The first lesson is that no matter how good your interface is you cannot possibly reduce the error rate to zero. This isn't a very new observation, but it carries a corollary that many people still fail to recognise. We'll get to that in a moment.
The second lesson is that the reason why getting your boxers on the wrong way round isn't such a big deal is because it's easy to correct the mistake. You simply take them off and put them back on the right way round.
If you can't stop people making mistakes no matter how hard you try then you have to accept this as a fact of life:
Your software must allow the user to correct each and every thing that they've told it. Even after they told it they were sure.
It doesn't have to be painless, but it does have to be possible. The corollary of all of this is that the higher the error rate the easier it has to be to correct the mistake.
Of course this is an idyll, a Utopia, but that doesn't mean we shouldn't be striving for it and it does mean you need a jolly good reason if you can't allow corrections!
How close is your software to passing the Boxer Usability Test?