GoogleNet conquers Mountain View One lamp post at a time (By Andrew Orlowski in San Francisco, Published Thursday 17th November 2005 01:55 GMT)
So, hey, I'll be interested to see how quickly Google gets it up (er... installs the net) and what the service is like once it's running. (the article suggests June, 2006 presumably)
What I find personally interesting in this article are the costs involved.
Google will rent 400 street posts ... for $12,600 a year ... 802.11 hubs at a density of around 20 to 30 per square mile.
My interest is that in the past I invested in Metricom, the developers of the Ricochet service. Ricochet wasn't and isn't well known, so let me give a little description. The service was 128kbps (or faster) Internet access that supported roaming at speeds up to 60-70 miles per hour. They deployed the service in 2000-2001, and it was an upgrade over a 28.8kbps service they'd operated since 1994 or thereabouts. To deploy the service they had to install boxes on street posts, they were powered by the lighting infrastructure (hence, one cost to the city is providing power), and were installed at a quarter-mile or so radius. It was a mesh network in that packets would travel across the poletop boxes until they found one connected to a wire, the wire went to a region-wide hub that then was connected to the Internet. Metricom also had to "rent" space on the street posts, which they called a "right of way" agreement.
What Google is planning will clearly have a similar design, just using WiFi boxes rather than the Metricom-proprietary network used for Ricochet.
Metricom failed for several reasons. One of them was the cost, network installation required a large infrastructure and Metricom spent in the neighborhood of $1 billion building their network across 10+ U.S. metripolitan areas. But Worldcom (MCI) also had their fingers in this game, with an investment and a high hurdle to meeting the terms of the investment.
Because WiFi involves a smaller radius per radio, blanket WiFi installations will mean more radios than the Ricochet design required. On the other hand the WiFi radios might be less expensive, considering that WiFi is available off-the-shelf while the Ricochet system was proprietary. Google won't be able to install the same commercial WiFi units we regular consumers use, because being out on street poles they will have to be rugged enough to handle weather extremes.
Hopefully as the Google boxes are installed the Ricochet boxes will come down. Every time I see one it reminds me of the failed investment that cost me so much money. I wonder if Google will also find this to be a failed investment for them. They are the current wunderkin corporation, but that doesn't mean automatic success for every line of business they enter. To be directly providing Internet access sure does change their corporate focus, and maybe they're taking a walk down the road that led to disasters at other previous companies ... such as the Excite-At-Home merger.
That is ... there's a neutrality one has in not owning any direct-to-user network infrastructure. That neutrality means you will tend to serve all the ISP's equally. But once the portal is tied to an ISP, the portal might tend to focus their service on that ISP and give less service to the other ISP's.
Google is such a good company. I hope they don't fall prey to that brand of portalitis.