RDVP Seminar (10/6/2004): Bruce Lusignan

Bruce Lusignan is the director for Stanford's Communications Satellite Planning Center, and has, for 30 years, been lecturing on rural connectivity worldwide. He mentioned several times "Oh, he was a former grad student of mine..." in the context of key government officials planning telecom policy around the world. He spent most of his time talking about a recent project in Peru, covering the regulatory environment, and the technical specs of the way the system could work. A couple of the fellows, Yann and Greg in particular, seemed familiar with the details, though much of it was over my head.

Fortunately, Bruce returned to language I could understand, talking about the business opportunities that exist both in running the infrastructure as well as those that were enabled by it. As I understood the technical part: VSAT satellites (like the PanAmSat and IntelSat networks) provide cheap global communications, but should be used only when the link cannot be made by multiple hops (via wi-fi) between "Smart" base stations. A single base station (equipment costs ~ $30K) can handle ~200 cell phones, and this is a manageable level for an individual entrepreneur to set up a store providing telecom to a region. Even at 1 or 2 cents per minute, the calls are still 50% profit.

(Quick ROI calculation: need 3M minutes of calls to recover the $30K initial cost. 3M/200 phones = 15K minutes per phone, or 250 hours of continuous usage. Assuming 1 hour/day of usage, cost recovery in 8 months. Sanity check: in a village of 2,000 people, each person is making $30 of calls to cover $15 of expense. Assuming $2/day income, an 8 month recovery is 6% of income on telecom expenses, and that person made 1500 minutes of calls 25 hours in an 8 month period--that seems a bit high. Probably need more like 4,000 people, (1.5 hours per month).)

Bruce pointed out that a big percentage of the cost was the solar panels, and that in many cases, there is a local grid that could provide the power more cost effectively. The cost for just the communications equipment is closer to $8K... The current revenue sharing model provides only 8% of the total to the store merchant who does most of the maintenance; the local telecom affiliate gets 20% for providing the pre-paid cards (though it's often the merchant's job to pick them up....) Bruce noted that once this infrastructure is in place, there are additional services that can be easily added: FM radio, for example, requires only a 200 watt transmitter (about $200) and individual radios are about $1 each. There's a built-in local market for football coverage. Even TV broadcast becomes feasible with more TV sets having builting MP3 decoders.

This also led to a discussion of internet cafes, and the assertion that a community of 1-2,000 people could support one. The fellows all piped in with their experiences of hourly rates for cybercafes around the world, ranging from about 25 cents/hour (China) to $1.50 (Mali) to $8.00 (NYC and business centers in Chinese hotels). We talked about systematically collecting the info to produce a "CyberCafe Index" like The Economist's Big Mac index that also factors in local salaries.

Bruce said that there had been a sucessful market place for cut flowers between Peru and Europe, where the growers would deliver according to orders placed on the internet, with barcodes to track logistics.

There were some questions about Voice over IP (VoIP) as being the ultimate solution and "spoiler" for much of telecom's sunk cost infrastructure, and a means of getting around the monopoly / regulation. Also about the effectiveness of daisy-chained wi-fi conenctions for data transfer, since latencies can be long. Greg pointed out that for asynchronous applications (like store and forward email) this was not a problem (because it had been explicitly taken into account at design time). So, in many respects, this model is a return to the early days of the internet in the US when communication links were not redundant and reliable.