The icon indicates free access to the linked research on JSTOR.

Quick! What’s the average velocity of an unladen swallow? Better do a search. But that quick bit of knowledge comes with a hidden cost: greenhouse gas emissions. How significant is the internet’s carbon footprint?

JSTOR Daily Membership AdJSTOR Daily Membership Ad

According to Columbia University earth and environmental scientist Joel Gombiner, writing in the journal Consilience, it’s substantial. People forget. To the casual user, the internet simply seems like a magic content fair. But powering the internet requires a lot of electrical hardware. For starters, that hardware needs to be manufactured, shipped to its destination, and installed in buildings, some of which are built for the purpose—all of that means emissions. The materials used to build that hardware, same thing: raw materials need to be mined or fabricated, then shipped to the manufacturing location.

However, construction and shipping impacts exist for all physical human-made objects, and in that context the manufacturing of internet hardware is difficult to compare to other sectors. So how about the operation of that hardware?

All the data in the world that collectively makes up “the internet” is stored on souped-up hard drives called servers. Servers are the repositories and movers of internet content, the physical locations where the data is stored that route data where it needs to go. When you search for “unladen swallow velocity,” the data containing the answer is retrieved and routed to your device by one or more servers. Many servers are stored together, where they not only require a lot of electricity to operate, but also  powerful and energy-intensive cooling systems. Many large internet operations, such as Google, locate huge numbers of servers in China, where coal-generated electricity remains common. These servers are on and running 24-7.

A bigger issue, however is that what makes the internet efficient to use in terms of answering Monty Python queries is highly inefficient in terms of power use. Using the example of a Google search, Gombiner explains that basically an entire copy of all the data on the internet to which Google has access might be stored on one server. However, to comb through all the data on one server is time consuming, so the standard search actually combs through multiple servers at once to generate a faster response. The amount of data used increases the power usage of a server. So, while reading a text webpage generates some emissions, something data hungry, like a video, generates many more. Then there is email, gaming, and whatever other uses you can imagine.

To calculate an exact amount of emissions for one search is difficult. Google scientists, who admittedly have motive to come in on the low end, came up with .2 grams of carbon dioxide per search, while, on the high end, a Harvard physicist calculated 7 grams. Whatever the exact figure, it’s clear that millions of people are running searches around the clock. And by some estimates, internet use accounts for around 2% of all carbon emissions. The clear solution is to power the web with renewable energy.


JSTOR is a digital library for scholars, researchers, and students. JSTOR Daily readers can access the original research behind our articles for free on JSTOR.

Consilience, No. 5 (2011), pp. 119-124
Columbia University