Tempest in a Tea Kettle
Evan Mills and Jon Koomey** - February 2, 2009
Mills is a Staff Scientist at the U.S. Department of Energy's Lawrence Berkeley National Laboratory, at the University of California.
Koomey is Project Scientist at Lawrence Berkeley National Laboratory, and Consulting Professor, Department of Civil and Environmental Engineering, Stanford University.
Letter submitted to the Sunday Times of London, a small excerpt of which was published here.
Erroneous carbon footprint estimates can steer energy users away from much-needed climate change solutions.
As a case in point, a hastily thrown together article in London's Sunday Times by Jonathan Leake and Richard Woods "Revealed: The Environmental Impact of Google Searches" (Jan 11) perpetuated misleading claims about the carbon intensity of the Internet, quoted one of us (Mills) out of context, misquoted other studies, and cited data that has not been well documented or peer reviewed and is therefore of uncertain reliability.
The article opens by asserting emissions of 7 grams carbon dioxide per web search (half that, they claim, of boiling a cup of tea), while Google politely responded that the number is 0.2 grams (about 1/35th of the author's value, and subsequently accepted by the Times). Muddying the water, the higher number probably includes energy used by the individual searcher's own computer [we will know for sure when the in-progress research it cites is published].
The article then uses a truncated version of Mills' statement about the energy-intensity of data centers to legitimate an effort to equate web searching emissions with those from global aviation. The unfortunate comparision to aviation actually includes not only datacenters but also personal computers, printers, telecommunications energy use plus the carbon embedded in the manufacture of those goods (how much carbon is embodied with the manufacture of that tea-kettle?). Most of this has nothing to do with web searches, or is otherwise outside the control of Google or any other search provider.
Now that we see how they used our quote, we would only add that a datacenter could actually become completely carbon neutral through judicious mix of energy efficiency and on-site renewables or purchased "green" power. This leads to the opposite bottom line as that tendered in the Times article. We sent a clarification to the Times, which eventually published a much-abridged version of our response.
To date, no one has conducted competent peer-reviewed full life-cycle analyses of the carbon footprint of the internet (or the alternatives), so the numbers cited in the Times article should be treated as no more than speculation by industry observers or reporters who might have an ax to grind. Information technology (IT) facilities do use electricity, but moving electrons is always less energy intensive and environmentally damaging than moving atoms. By largely ignoring the structural changes in the economy that IT enables, the Times has painted a misleading picture of the environmental effects of IT. This myth was actually put to rest a decade ago. More on the Internet Energy Myth (from the Epilogue of Turning Numbers into Knowledge, 3rd edition, 2017). While it is important to improve the energy efficiency of IT facilities (and the industry is making great strides in that effort, particularly Google), it is the NET environmental impact that matters, not the direct electricity use and carbon emissions of these facilities treated in isolation.
Indeed, driving an average car just one mile and back to the nearest library to manually search for information produces more than 100-times more greenhouse-gas emissions than a web search,* even given the dubious numbers proffered by Leake and Woods. Just the emissions associated with manufacturing the paper upon which the Times article was printed (not to mention, running the newspaper or transporting the papers) eclipses those from reading it online.**
* Based on a US EPA assumptions for typical autos, a two-mile roundtrip drive to a library to search for information would entail emissions of nearly 900 grams CO2. This is about 4500-times that of a single web search, according to Google's estimate of 0.2 grams of CO2 per search.
** According to EDF's Paper Calculator, one pound of newsprint (with 50% recycled content) is responsible for about 2 pounds of CO2-equivalent (~1000 grams!). Assuming 30 pounds of paper per ream, and 500 sheets (24x36 inches) of newsprint per ream, there would be about 71 grams of CO2-equivalent for a single sheet of newsprint. This is 350 TIMES that of a search per Google's numbers and 10-times that of the dubious number in the Times article. These values for paper don't not even include the secondary embodied energy and carbon associated with publishing and distributing a newspaper.
Thanks to Klaas Jan Kramer for helpful references.
Original article [Online version]
Our response [PDF]
Alex Wissner-Gross' article (part of the basis of Leake and Woods article)
Clarifications by Wissner-Gross
Taylor, C. and J. Koomey. 2008. "Estimating energy use and greenhouse gas emissions of Internet advertising." Working paper for IMC2. February 14 [PDF]
Data disclosed by Google [NY Times article]
LBNL research on energy management in data centers