A coworker posted on Firefox 3 and Mac addicts, and mentioned:
Other than background windows looking like foreground windows, all of his issues are either extreme power-user features (AppleScript support), things that are easily arguable (clicking on the URL bar selecting the URL), or things that Firefox could quickly steal from Safari and should make everyone happy (submenus with additional history items in the History menu.)
I started this as a comment there, but I wanted to expand it more. Now I know what he meant, but I don’t like the word “arguable” in that sentence. He suggested “subjective,” but I think that makes it worse.
Interfaces are thought out or not. Interfaces can be made simpler or more complicated, in one aspect or another. Often simplifying one aspect will make another more complicated. But interface design is neither arguable nor subjective. Instead, the effectiveness of an interface is something much simpler: measured or not.
In user interface design, the impact of an interface can be measured across your prospective user base. It may not be practical for you to do so1, but it can be done. That means it’s not subjective. Don’t write something off as “arguable” — if it matters, determine the truth.
This is not to say you can determine the perfect interface by measuring, but you can determine the superior of two interface designs.
If it’s something your users will feel strongly about and there’s significant disagreement, make it an option, but that should be reserved for cases where it’s really important to the user. But that’s tangental: User preference for interface varying for a particular user doesn’t mean interface design is subjective. And I think a good first pass is to default to least astonishment — which in today’s world, will mean consistency with the spirit of the platform you’re targeting.
- In fact, this is probably the rule rather than the exception. [↩]
April 22nd, 2008 at 9:55 pm
APPLY_FLAME_RETARDANT += 1
I think that ‘arguable’ and ’subjective’ (in that usage) are passive/aggressive, suggesting that the writer doesn’t just think it’s possible to argue it, but that the argument is either obvious or futile.
As for “one true way,” that’s arguably a straw man, which is to say that it’s both futile and obviously suspect. Which is itself a self-referential passive/aggressive argument, and a straw-man (proving nothing). Measuring the perfect interface, however, across a large sample set, includes the possibility that there may not be one single result: in fact the result may show many clusters of modes that fit various styles of use. Averaging such a measure would be an obvious statistical calculation error.
The short of it: there may be multiple best ways to solve any given problem (including user interface problems). Or, there may be one. Or, there may be zero. Arguing that only one possibility exists is arguably insane, without looking at the measurements of that particular problem, and responses to solutions.
There are elegant balances, of course, which distill one (or more) modes in a way that may make you smile. In the philosophical sense, this is itself the one true way, which is to find a pleasing balance to an underlying, messy reality. And not just a balance, but that one true balance that makes anyone with any sense feel warm and fuzzy inside. Warm and fuzzy, of course, is still fully subjective. You can measure that people think they feel it–but actually measuring objective reactions to complex balances is buggy and questionable.
Statistically balance is difficult to prove. You can measure the parts, but understanding the effect requires understanding the interactions and how they make you feel. So you have modes to a problem, balances of solutions to those modes, and perceptions of those balances. The state space gets large quickly, and while some of the parts are theoretically measurable, it’s not practical to do so, especially when the important measures are subjective.
April 22nd, 2008 at 11:05 pm
I’m not saying you can determine One True Interface through testing. I am, however, saying that you can determine the better of two interfaces through testing.
In this example, clicking an URL to select it all? Test it. Safari does this brilliant trick where clicking the URL bar selects it all, but clicking a letter in the URL puts the insertion point there. It’s actually a standard Mac OS X thing, not just Safari. Is there an advantage to doing it some other way? I really doubt it. If you can show an advantage to some other way, go for it. In the absence of that evidence, stick with the system standard.
April 23rd, 2008 at 6:49 am
There’s always a better way (and it’s this belief that fuels innovation) … but sticking with a standard is a sensible default.
April 23rd, 2008 at 8:42 am
Absolutely. Some of both the coolest interfaces and the nicest subtleties I’ve seen come from violating the standard. One of the simpler ones I’ve seen is the way Address Book on Mac OS X shows fields when editing contacts.
(On the other hand, I don’t think I’ll ever get used to the way Spotlight is invoked on Mac OS X.)