Under estimating future trends - like bandwidth requirements or mobile adoption

James Enck questions many of the estimates of future consumer bandwidth requirements, for example

BSkyB has publicly stated a working assumption that end-user bandwidth demand may double every five years.

and others, leading to:

I am uncomfortably reminded of other previous industry predictions which proved wide of the mark - "no one will ever need or want their own computer," "SMS has no potential as a consumer service," "mobile penetration will peak at 20%," "the public internet will never support acceptable quality for voice," etc.

I agree with James and would go further to suggest that doubling in five years is completely ludicrous — representing just 15% increase per year.  When I look at my personal, residential data connection, my capacity has grown from a symmetric 1200 bps Hayes modem in 1982, to an asymmetric 15 Mbps/ 2 Mbps Verizon FiOS service installed in the summer of 2005.  That's a 23 year history in which my download bandwidth grew 51% per year.  Even my lame 2 Mbps upstream service represents a 38% per year growth over the past 23 years.  Why should the next five year's growth drop to 15% per year?

Interestingly I was looking at a similar estimation error in a different field just a few hours ago — projected growth in the number of mobile subscribers for the next four years.  The particular report I was looking at was from one year ago.  I won't embarrass the firm, but they are globally recognized telecoms market analysts.  They estimated there were 2.1 billion mobile subscribers in the world as of December 2005 and they projected there would be 642 million new subscribers added by December 2009.  This is nuts!

According to their own estimates (confirmed by ITU data and other sources) there were 343 million new subscribers added in 2004 and 345 million added in 2005.  Why would the rate of additions suddenly drop to 160 million per year? or did they expect the rate of growth to tapper off to near zero in four years?  What were they thinking?

I don't know the answer for under-estimating bandwidth requirements, but I have a theory about under-estimating mobile phone adoption.  People don't believe Moore's law will continue.  If you expect mobile phone infrastructure will always cost what it cost in 2005, and people will never get wealthier (despite the positive economic benefits of mobile phones), then perhaps you could conclude there's a near term limit on how many people can afford mobile phone service. 

But Moore's law does apply!  Every year the capital cost to build mobile networks declines and every year the cost of low-end handsets goes down — in both cases, despite the ever increasing functionality of the mobile networks and handsets.  More than that, countries with growing mobile phone adoption also have growing economic prosperity, so more people can afford the every lower cost service.

Indeed, all indications are that we added more mobile subscribers in 2006 than in any previous year.  Even better for me personally, rumors are that Verizon is about to up the bandwidth limits on my current service from 15/2 Mbps to 20/5 Mbps.  Bring it on!

Comments

[Stephen Smith]

I have a colleague who does a lot of organizational development work. He's fond of saying that we tend to overestimate what we can do in the short term, and underestimate what we can do in the long term.