Is that 7g or 0.2g CO2 per Google search?

You may recall news stories last month claiming that a google search results in 7g of CO2 emissions.   This story resulted in a storm of comment and reporting, a clarification from google (0.2g per search), and somewhat of a clarification from the original study’s author.  But all the resulting hoo har goes to show:

• The original claim was woefully unclear as reported
• Releasing research headlines without the research is troublesome and results in misunderstandings
• We’ll need to get a lot better at identifying what we are actually measuring when talking energy and CO2

I want to break this story down and inject some facts in, and hopefully we’ll learn something in the process.

So, starting at the beginning:

The Sunday Times reported on January 11 that a Google search produced about 7g of CO2.  In their words:

Performing two Google searches from a desktop computer can generate about the same amount of carbon dioxide as boiling a kettle for a cup of tea, according to new research.

While millions of people tap into Google without considering the environment, a typical search generates about 7g of CO2 Boiling a kettle generates about 15g.

Now, you’d hope the rest of the article would go on to clarify this a bit.  That 7g per search.  What does that include? Where are the boundaries drawn around what a search is?  Not explained.  So, that get left to indivudual interpretation and that’s where this sort of measurement and claim gets messy and there is a resulting storm of voices claiming This and That.

Google quickly posted a blog post and said that the energy required by Google’s servers to handle one search is 0.2g CO2.

Supposedly measuring the same thing, but we have over an order of magnitude difference?   This comes down to what is actually being measured, as later clarifications revealed.

The original 7g of CO2 per search was actually made up from several searches and a few minutes of time sitting at a PC, and it included the energy of the PC used to start the search, not just Google’s servers.  Here’s the Jan 16 clarification by The Times:

A report about online energy consumption (Google and you’ll damage the planet, Jan 11) said that “performing two Google searches from a desktop computer can generate about the same amount of carbon dioxide as boiling a kettle” or about 7g of CO2 per search. We are happy to make clear that this does not refer to a one-hit Google search taking less than a second, which Google says produces about 0.2g of CO2, a figure we accept. In the article, we were referring to a Google search that may involve several attempts to find the object being sought and that may last for several minutes. Various experts put forward carbon emission estimates for such a search of 1g-10g depending on the time involved and the equipment used

Bingo. That’s the detail we originally needed.  It ain’t about Google’s servers or the search itself, but about you sitting down in front of a foot-warming PC with a big, bright screen and tapping away for a bit trying to find something out.  And we now have a range of 1g to 10g depending on circumstances.

So, the mention of Google at all in the story is pretty spurious,  they claim 0.2g for their part of the search, the rest is elsewhere.   A more correct statement could to be something like… “Using a PC and the Internet produces CO2 at the rate of between 1 and 10g CO2 per few minutes depending on your computer setup and what you are doing” (or something like that, please don’t quote this statement).

Basically, the Sunday Times got it wrong.  They did the classic lazy blame-somebody-else story, blaming the CO2 on Google, when it is really much more about a home PC and how it is used, and the rest of the Internet equipment used to move all that data around.

One more quote, from a followup article from TechNewsWorld put it basically to rest:

One problem: the study’s author, Harvard University physicist Alex Wissner-Gross, says he never mentions Google in the study. “For some reason, in their story on the study, the Times had an ax to grind with Google,” Wissner-Gross told TechNewsWorld. “Our work has nothing to do with Google. Our focus was exclusively on the Web overall, and we found that it takes on average about 20 milligrams of CO2 per second to visit a Web site.”

And the example involving tea kettles? “They did that. I have no idea where they got those statistics,” Wissner-Gross said.

An average 0.02g of CO2 per second.  That’s 1.2g per minute, or 72g CO2 per hour.

Contrast that to driving your car, which likely produces 200g CO2 per km or more.  Drive 1km, or browse the net for nearly three hours?