The USS Quad Damage

User Friendliness vs Expectations

Old ladies should NEVER be trusted

Last night I saw an episode of Top Gear where the presenters got their mums to test out some micro cars. Some of the cars were “ordinary”, and some less so. For example, one had electric sliding doors, and starting the car was non-trivial. The expectations of the old ladies from cars they’ve driven in the past were not going to help them in some of these cars. The least “ordinary” car took about twice as long to start as a car that had everything in a familiar place.

At this point the presenters declared the most “ordinary” car as the most user friendly. At this point I had a catharsis I’ve long since lost, but I’ll try and re-gain it here. User friendliness is often taken as similarity with past experiences.

That is, something that is user friendly will present a familiar interface to the user. At first blush, this seems like a fair enough statement. However, it niggles me. One of the reasons is that if something new is invented, the only way to make it user friendly is to (somehow) make it familiar to the user. It would be akin to saying that a soft-phone should look like a real phone.

Clearly, this makes for an extremely inefficient interface. Even user friendliness nuts think it’s stupid to make software look like something in real life. When making an example of bad UI design a fair chunk is spent on trying to mimic some aspect of reality — To make a desktop seem like a physical desktop, or a soft-phone seem like a real phone, or to make an mp3 player look like a cassette or CD player, yet they still dismiss an unfamiliar interface for being “too difficult to understand”.

I think can catch the contradiction if I try: When talking about ugly VoIP clients matt mentions Gizmo and Skype as exemplary examples. Want to know why? Because they look like IM clients! That is, they’re familiar, despite the fact that familiarity was the very thing that made all the other phones “ugly”.

In order to find an example of good UI design, we must go back through the history of IM clients. IM clients copied their UI from IRC clients. mIRC is a good example of a typical IRC client, and I’m pretty sure on windows you wouldn’t use any other IRC client.

If a usability expert were to talk about the usability of mIRC, I’m sure they wouldn’t find it very usable. It’s customisable, it’s configurable, it’s got thousands of options under obscure menu items. It’s not something that you picture “ordinary people” having on their desktops. However, I think everyone who used IM clients had used IRC before then, because IRC fed the same need as IM before IM was around.

Unfortunately, I have no perspective on how easy or hard it was to use IM for the first time, because I used IRC heavily before I installed ICQ. I had familiarity on my side. If I was an old lady, maybe IM would’ve been a confronting task.

I think familiarity, therefore, should not be used as the benchmark for user friendliness, and usability experts should think about whether their opinions on some device are coloured by past experience.

Incidentally, I’m going to argue that consistency, as distinct from familiarity, is also a bad metric for usability. Consistency is making your device look like all the others in the same “area”. Back in the day, Everything in your TV cabinet was black. This was good. It didn’t affect usability so much as to make everything seem a part of a consistent whole. Now a lot of devices are coloured silver, and this makes all the devices look different and unclean, and the extra colour adds clutter and confusion. While I like consistency, and it might mean that no one will use your device because it’s inconsistent with the rest, it doesn’t make a device more or less usable.

So let’s re-iterate: Familiarity is a bad metric for usability, because familiarity is different to different people, and conistency is a bad metric for usability because it’s purely stylistic, and does not affect interaction.

Note: I’m not saying that these are useless metrics, period. In fact, familiarity and consistency will probably determine if your audience will use your device at all, because they probably just won’t bother to use a device if it’s either inconsistent or unfamiliar, but just because a device is familiar and consistent doesn’t mean it’s usable.

Just to give an example: Office 2007 now has something like a ribbon with all the options on that instead of using menus. Presumably this solves some problems that were insurmountable with the old menu system (for example, huge menus that fill up the screen). However, this means everyone who learnt your software has to learn it again. The inconsistency and unfamiliarity is frustrating and will probably end up making everyone hate Office 2007, except for those who haven’t used any previous versions. It’s the same with learning to type DVORAK, or using a track-ball, or learning a new (non-programming) language. It may be frustrating, and it may not seem “user friendly” but really it’s just the familiarity, and has no bearing on usability.

So what is a good metric for measuring usability?

A good paradigm: Every good device puts you into a world which is internally consistent and internally familiar! For a simple and well-made device it will likely also be externally consistent and externally familiar. i.e: The paradigm will be one found in other similar devices, but that has no bearing on the usability, which only depends on the internal consistency and familiarity. That is, the device will behave similarly for similar actions. It must have lots of repeating patterns and ways of interacting. It must build habits which are useful in the operation of the device.

In conclusion, you shouldn’t work very hard to be like everyone else (from a device perspective). People will probably find you frustrating if you’re inconsistent with everything else, but those who persevere won’t look back.