We all have human frailties - there's no such thing as a 'dumb mistake'

While the design was technically perfect - it did what it was designed to do - it resulted in about 20 unnecessary deaths in just three days.

• Atlanta Olympics bomb threat: a call was made to 911, "there's a bomb in Centennial Park, you have 30 minutes". Dispatchers couldn't find Centennial Park's Street address in the system, and couldn't add the call to the dispatch roster until it was entered correctly.

It took a vital 10 minutes to find it - eventually they called someone who knew the phone number, then rang the number and asked for the address. By the time they were able to alert the police, the bomb went off, killing one person and injuring dozens.

• Mission-critical piece of software: 80 per cent of staff in a large company client of ours used this system. It took 9 months to learn, but the staff using it worked there for an average of just 12 months.

It took six minutes for experienced users to do the most common task (which our new design whittled down to about one minute). When measured across all the staff over a year, this task alone amounted to millions of dollars in lost productivity.

Why do people design this way?

In an environment of imminent deadlines, financial pressures and creative flair, the focus naturally seems to fall on the technology or the product's functionality, rather than how humans will use it.
Some projects don't even have a designer and the most junior programmer ends up flinging an interface together in a couple of hours.

But usually it's because the people on the technical and design teams aren't typical of the people who'll use the system; they know things users don't know and see the system in an utterly different way.
Take the Florida ballot. The designers of the ballot needed to design a space for all ten candidates plus one extra slot for a "write-in" candidate. They saw it as a table of two columns with a bunch of rows in each column, with a line of buttons between them.

But the voters weren't looking at it from that point of view. They just wanted to vote for one candidate. This difference in outlook is vital - it makes or breaks a design.

The picture below shows how the actual ballot looks when attached to the voting machine, as seen from the perspective of a medium-height voter.

Say I'm a voter in that election, and I want to vote for Gore. The natural thing to do is to scan down the first column until I spot his name. "Ah", I'd think, "he's the second one in the list, so I'll press the second button". And that's where I'd stop looking, because I've found what I want. I'd tend to ignore the arrow, because I've found what I want and, like most people, I don't read instructions.

There's no reason for the voter to look at the other column at all, let alone analyse whether they've picked the right button or not. They just assume that they have.