The Uncanny Peak of Wrong Metaphors

Last week, Tim O'Reilly wrote about a Bill Higgins post called "The Uncanny Valley of User Interface Design."

The metaphor Higgins brings to the table to help explicate his point is that of "The Uncanny Valley," which Higgins explains well. For those unfamiliar with the concept, the basic idea is that we feel more and more empathy towards robots (or non-human characters, such as animated cartoons or video game avatars), as they begin to express more human features. At some point, though -- the edge of the Uncanny Valley -- they become too lifelike for comfort, but not lifelike enough to really fool us. Something is missing. Their appearance strikes us as just... wrong. Odd. Uncanny.

Higgins goes on to cite two films that are often used as examples of having fallen into this valley;  The Polar Express and Final Fantasy. I've seen both films, and I agree (especially in the case of The Polar Express... it just ain't right for some reason).

It's a well-thought-out article that has, as its main premise, the idea that the user interface for a piece of software should:

"...remain consistent with the environment in which our software runs. In more concrete terms: a Windows application should look and feel like a Windows application, a Mac application should look and feel like a Mac application, and a web application should look and feel like a web application."

So, the metaphor here, linking software UI to robotics/animation, is that we want something to not behave and/or appear to be too strangely like something else, while not actually being exactly like it. His main point is that Web aps that try to look like desktop aps, and vice versa, aren't really doing anybody any favors.

Though well written and supported with a couple good examples, I disagree with the premise, and I think the metaphor is badly chosen. 

Metaphor first. We all expect people to look like people because, for the most part, people look like people. There is an instant -- some biologists/anthropologists argue, innate -- connectivity at birth between newborns and human faces. We respond to smiles. We track eye movements. We can, without any training, mimic facial expressions. We are hard-wired, at least emotionally, and probably chemically/biologically, to know the human face better than almost any other object in the world. When you work with human faces -- as an artist, cartoonist, director, animator -- you are dealing with one of the most sensitive and universal inputs/outputs available.

What does a "link" look like, biologically speaking? Or a block-quote, menu bar, button, etc. Yes, there are conventions... which change. Yes, currently a Windows ap looks different than a Mac or Linux ap or a Web ap. And if you gave me a program that looked 90% like Microsoft Word -- one of the most succesful and widely used programs on the planet -- but then had a couple fundamental features that were radically different... I'd probably have some issues with it. At least from a learning perspective, possibly from an adoption one as well.

Unless those differences were really cool and easily grasped.

Time and time again, I've used new software that broke some existing UI rule or convention... and surprised me pleasantly. When done well. I will agree with Higgins that purposefully making an ap look like what it's not in order to glean some kind of borrowed brand shine is a bad idea. The plethora of "Web 2.0-y" glowing buttons, 3D tabs and various shiny bits, when applied to standard interface elements is, frankly, goofy.

Some might argue that much of this 2.0-look is an attempt to make Web aps look more like their desktop counterparts. In many cases, I expect that is the case. Design in everything follows trends. Designers often work for clients who don't understand the underlying design reasons, but can keep their eye on trends and popular elements and who want their little slice of the Web to look "cool like that link I sent you the other day. Make it cool like that." It doesn't matter that the site design being changed is one with a standard home page and six children, and the one being emulated is a robust, interactive, social application with all kinds of moving parts. "I want it to look cool like that" is, I think, as major an influencer of UI design as it is of fashion in any industry.

All that being said, however... I have real problems with the idea that UI needs to "look like" what it used to look like, or "should" look like at some level that, according to this metaphor, is emotionally equatable to that of the human face.

Higgins gives as an example, Zimbra, and compares it to Gmail. He says: 

"To me, Zimbra doesn’t in any way resemble my mental model of a web application; it resembles Microsoft Outlook... On the other hand Gmail, which is also an Ajax-based email application, almost exactly matches my mental model of how a web application should look and feel... I prefer [Gmail b]ecause over the past twelve years, my mind has developed a very specific model of how a web application should look and feel, and because Gmail aligns to this model, I can immediately use it and it feels natural to me."

Here's my problem with that. For me, "email" is the "human face" of this metaphor, rather than the development/deployment platform. Users (like myself) who have spent 12 years in corpoate environments with various versions of MS Outlook would find Zimbra much more "natural" than Gmail. I have a Gmail account, and don't use it much, partly because I find the UI to be so much different than MS Outlook, which is patched directly into my cerebral cortex for about four hours a day.

For Higgins, humanity is found in the platform. For me, it's found (in this case) in the tool interface. He thinks a Web ap should look like a Web ap. I think that email should look like email... and, for me, that means Outlook. I'm not saying I'm right -- but that different definitions of "natural" in a software design context need to be considered. In the end, what works is natural. It's not turtles all the way down for UI development; we can actually do things as businesspeople and managers to see if our users respond well to various changes.

We only have one model (though billions of examples, obviously) of the human face. It is uncanny, indeed, when a replication of it gets close... but not quite close enough. In software interface design, I'm not sure that there is a universal model. And, if there is one, that it would be at the platform level. MS Word, Photoshop, TrueSpace, World of Warcraft... I run all of these on Windows XP. If they are all examples of what should "feel natural" for a Windows ap, then our anthropomorphic uncanny valley would be filled not with proto-human forms... but with lions, tigers and bears.