On Vint Cerf, Google, dark fiber and video
When we reached Vint Cerf in his MCI office several weeks ago to ask about a rumor that he was going to Google, he quickly rebuffed us, saying, "There are lots of rumors about me. Don't believe everything you hear.'' We realized when we hung up the phone that it was an artful dodge. And, lo and behold, MCI is scheduled to announce this morning that Cerf is indeed joining Google. (The Google release is here.)
Cerf, as old Net hands know, is one of the guys who in the 1970s developed TCP/IP, the underlying architecture for the Internet (he's often called the "father of the Internet").
It's almost sick how much high-profile talent Google continues to amass, often without even trying. Cerf, Apple's Andy Hertzfeld, Alta Vista founder Louis Monier, Adam Bosworth of XML fame, as well as Bill Coughran and a host of other Bell Labs vets, to name just a few.
Some might view a 62-year-old engineer whose most celebrated accomplishment is more than 30 years behind him as being past his prime for a young company like Google.
But Cerf, MCI's senior vice president of technology strategy, isn't close to hanging up his spurs. In our brief conversation with him yesterday, he seemed hungry - impatient even - to get his hands dirty again after 11 years of dealing with policy and regulatory issues for MCI. He's got ideas - some of them "half-baked,'' he said - that he wants to dig into and turn into reality.
Cerf sounded like the kind of person who's seasoned enough to see the sweeping arc of technological possibilities, but who probably still likes to roll up his sleeves and write code every once and a while.
"To hell with this father of the Internet stuff,'' said Dave Burstein, who writes the DSL Prime newsletter, of Cerf. "He's a damn good engineer. It's the challenge of a real hard interesting problem. That's what Google offers.''
Speaking of Burstein, we've been trading emails recently about Google and its ambitions. Burstein is convinced that Google's worldwide buying spree of dormant fiber optic cable networks will ultimately support "the world's largest video server network,'' what Burstein calls "the largest TV 'anti-network' in the world.''
"The dollars in video are large, and Google (with servers and worldwide fiber) will have strategic advantages,'' he says. "Spend some time at video.google.com and think what it can be.''
We'll quote more from one of his recent e-mails:
"[The] key idea is that Google intends to become the most important video carrier on the planet, and is developing the servers and fiber network to make that possible. As television shifts to the net, only Yahoo and perhaps British Telecom are in position to compete. The ABC's and NBC's of the world are outclassed.
"Google has told their folks to plan services 'as though the servers and delivery' cost next to nothing, although the real cost in 2005 remains in the $billions. They know that will come down with Moore's law, and they are pricing down the learning curve. Once they build the basic network, the marginal cost of doing more will be amazingly small.
"To make that concrete, a company like Movielink or Akimbo makes a business plan around a cost of about 10 cents an hour to serve and distribute their shows at full-screen quality. That means that even if the content is free, they have to generate significant revenues from every show. That's why 2 minute movie trailers, or video that shows in a three inch window, dominate the web. The infrastructure costs are manageable if you are selling movies at $2 or $4, but a tough obstacle if you want to be ad-supported or cheap in volume.
"Google is thinking ahead, building a network designed to bring that cost to a penny or two per hour by 2008. If they get video for free, they can afford to serve it just for the related ads. There is an unbelievable amount of video on the shelf not making money that can now be distributed through Google. Some will be free, other stuff as cheap as necessary to find a market."
More grist for the mill.
Links to blogs that reference this entry: