Improving Software Usability?

kevin_conaway asks: "Software usability is one of the hardest things to get right. Writing good, usable software is the holy grail of software development, yet few developers give it more than an afterthought. As a professional developer, I delight in writing software for other developers but shy away from writing an interface that the end users will see. What resources/books are recommended for improving your Human Computer Interaction (HCI) / software usability skills?"

Comments

here are a couple:

[yagu]

Don't Make Me Think and The Design of Everyday Things... two of my very favorite books.

"Think" is more web centric, but has many tips and insights, and is an accessible read cover to cover.

"Design" is a bit more pompous, and I don't agree with all points, but I give it high marks for making you take a different look at things you'd always taken for granted (Microsoft asked me a question at my interview from this book, btw).

A few more thoughts: don't confuse usability with user responsibility. If a task if tediously complex, it's going to be difficult to design a thin elegant easy-to-use interface. For example, photoshop can be amazingly obtuse to use, but there's a reason. Overall I give photoshop a "5" (out of ten) for their ergonomics, but I give them a "10" for what their application can do. I consider it partially my responsibility to climb that learning curve to do real work in digital graphics.

On the other hand, the unusable applications out there are infinite. My favorite example is Windows Media Player. I still have to figure out what to do just to play a CD with WMP. (And what's with the disappearing window?)

(Here's an interesting non-software example of horrible design: my parents have an RCA TV, not that old, but not HD. It has Videos 1, 2, 3 input, Cable/Air input, and VCR. There's a "SETUP" button on the front panel that lets you change the signal input from Cable/Air to VCR (or something like that), but the only way you can get Video 1, 2, or 3 is by tuning the TV channel to 91, 92, or 93 respectively. Until I found the manual and got to page 60 I was convinced the TV was broken.)

My favorite example of transcendental usability: Google.

(Some runners up: Picasa; Amazon.com (one-click), wish list, etc.)

(Also, I am opposite as to who I like to write for: I cringe when writing for other professional software developers, they're some of the biggest whiners about "what should be". I do however delight in writing software for clients. If you do it right, it's a genuine high.)

Interface design != Software design

[teh moges]

As much as everyone here loves to create their own programs and websites, for professional jobs, it must be known that those that create the software should NOT be responsible for designing the interface. Its a challenging field. While almost everybody here can create a good design without thinking, creating a great design is alot harder. Its the same with everything. Using certain software, ANYONE can create a good website. It takes skill to create the great ones though. Using certain software, the company I work for has their interns creating press releases. They work, but they aren't great. Anyone can design a logo, but theres a reason the big companies hire design artists. The very same is true in interface design. If you are worried about it and your budget can afford it (it should be budgeted for anyway), hire an interface designer.

Users are the best resource

[miyako]

The best resource for making sure your software is usable is to watch people use it. While large companies can afford professional UI designers and formal usability studies, even a humble F/OSS developer can do some simple UI testing.
When I'm working on software that is intended for users who are not developers or otherwise computing professionals, I usually try to get a regular user to sit down with my software for a half-hour or so and I watch them use the software. Generally, I just say something along the lines of "hey, wanna do me a favor? play around with this program for a bit and tell me what you think". Then watch over their shoulder. Generally this is a good way to get a list of what sorts of things are poorly placed "how do I...?", things that are confusing "what is this?...", features that users will like "can I ...?" and it's a good way to start finding bugs that only a user will discover.
A few tips that I've found doing this include
If any option is unavailable then it should be obvious WHY it's unavailable.
No matter how obvious your icons are, they should ALWAYS have text with them.
Avoid dialog boxes as much as possible
If you make your program look too much like another program, then you better make sure it looks and works exactly like that program. In other words, either stick completely with the standard way of doing things, or do it completely different. If you take some common UI element and tweek it, then you'll just confuse users. Menu bars tend to be the most common violators of this.
Understand color. A lot of applications throw colors around willy-nilly, if you are going to use color then study up on color theory and learn what colors go together, what colors are calming, etc.

Usability Studies are a must

[jchenx]

I think too many companies focus just on heuristic evaluation. That's basically paying a UI expert to tell you what to do and what not to do. A lot of companies won't even hire a usability expert, instead relying on their own engineers to "read a lot of books" and try to wing it.

This is bad.

Just like how software engineers should not be trusted to test their own code, they should also not be trusted to do "good usability". I'm saying this as a software engineer, who also has a Masters in usability engineering and has been in the field for a few years. Too often I'm surrounded by fellow engineers who think they know what's best for the user. Also, they'll claim that a certain design is best because it also makes for a "clean UI" and "clean code design". Then we sit users in front of the application, and all hell breaks loose.

Don't do this. Spend the money to hire a good usability expert, and have THEM perform proper usability studies. Good usability is NOT necessarily about a "clean UI" or "clean code". It's about a product that people know how to use. After this is established, it is then up to the engineers to make sure the actual implementation itself is clean, extensible, un-cluttered, etc. Not the other way around.

A few simple guidelines

[c0d3h4x0r]

Just by following a few simple common-sense guidelines, you can drastically improve the usability of any given software:

• Simplicity: keep things as simple as possible. Fewer options/settings/etc means less to have to figure out. If you must provide an option for something, supply a reasonable default. The user should never have to configure a bazillion options before being able to just use the program.
  
  
• Sane hierarchical organization: The human mind can only processes and deal with the external world by grouping, categorizing, and thinking of a collection of items as a single "chunk". So sanely organize options in hierarchical menus; sanely organize navigation into a tree-like path; etc.
  
  
• Direct manipulation: users always expect to be able to direclty manipulate an on-screen object by clicking directly on it, dragging it, etc, so design your UI that way. (Example that violates this: a listbox with items in it, with buttons underneath that must be clicked to act on the selected item, rather than allowing the user to right-click on the item itself to get a pop-up menu.)
  
  
• Data transparency: there should always be a way in the UI for the user to clearly see the information they want organized in the way they want it, and it should never be a mystery to the user where some calculated field came from or how it was calculated.
  
  
• Terminology: keep technical lingo out of the program's UI as much as possible, and make all text and phrasing clear to non-technical users. Whenever an error occurs, present a dialog that clearly explains the nature of the error and which also suggests a course of action that might rectify the error.
  
  
• Real-time on-screen feedback: when a user performs an action, they need immediate feedback to know whether the action succeeded or failed. Real-time responsiveness is super important. If a user-initiated action will take longer than about 0.1 seconds to achieve the ultimate result, then you need to put up a suitable progress indicator that updates itself responsively as the operation proceeds. If the user clicks on something and they don't realize the system is just busy processing their request, they are likely to click on it again and again without realizing the first time succeeded. (Example that violates this: you launch an app from the Windows XP start menu... you don't hear the hard drive churning or see a hard drive light because you are remoted in via Remote Desktop... no visual on-screen cue is given that the app is actually loading up... so you try to launch it again... in the end you get 2 or 3 instances of the app).
  
  
• Don't assume user expertise: always assume your user knows NOTHING about computers.
  
  
• Scenario-based design: don't merely dump a bundle of functionality on the user; give them a program that guides them through all the steps needed to solve their scenario. It's the difference between handing someone a graphing calculator and handing them a math expert.
  
  
• Users won't read, and shouldn't have to: users don't read text -- it's a proven fact. Nor should they generally have to. For most people, reading is an unpleasant expenditure of valuable energy and time that could instead be used getting something done. If you need more than one brief sentence in the UI to explain something, then your UI is too unusable and you're just leaning on text (that won't get read) as a crutch. Users should be able to jump in and start being productive with your software without having to read a manual, README file, or any other long-winded explanation whatsoever.
  
  

Re:A few simple guidelines

[isj]

Don't assume use expertise

I am one of the few people that have read the CUA guidelines, and they make at lot of sense, although many of the specific details are now outdated. The CUA guidelines say that you have to first decie if the UI is going to be a standard UI where it has the conform to system defaults and in general user's expectations; or if the UI is a so-called walk-up-and-use UI (such as ATM interfaces). The difference is that the standard UIs have to conform to standards but can contain many features, while the walk-up-and-use UI has to be simplistic and require absolutely no learning, but can break any standard as long as it makes it simpler to use.

So the guideline should be:

• Know your users: Don't assume your users have no expertise, but neither assume that they have. Find out. This impacts not only the program UI but also the documentation. If you do not know your users (or the intended target group) then the program is always too simple and to complex at the same time.
  

Re:A few simple guidelines

[Haeleth]

Just by following a few simple common-sense guidelines, you can drastically improve the usability of any given software.

You appear to be claiming that the guidelines you quote are universal and apply to every conceivable bit of software. I would beg to differ.

Simplicity: keep things as simple as possible. Fewer options/settings/etc means less to have to figure out.

Photoshop would not be a popular program if it only provided the five most common graphical operations and only let you configure three aspects of each. It's successful precisely because it's insanely configurable.

In other words, while there is indeed a place for simplistic software (witness the popularity of basic photo editing programs that literally only offer crop, size, and red-eye removal), it is by no means desirable for all software to be simplistic. Complex tasks require complex interfaces. Simplistic interfaces limit users to the options you have chosen, and I find it hard to believe that you can imagine all the possible ways a user will want to use your program.

Don't assume user expertise: always assume your user knows NOTHING about computers.

You cannot possibly believe this applies to all software. Are you seriously saying that the writers of a kernel debugger should assume their users know nothing about computers?

Scenario-based design: don't merely dump a bundle of functionality on the user; give them a program that guides them through all the steps needed to solve their scenario. It's the difference between handing someone a graphing calculator and handing them a math expert.

If a mathematician asks for a graphing calculator, you think they should instead be handed another mathematician?

Yes, simple tasks should have simple interfaces. Anyone should be able to answer emails, browse the web, write letters, keep an address book and diary, and print photos from their digital camera.

But that does not mean that simplicity should always be prized over functionality. Some things in life are tough. I've never driven an 18-wheeler: I would not expect to be able to sit down in one and take it across the continent. I've never filed a lawsuit: I would not expect to be able to waltz into a court and win a billion-dollar case. Why should I expect computers to be any different? If users won't read manuals, that is the users' problem.