According to Cisco there’ll be an average of two internet connected devices per person by 2015. The company also believes there’ll be almost three billion internet users by that date.

The stats are part of the company’s latest “Visual Networking Index” forecast, which aims to predict future net use and its contributing factors.

The headline figure is that by 2015, the total amount of traffic throughout the year will be right around the one zettabyte mark: in other words, one trillion gigabytes. To put that into context, the company is forecasting that the data growth between 2014 and 2015 will be more than the total data used in 2010.

While the logic of more users and more internet devices meaning more traffic makes sense, some of the company’s other arguments are more shaky. It claims average broadband speed doubled over the past year from 3.5 Mbps to 7 Mbps, and will reach 28 Mbps by 2015.

The current 7 Mbps figure sounds very high for a global average, and even if that is the case it may simply be advertised theoretical maximum speeds, which bear little resemblance to reality.

Of course, if that’s applied consistently, the trend might still be valid. However, this ignores the point that just because maximum speed increases, total traffic may not increase at the same rate. In many cases, people may simply download the same amount of data in a shorter time.

The biggest problem with the forecasts is that they appear to fall into the inferential statistics trap. Put simply, this means looking at a historical trend and assuming that it will continue to grow in the future, without taking into account limitations on that trend. In this situation, the limit is that the new Internet users are most likely to be in locations where for either economic or geographic reasons, maximum speeds are considerably lower than today’s average.

The best example of this is a case study noting that there were an estimated 37 Elvis Presley impersonators worldwide in 1977, a figure that grew to 48,000 by 1993. Extrapolating that trend would mean that by 2010, there would be 2.5 billion Elvis impersonators, around one-third of the world’s population.

That proved not to be the case.

(Picture credit: Paul Smith / Martin Fox)