Interview: Doug Edwards, author of <em>I'm Feeling Lucky: The Confessions of Google Employee Number 59</em>...
...Page's idea to serve unscreened, customer-created ads next to search results, fearing it would unleash offensive content that would damage Google's brand, only to see the AdWords programme become the main source of Google's billion-dollar advertising revenues.
Another Google success that Edwards was set against were the Google Doodles, the sketches that adorn the Google logo to mark special occasions, as he was anxious about their effect on brand identity.
"I learned to be receptive to alternatives, and not just to assume that conventional wisdom is the best practice because that's the way people have always done it," he said.
"Too many organisations feel like they're being innovative when they make an incremental change. What Google taught me is, why make an incremental change when you can change the entire industry?"
Keeping this iconoclastic approach from descending into anarchy is Google's love affair with data. Edwards said every idea needs to be backed up with hard data to show why it is the best option and why it has been a success.
"There was always the sense that the most important thing to do is to get real, not theoretical, feedback and to not get lost in the debate about product features," Edwards said.
"It was about trying something, and what went hand in glove with that was that if you tried something and it was the wrong thing then there was a willingness to kill it."
If the data looks bad then no idea is too big to fail, Edwards said - whether it is Page's idea for an affiliate scheme that failed to drive traffic in the late 1990s or the now defunct collaboration platform Google Wave.
"I found it quite refreshing after having been at a company where you took years to plan and launch something and, regardless of the feedback you got in the market, you just kept it out there because you spent so much time building it," he said.
Computer says no
Of course, the flip side of reducing every decision to an algorithm, where X has to be greater than Y for Z to be a success, is that not every factor can easily be measured and assigned a value.
As someone whose job involved manipulating sentiment, not code, Edwards argued that data shouldn't always win the day.
"This was at the heart of some of the issues I had with the company, because I was always a believer that...