Josh On Design

Usability is the number one concern when designing software. And I don't mean "today with modern computing and the lack of new types of software usability is easier and growing in importance". I mean computers have no other purpose than to serve humans, and a better computer serves humans better. End of story. Software that doesn't conform to the way humans do things is bad software. (bad!) I don't care how cool and awesome your software is, if it's not understandable then it's useless.

A long time ago I thought that computers were hard to understand. This also made them cool. Then I worked as a lab assistant for a few semesters and saw plenty of smart students having incredible problems using computer software, often for tasks I deemed trivial. Suddenly I was hit by a revelation: computers are here to serve people. If a human can't use it, then the problem is in the software, not in the human. This kicked off a change in my mind and I switched my specialization to usability.

In the ten years since I started writing software professionally virtually every core problem I have found with software, be it commercial or open source, big or small, 3rd party or something I wrote myself; has been a usability issue. Not always a User Interface issue, but Usability nonetheless. I once wrote an entire language and toolkit for building wireless applications. I designed an elegant system that handled every task on my features list presented by management and my own ideas of how it should work, yet I never considered how a third party developer would use it. Once I got real users I ended up reimplementing a entire programming language to fit their requirements instead of doing the obvious: embed and existing and debugged language inside of my toolset. I wasn't thinking about the usability: the big picture.

This is my request of all software developers out there, especially the open source ones. I implore you: think of your users first, not how you would want to do it. Your UI will be subconciously molded to fit your own algorithm if you don't think of your users first. We Java developers should be familiar with concept since we often separate interface from implementation. When you design that new widget, first think of what the user wants to do with it, not how you want to implement it. Implementation is secondary. (at best!)

Here are some easy things you can do to make your software more usable.

• Walk away. Seriously. Work on something else for a day. Then come back and run it. Did you have problems getting it to start because you forgot the exact incantation required? Did you click the wrong button because it wasn't where you subconciously expected it to be? Did it look ugly? Did it take longer to run than you remembered? You have to get some distance from your program to accurately evaluate it.
• Draw it. Literally draw the interface on a piece of paper. Make some notes. Does it look good? Does it look too big or small? Could you imagine doing everything in the program from what you see on this paper? What's missing? If you can't draw the interface then you certainly can't code it.
• Get a second opinion. You wrote the software so you are too close it to it be objective. Get someone else to look at it. Show them a screen and ask them to tell you what it does. Ask them to explain how they would do something. If they can't then you need to redesign the screen. They say that programmers can't be interface designers. That's not true. A single person can program and design interfaces, just not at the same time. You need distance.
• Practice uninstalling. If you can uninstall your software easily, then installation must already be a snap. There are plenty of tools out there to make managing your software easy. JNLP. EXE builders. Prefab installers. Pick one. Don't use a commandline with custom options.
• One paragraph description. Describe your software in one paragraph. If you can't then it's doing too many things or you don't understand the problem well. Cut it back.
• Remove features. We often judge a program to be the most elegant if it can do as much as possible in the fewest lines of code. Instead I ask you do judge if the user can do the most possible work with the fewest features. Think of it as the Zen of UI Design.
• Don't reinvent the wheel. Your job is to get a program to your users that lets them do as much work as possible as quickly as possible. Shipping it soon is even better. This quickest way to do this is to not reimplement existing code if you can possibly avoid it. Sometimes you find an open source library that sort of does what you want but not quite. Don't reimplement it. Email the developers and ask for an extension point. You'd be surprised of often they will change the code just for you. And when you write your own reusable software, provide extension points. Then document them. If someone emails you for help, help them.
• Don't invent the wheel either. The quickest way to do this is to not invent new interfaces. I know that your sticky slider widget is cool, but would a simple drop-down box be better? Unless your widget is replacing at least three others you should leave it out.

I'd like to add that is applies not just to software but to the documentation and websites as well. SourceForge, for example, has to be one of the hardest to use websites ever. Each project should have a single paragraph that says: "This is what I do, this is my current status, and this is what to download". Please: think of the users.