So we clearly haven’t yet mastered the science of user interface here in the 21st century. But what is it we’re striving towards? What’s the perfect user interface? In, say, a thousand years, when we have unlimited computing power and unlimited energy (like the characters of my novels Infoquake and MultiReal), what kinds of user interface will we be using?

Let’s take the question one necessary step further: do we really need user interface at all? Or are we evolving toward the point where intelligent tools automatically understand what we’re trying to do? In a thousand years, will the concept of giving commands be obsolete?

Software developers are taking the first tentative steps in that direction now. Apple’s Steve Jobs has always taken that “benevolent dictator” approach: we’ll decide what you, the user, need to handle, and the machine will just automatically handle the rest. Take disk defragmentation, a software task that only the wonkiest of technowonks has any interest in controlling. There isn’t any standard disk defragmenter for Macs, but that’s not because Mac hard disks never need defragmenting. OS X simply does it for you behind the scenes, as this article on the Apple website makes clear.

Microsoft is moving in this direction too. One of the advantages that Windows users have historically held over Mac users is the fact that it’s generally easier to get under the hood and tweak the gears that make the system work. But that’s going away. Not only because OS X has brought command-line tweaking to the Mac, but because Vista is taking away a lot of tweakability from Windows. Disk defragmentation under Vista is a simple on-off proposition; flip it on, and the OS will handle it as needed. Likewise, throughout the operating system, interfaces that were once cluttered with hierarchical menus and interactive dialog boxes are giving way to much smaller lists of context-sensitive tasks. (For more of my thoughts on this, see old blog posts Don’t Worry, Vista Will Handle It and Look Ma… No Program Menus!)

It’s the same long-term trajectory of user interface we’ve seen in automobiles. Look at the user interface for the Model T (pictured, below; original photo, with explanations and more detail, here). Most modern automobiles have reduced this to a standard set of four controls — the gas, the brake, the steering wheel, and the gear shift. It’s not that the car doesn’t still need all those functions, but now the car handles everything itself. It’s not exposed to the end user. If you believe the so-called experts, we’ll all be zipping around in self-driving robot cars within a generation.

Follow this trend several hundred years, and where does it lead? I talked previously about elevators that automatically know which floor you’re going to via RFID chips in your apartment keys. Why couldn’t that work elsewhere? Maybe you’ll pull into the Starbucks parking lot and find your usual soy milk decaf latte waiting when you get up to the counter. Maybe the refrigerator will automatically order more eggs from the store when you take the last two out. Maybe the polling station will know that you’re a member of the Christian Coalition and have a ballot all queued up with Mike Huckabee’s name checked when you get up to the voting booth.

There’s something very unsettling about these scenarios, and it’s not just the potential privacy hazards. Humans want to be in control of our environment; we instinctively resist environments that control us. Not only that, but we quickly grow bored with environments that coddle us. Humans are designed for dynamism, dissatisfaction, and change; despite the stereotype of modern man as couch potato, as a species we don’t handle stasis well.

So we like to be in control of our surroundings. But how much of this control is just feel-good illusion? When you order a hamburger at Burger King, sure, they’ll make it your way — as long as “your way” only involves their nine predefined toppings. And when you ask for lettuce, you can’t control how much, or whether they use shredded iceberg or delicately layered romaine, or whether it comes from West Virginia or Peru or Ecuador. Burger King’s real slogan should be “Have It Your Way, As Long As Your Way Falls Within the Narrow Parameters of Our Way.”

You don’t have much control over Google search results either. Try searching for “Bob Dylan.” You can choose to click on any one of the 25 million results pages you want — but Google determines the order in which they appear, which is tantamount to choosing your search results. (Try selecting the 4,523rd result sometime.) You can select “Advanced Search” and filter those 25 million results a number of ways, but you can’t choose the algorithm that Google uses to determine search results. Nor would you want to, because you’re not a computer scientist specializing in advanced information processing. If Google allowed you complete and utter granular control over every aspect of your search query, you’d either go insane or you’d never get anything done.

So is the Burger King experience a premonition of our future? Do we need to just trust the benevolent dictatorships of Google, Microsoft, and Apple (not to mention Burger King)? Is the future of user interface just a big pie of machine control with a thin crust of user choice on top?

As frightening as that scenario is, the opposite extreme is equally worrisome. It’s the future of total individual control. And boy, would that future suck.

We’re all familiar with the totalitarian sci-fi future where Big Brother boxes you in to a world of limited choices. 1984, Brave New World, Logan’s Run, etc. But what about the world of ultimate choice, where you have to control everything? The world has gotten smaller, our capabilities have grown larger, and the number of choices we have to make is bewildering. Once upon a time, you could choose to be a blacksmith, a farmer, or a priest. Now your career choices expand into the hundreds of thousands. Your parents went to the store and bought apples. Just apples. We go to the store and have to choose between Granny Smith, Macintosh, Fuji, Braeburn, Pink Lady, Red Delicious, Gala, Pippin, and Rome Beauty.

Big deal, you think. So I have to choose between a dozen brands of apples. How’s that a bad thing? It’s not. But what happens in thirty years when you’re expected to specify the size, tartness, color, firmness, ripeness, and pesticide of every piece of fruit you buy? What happens in 150 years when you can bioengineer your own hybrid apple/pear/mangoes right in the store while you wait?

If you want to see the beginnings of the future of total individual control, look at Linux. The Wikipedia list of Linux distributions catalogs around 200 different flavors of Linux. DistroWatch has much more. And these are just the prepackaged bundles of Linux. The hood’s wide open and the tools are sitting right there on the dash, giving you complete and total freedom to replace anything you like.

But who can deal with that kind of freedom? Unless you’re the kind of guy who likes to write display drivers in your spare time, you probably don’t have the time, the resources, or the expertise to make informed decisions about all of that. Perhaps one day we’ll all have neural implants to help us cope with all that cognitive processing. But until then, even the Linux geeks rely on consortiums of developers to make those decisions for them.

My point is not to bash Linux or to get into the whole open-source-versus-proprietary discussion — please, God, I don’t want to get into that right now. Rather, I’m pointing out that whether you use a MacBook Pro, a Dell Inspiron with Windows Vista, or a custom box with Kubuntu Linux, you end up relinquishing control. There’s only so much time you want to spend fine-tuning your computer, so instead of letting Microsoft make your decisions for you, you let a worldwide network of open source developers make them. We can argue about whether that makes a better operating system some other time; the point is that the practical effect of too much control on user interface is… giving up control.

So we’ve got the two extremes of User Interface Hell: the world of the benevolent dictator, where your control over your environment is deceptively limited; and the world of ultimate freedom, where you’ve got so much control that your ability to get anything accomplish is equally limited. Both of those extremes are equally unlivable; and you’ll notice that what those futures share in common is a lack of common-sense user interface.

Obviously we need happy mediums. We need to reconcile these two extremes, and simply, reductive user interface is the key.

The machinery that runs your information technology grows more intricate by the day, as does the machinery that powers your car. (Hamburgers, thankfully, seem to have reached an evolutionary plateau.) Despite what some Slashdot readers may fervently wish, I don’t think we’re headed for a future where everyone tweaks their own Linux distribution. If the freedoms we gain from our technology is the time and luxury of tweaking our technology, then we’ve gained nothing.

What often gets overlooked is that user interface isn’t a technological issue; it’s a sociological issue. Bad user interface limits freedom, it limits capability, it disempowers minorities. Think of how much difficulty your grandma has using the ATM. Technology has become too integrated into our society for us to leave people behind through insufficient user interface.

So what form will these perfect user interfaces take? To be continued…

In my first ramble about user interface, I used the toaster as an example of something that is erroneously thought to have a perfect user interface. Perhaps a more apropos example for most techies is the Internet search engine.

Think of any piece of information you’d like to know. Who was the king of France in 1425? What’s the address and occupation of your best friend from junior high school? How many barrels of oil does Venezuela produce every day? Chances are, that piece of information is sitting on one of the trillions of web pages cached in Google’s databases, and it’s accessible from your web browser right this instant.

You just have to figure out how to get to it — and Google’s job is to bring it to you in as few steps as possible. It’s all a question of interface, and that’s why user interface has been Google’s main preoccupation since day one.

It might seem the model of simplicity to click in a box, type for a search term, and click a button to get your results. But the Google model of searching is still an imperfect process at best. You may not realize it, but there are still a number of Rubegoldbergian obstacles between you and the information you’re trying to get to. For instance:

1. You need to have an actual machine that can access the Internet, whether it’s a computer or a cell phone or a DVR.
2. That machine has to be powered and correctly configured, and it relies on hundreds of other machines — routers, satellites, firewalls, network hubs — to be powered and correctly configured too.
3. You need to know how to log in to one of these machines, fire up a piece of software like a web browser, and find the Google website.
4. The object of your search has to be easily expressed in words. You can’t put an image or a color or a bar of music into the search box.
5. Those words have to be in a language that Google currently recognizes and catalogs (and your machine has to be capable of rendering words in that language).
6. You have to know how to spell those words with some degree of accuracy — which isn’t a problem when searching for “the king of France in 1425,” but can be a real problem if you’re looking for “Kweisi Mfume’s curriculum vitae.”
7. You need to be able to type at a reasonable speed, which puts you at a disadvantage if you’re one-handed or using imperfect dictation software.
8. Google has to be able to interpret what category of subject you’re looking for, in order to discern whether you’re trying to find apples, Apple computers, Apple Records, or Fiona Apple.

Some of these barriers between you and your information might seem laughable. But it all seems so easy for you because you’re probably reading this from the ideal environment for Google, i.e. sitting indoors at a desk staring at a computer that you’ve already spent hours and hundreds, if not thousands, of dollars to set up. If you’re running down the street trying to figure out which bus route to take, the barriers to using Google become much steeper. Or if you’re driving in your car, or if you’re a Chinese peasant without access to 3G wireless, or if you’re lounging in the pool, and so on.

Even in the best-case scenario, after you jump through all those hoops, you usually have to scan through at least a page of results from the Google search engine to find the one that contains the information you’re looking for. Google does no interpretation, summarization, or analysis on the data it throws back to you. Some search engines do some preliminary classification of results, or they try to anyway, but it’s generally quite rudimentary. Chances are you’ll need to spend at least a few seconds to a few minutes combing through pages to find one that’s suitable, and then you’ll need to search through that suitable page to find the information you want.

I don’t mean to minimize the achievement of the Google search engine. The fact that I can determine within minutes that a) the king of France in 1425 was Charles VII, b) my best friend from junior high school is currently heading the division of a high-definition audio company in Latin America, and c) in 2004, Venezuela produced 2.4 million barrels of oil a day — this is all pretty frickin’ amazing. But that doesn’t mean we shouldn’t note the search engine’s shortcomings. That doesn’t mean we shouldn’t point out that there are still a zillion ways to improve it. There’s still a huge mountain to climb before we can call Google an example of perfect user interface.

But don’t worry, because Google’s on the case.

Google has been making a mighty effort to break out of the web browser for quite some time. Not only have they been pushing their browserless Google Desktop app for some time, but they’re also quite open in publishing their APIs and trying to get you to hook into Google from other places. Cell phones, iPhones, car dashboards, public kiosks, refrigerators, digital chopsticks, Bluetooth-enabled dog collars, etc.

Why? A few years ago, we might have said that they were trying to escape the monopolistic grip of Microsoft and its Internet Explorer browser. But now that Firefox has made serious inroads on IE’s dominance — they’ve got around 16% global market share, 20% North American market share, and 30% European market share, if you believe the latest statistics — it’s not such a big concern.

No, the main reason Google’s looking for new avenues for search is that the standard WIMP (Window, Icon, Menu, Pointing Device) user interface is a dinosaur, and right now it’s late the Cretaceous Period and there’s a big fucking meteor zipping across the sky.

Forget about the distinctions between Mac, Windows, and Linux — they’re all inefficient. While some computer operating systems may work more smoothly than others, they’re all based on the principles developed by Stanford researchers and Xerox PARC engineers in the late ’60s and early ’70s.

What’s wrong with the WIMP interface? It’s a nice all-purpose interface for general tasks, but it falls down on the job on just about any specific task you give it. As software has grown more complicated, the WIMP interface has failed to keep up. Programs like Microsoft Word have become mazes of hierarchical menus and drill-down dialog boxes, and operating such programs efficiently has become an exercise in rote memorization. Shoehorning the computing power of a 2.4 GHz dual-core processor into seven or eight subcategories and a row of increasingly tiny icons is kind of like running an M1 Abrams tank off an Atari 2600 joystick. You’re wasting potential.

Software manufacturers are now toying with a host of WIMP extensions and alternatives like the Office Ribbon, which try to unearth options that had been buried four menus deep for years. And while the Office Ribbon is pretty nice, it’s ultimately limited. You’re still dividing up a list of possible tasks into seven or eight subcategories, and expecting users to drill down to find the item they’re looking for. The Ribbon works fine for Office 2007, but it’s certainly not going to cut the mustard in Office 2020 (if such a thing even exists then).

But let’s take things one step further. Forget the WIMP interface — the computer itself is just an intermediate step, headed soon for the great Recycle Bin in the Sky.

As the MacBook Air has demonstrated, the physical machine itself is disappearing. People have been talking about the concept of “wearable computing,” and experimenting with gadgets like the Senseboard, which allows you to project a virtual keyboard and type on any surface you like. Computer manufacturers are looking at the mouse and realizing, heck, you don’t need an intermediate plastic device that represents where you want to point on a computer screen. You can just touch the damn thing yourself and make it do what you want. Thus the creation of Microsoft Surface and devices like the iPod Touch.

The point I’m trying to make with all this is that we’re still in the Dark Ages in terms of user interface. You may feel pretty content with your little plastic box showing little two-dimensional pictures on a little 17-inch screen. But it’s just an interface, and a ridiculously inefficient one at that, and it’s going away. Soon.

So if computers are going away, where do we go from here? Do we still need user interface? Coming in the next article…

***

(Of course, let’s not forget that all this time I’ve just been talking about one very narrow application of user interface, and that’s interface as a gateway to information technology. But what about user interface in the real world? After all, your car’s got a user interface, your hedge clippers have a user interface, your TV has a user interface, and so does every elevator you’ve ever ridden.

(Take the standard elevator. Elevators are extremely dumb machines. They spend large amounts of time sitting on the wrong floor. When you walk up to the elevator, the only interface you’ve got is a simple two-button panel that asks whether you’re going up or down. People often end up piling into multiple elevators that are going to the same destinations, requiring all of the elevators to stop at multiple floors. The buttons for opening and closing the doors once you’re in there are a bad joke — by the time you find them, it’s either too late to stop the doors or just an unnecessary extra redundancy.

(How come the elevators don’t know where you’re going already? If you’re in a strange building, that’s understandable — but why should you have to push the same button for your apartment or office every day? Couldn’t the building automatically sense that someone’s waiting for the elevator via motion detectors? And couldn’t it automatically sense which floor you’re heading to by reading an RFID chip in your key? Hell, the elevator should start making decisions about which elevator to send and when as soon as I enter the parking garage.

(So just like computers, these real-world interfaces are rife with inadequacies too. They’re just waiting for a revolution in user interface.)

***

(Sources for the images in this article: “Google Is a Giant Robot” by Stuart Brown; screen cap of Google Desktop from the unofficial Google Operating System blog; and the original WIMP interface for the Xerox Alto, circa 1973, from the Encyclopedia Britannica.)