* Minimise the total number of clicks needed to access the sum of the program's functionality.
* Make frequently-used functionality easier to reach than rarely-used functionality.
* Don't have more elements in the UI than is easy to parse quickly.
* No surprises.
However, after reading this article, I get the feeling that it's a more arcane art, which not many people know. What are some simple rules about good UI design, in your opinion?
Your first three principles are about efficiency. That's a big part of usability. When you reduce clutter, you make it easier to use by clarifying the next action the user should take.
But "No surprises" is rather complex. It means that you not only have to explain what you are doing, but also work with the user's mental model, which might be totally wrong.
Here's an example from my current project. We are designing a web form to upload images and video to Wikipedia. In 2010, the users' model of media uploads comes from Facebook, ImageBucket, and maybe Flickr. Even after you make them click on a box that says they give up their own rights before proceeding, they don't understand what they did. Apparently people's mental model just continues on blithely until it smacks against a major contradiction.
So we've tried various metaphors. "License your work" was replaced by "Release rights", and the next iteration is going to be "Donate your work". That wording was the suggestion of a usability testing and analysis firm we hired, by the way. So it does take a certain amount of inspiration to come up with alternatives, but recognizing where the problem areas are is just a matter of measurement.
Inspiration is aided by just having a bigger bag of tricks, too -- knowing common problems and solutions. For instance, if there is so much to do on one page that users are finding it overwhelming, it is helpful to break it into multiple steps (a "wizard"). That's another approach we have taken.
Anyway, I'm not sure if my expertise with usability is so deep that I could pronounce just a few principles. Maybe "reduce cognitive load to the minimum" is one, which has been more memorably distilled by Steve Krug as "Don't Make Me Think". Since we want to avoid onerous information transfer and cognition, there are some models of communication that might be helpful here; a classic text on human communication is The Responsive Chord, by Tony Schwartz. He argued that most communication is about activating stuff you already know (at the right moment), which you can see at play in the example I gave.
But ultimately it's about motivating human behavior, so it's only about as complex as the human heart. ;)
Personally, I spend a great deal of time putting myself in the mind of the user, trying to figure out how we as humans visualize things and simplifying the actual underlying task at hand.
The list you summed up does work for 95% of all cases - especially if you're just thinking about a "website". However when you're trying to create something new or working with some new concepts you have to imagine how the user pictures things so that it matches his thought pattern as much as possible.
This also includes placing things were the user is most likely to look for them etc.
So your right with the number of clicks, but you also have to make sure that the user can complete the task (effectiveness), and that they are happy (satisfaction). From our own research the three; effectiveness, efficiency and satisfaction are very highly correlated together.
Usability is so much more about mental models than it is about efficiency, about helping the user feel confident & safe, than about a lack of surprises. Sensible hierarchy and visual relationships are a big thing that most people seem to forget.
Workflow, too, is critical. I've used a lot of software where every screen is mathematically, scientifically "usable" but the software overall was crummy because it duplicated a workflow from another piece of software which had got it wrong initially.
For example: 100% of email clients. Feed readers. Project management tools. They're all the same.
Surprises are actually very good in many cases. Even "ease of use" is something you don't actually always want.
My interfaces break all sorts of rules like "use standard widgets" and "conform to expectations" and "no surprises" -- but that surprise then results in an "AHA!", great enjoyment and power, and that's why people write love letters to our customer support account.
For example, in the information retrieval community, companies like Google continue to employ a lot of people to mark search results as relevant or not. Even with all this data, it's still hard for us to quantify what is "relevant" to a user. Likewise, it is hard for us to quantify what is "usable."
But even though the programmer may own this toolbox of quantitative methods and whatnot, most of them do not use it when it comes to usability. Why? Because the toolbox is quite empty. The advantage that these "snobby" designers (just like how programmers are "snobby" about their choice of programming languages and tools) have over programmers is that designers have more empathy towards the everyday man. So there is some emotion involved. Sure, there might be a science behind all of this, but since no one figured it out yet, we'll have to go use our intuition of what a good design is.
In the future I imagine there being plenty of UX consultants who, rather than having a strong education in the psychology of websites, instead is experienced with the tools necessary to evaluate a design and make recommendations. Then there will be graphic designers, marketers, etc who are actually qualified to work with usability.
Once you do start analyzing, what do you do with the results? You can try to make small tweaks -- it can be worthwhile if what you have is already good. If what you have is bad however, no amount of rigorous analysis is going to solve the artistic problem of coming up with a cohesive vision of something better.
The way I describe this to people is: "Usability science can only test a hypothesis. It doesn't give you a hypothesis to test."
Not to mention the problem of hill-climbing… you may think your test results are amazing, but you can't tell if you're really at a peak or in a valley with a lot of mist obscuring a potentially even higher peak..
UX designer = "mental space"
User Interface Designer = "white space"
User Experience Designer = "mental space"
Now how arbitrary is that? What is it about "Experience" that implies thinking and "Interface" that implies a lack thereof? What about all the designers unfortunate enough to come before the term "UX" was invented? I suppose Jef Raskin was only ever about white space.Let me suggest an improvement
Mediocre Designer = "white space"
Good Designer = "white space and mental space"
But the real problem is that most shops can't measure usability in a fine-grained way. And yet everyone does have some idea that it's really important. Consequently they are bamboozled by anyone who purports to have such expertise.
For the most part, so-called usability experts are really just graphic design snobs. They have a certain aesthetic they like, which is arguably "cleaner" and has more white space. So the page might be better organized but that's nowhere near what true usability is.
Real usability work is about trying to understand how the user is thinking. What their mental model is. What their goals are. In what context they are trying to achieve a task. Not just how their eye is scanning for information.
Anyway my point is, when it comes to real usability work, programmers are just as (un)qualified as graphic designers, marketers, or any other kind of "product" person. Arguably the designer is supposed to have a larger toolbox when it comes to possible graphic treatments, but the programmer also has a toolbox of quantitative methods, experimentation, and a deep understanding of how the system could work.