A tech autopsy of Google's failed communication platform.

In a short blog post on Aug. 4, Google announced that it was pulling the plug on Wave, the "real-time communication and collaboration" Web site that the company launched last year. This wasn't a surprise. "Wave has not seen the user adoption we would have liked," Google explained diplomatically—something akin to describing the Hindenburg's last voyage as being a tad bumpy. In reality, Wave had been deserted for months. Although it won a standing ovation when the company first unveiled the program last year, many early users reacted as I did when Wave went live—I was instantly turned off by its complexity and by the way it confounded online social conventions. (Especially annoying was its always-on "live typing," which revealed your textual fumbles to the world. Thanks, Google!)

The biggest problem with Wave, though, was that nobody seemed to know why it existed. What kind of users would benefit from Wave? Which online tool was it designed to replace? After poking into Wave semi-frequently during its early months, I never found a compelling reason to go back. I wasn't alone.

Google has downplayed Wave's demise, suggesting that such mistakes are a necessary part of the company's experimental culture. "Remember, we celebrate our failures," CEO Eric Schmidt told reporters earlier this month. "This is a company where it's absolutely OK to try something that's very hard, have it not be successful, and take the learning from that and apply it to something different."

I agree with Schmidt that Google's penchant for experimentation is laudable and one of the reasons for its longstanding success. But what's the point of releasing an experiment into the wild when you've given it no chance to succeed? A Google rep, who told me I could characterize our conversation but not quote it, indicated that the company always thought of Wave as more of a concept than a product. Google is bullish about the Web—since it makes most of its money from the Web, it has a corporate interest in making sure that the Web browser remains a place for technical innovation. Wave, which was one of the most technically sophisticated Web applications ever built, was meant to show off what's feasible in a browser—in particular, that a Web app could be every bit as responsive and functional as something running on your desktop or mobile phone.

But most companies don't release their concepts to the public. After showing off a futuristic vehicle at an auto show, a carmaker will pare it down for mass consumption. That process includes deciding on a particular market segment for the product—if you're releasing a new SUV, you might position it as an entry-level model aimed at young families or a luxury model for wealthy retirees, but not both. After that, you'd figure out how to appeal to that market segment—for example, you might run ads showing young families why your SUV would be perfect for them.

Google seems to have skipped this process—essentially, business school 101—with the release of Wave. The company spent thousands of man-hours building the site—a team of five (and eventually more) employees based in the company's Sydney office has been working on Wave since 2007. To be sure, the team displayed a great deal of tech wizardry. The company, though, seemed completely unsure about the end goal of that wizardry, as well as how to describe and sell it to users. As far as I can tell, none of Wave's support documents answer the most basic question about Wave: What's it for? In some ways, then, Wave was a quintessential Google product and a quintessential Google failure: A company teeming with money and programmers spent a long time engineering a solution to a problem it couldn't even describe.

This is not to say that Wave was awful. It wasn't. Many of its concepts proved useful to a select kind of user base: people who work in small teams that need to collaborate on documents, especially if that collaboration crosses large geographical boundaries. Among other people, this might include writers, Web developers, and probably a lot of engineers who work at Google.