Unless you're writing programs for a bunch of burned out computer geeks, your user isn't you. ... This is very hard to get through somebody's head; it's very hard to get rid of this notion that what you like your user is going to like... Again, your user is not you.Platt urged his audience to consider the needs of the user, and not the developer, when designing software. While this may seem obvious, he used several quick survey questions to drive home his point: that the users of his audience's software are very different from the developers themselves. For example, through show of hands, Platt identified that most of the audience drove a car with a manual transmission, something that is harder to learn, harder to use, but gives you better control. Apparently that crowd of developers found this a good trade-off. In contrast, Platt pointed out, only 12 to 14 percent of automobiles sold in the U.S. have manual transmissions! Clearly, his was not a "typical" crowd when it came to car design decisions.
Normal people do not drive stick shifts. Why? Because they don't care about the driving process in and of itself. It's a means to an end. They don't want to drive somewhere; they want to be somewhere.Users have work to do: goals to achieve, people to communicate with, errands to complete. Our software is incidental to this process... when it's working well. When it isn't - if it's awkward, intrusive, or forces unnatural workflow on them - it's getting in their way and this kind of visibility probably isn't good for our products. The classic example of this, of course, was MS Office's "Clippy," the helpful paperclip (now, finally, extinct), but less obtrusive annoyances can garner equal negative attention.
[laughter in the audience]
It's an important distinction. You think your users want to use your software. They do not want to use your software. They want to have used your software."
In the spirit of "do the simplest thing..." Platt suggested we need to "Make It Just Work" and offered five points he considered essential to make software "just work":
- Add a novice to the design team - someone that doesn't know the implementation of the software.
- Break convention when needed - the old way isn't necessarily a good way.
- Avoid feature "silliness" - don't let the obscure features get in the way of the most desired features.
- Instrument your application very carefully - it's hard to find out what the "silent majority" thinks, so useability testing can provide hard data needed to carefully think through instrumentation.
- Consider whether design decisions are taking you closer or farther away from the software *just* working.