more on "Let them eat megabits"

[Dave Farber <dave () farber net>]

From: Rahul Tongia <tongia () cmu edu>

Dave,

For IP consideration.

A response to Eli Noam's and other discussions on ICT and development.

Rahul
********
I think Heather Hudson and Taylor Reynolds responded well to Eli Noam's 
column "Let them eat megabytes" (emphasizing leapfrogging and applications) 
but I think a few clarifications and points need to be made.

The entire definition of "broadband" is somewhat murky, though some 
governments and entities define it varyingly as connections between 128+ 
kbps to those over 640 kbps (if not more).  When Japan and Korea already 
have connections in the 10+ mbps range, my 640 kbps DSL sure doesn't seem 
that broad.  The attributes that really matter, especially for developing 
countries (er, emerging economies or nations in transition) are that the 
connection is always on, and that it is "all you can eat" (flat rate 
pricing).  The actual speeds depend on what applications you want, and that 
is a chicken-and-egg issue.  The other, equally important issue is the 
pricing (not just flat rate) -- it needs to be affordable.  The recent ITU 
report that included DAI (digital access indicators) does a great job of 
capturing this information, whereby dial-up costs multiple years' worth of 
salary in many countries (when including local phone charges).  Contrary to 
first impressions, that is not simply because of low earnings (smaller 
denominators) but also because of costly connectivity (because of a 
multitude of factors spanning poor design, poor volume/economies of scale, 
bad regulation, and simply pricing things for high profitability/low volume).

I was somewhat perplexed by Eli Noam's statement "Thus, the money for about 
three broadband upgrades could instead support one basic connection of a 
new user to a network."  This is not an apples-to-apples comparison, nor is 
this pure substitution.  For starters, the $250 upgrade to broadband 
(though this can now be done for less, in some cases) only comes after an 
initial investment in physical infrastructure like copper/fiber (even with 
wireless).  *More importantly*, the users who ask for broadband are not the 
same as those who seek but don't have basic dial connectivity.  Gone are 
the days when there were waiting lists of years for a phone connection (in 
many countries).  Cellular and competition (and new technology) changed 
that.  In Indian cities, one can get a new phone line within 1 day (!), 
compared to the week Verizon asks.  So why doesn't everyone have a phone 
line?  Lack of affordability (combined with user perceived demand).

One solution is new technology that can bring down costs significantly. 
Here, I think we should be willing to relax some technical standards, 
within reason.  When considering VoIP, people complain that it doesn't have 
the five 9s of reliability of copper voice (as if phone lines, let alone 
power systems in many developing countries, have three 9s of reliability). 
Also needed are new business models, such as community based (kiosk) 
access.  Reselling should be allowed, and unlicensed or freely licensed 
(for basic users).  India found enormous success with these telephony 
kiosks in the late 1980s, because the government (for a change) allowed 
entrepreneurs to set these up and profit share.  Similarly, cellular phones 
used as local phones are exceptionally appealing to many in developing 
countries, since the same phone number works on their job (often, the 
informal sector) and at home.

How much bandwidth do people need?  There are few developing countries 
aiming for megabits to every home.  But as Taylor Reynolds pointed out, 
building a new network just to do dial-up copper is a bad idea. 
Unfortunately, the poorest actually need the most bandwidth, to counteract 
a lack of literacy.  Many of us do text emailing quite comfortably, but 
others need voice messaging, if not pictures and graphics to help them use 
the system.  This is all the more true when considering interaction with 
business/government, and accessing public information, not just contacting 
friends and family.  In a new model, bandwidth might replace some end-user 
hardware, which has been a major stumbling block for users to buy in to 
using ICT.  For example, voice recognition, difficult to do on a cheap 
handheld or thin client, can be done at a local or even regional server, if 
there is reasonable connectivity.

Applications give us a hint as to why developing country users NEED 
bandwidth.  A colleague went to India a shortly while back.  His laptop 
required its weekly dose of Windows and anti-virus update.  Size, 8.3 
megabytes (a major update).  His dial-up, 28.8 kbps.  In practice, it took 
nearly 6 hours, and he was lucky it didn't disconnect on him.  His cost for 
that update, about $6, or almost a week's average income in India.

Rahul Tongia
CMU

-------------------------------------
You are subscribed as interesting-people () lists elistx com
To manage your subscription, go to
 http://v2.listbox.com/member/?listname=ip

Archives at: http://www.interesting-people.org/archives/interesting-people/