Does not compute: Avoiding pitfalls assessing the Internet's energy and carbon impacts
A recurrent theme is that well-intentioned research often overestimates IT’s electricity use and climate impacts, sometimes by orders of magnitude. These results then become ‘‘factoids’’ that spread quickly as people share them and the media report them.1 The problem, of course, is that incorrect numbers can have real-world consequences when widely believed. Consider the growth of global Internet traffic in the 1990s, as described by Coffman and Odlyzko.2 Those data flows doubled every year or so for years but doubled every 100 days for parts of 1995 and 1996. Growth reverted to doubling every year after that, but extrapolations based on ‘‘doubling every 100 days’’ led to vast overinvestment in network capacity around 2000. In 2002, more than 97% of fiber capacity sat unused,3 and because electricity used by operating network equipment is roughly constant, regardless of whether it is used or not, that overcapacity had significant energy implications. Consequential mistakes can result when analysts’ inherent curiosity about an important topic area collides with a pervasive lack of accurate and up-todate information. IT changes so quickly that most data characterizing it are obsolete in short order, and people’s inability to accurately predict the effects of exponential change in multiple dimensions can make things even worse. The best data on IT’s energy and emissions characteristics are closely held proprietary secrets among tech companies, which compounds these problems.
Search for the Publication In: