A Berkeley study from 2011 shows that the Internet only about 2% percent of power worldwide, whereas transportation uses about 60%. This is about one-fourth of previous estimates putting worldwide power consumption at about 10%.
The paper advises that the Internet is not worth worrying about relative to “dirty” physical services, and that virtualizing products and services from the physical to network is all good.
These numbers are lower that many reported elsewhere. Other estimates have varied up to 10% of the U.S. electrical supply. What gives?
The Berkeley study ignores the fact that Internet use is uneven worldwide. In many countries the Internet has low penetration, according to the Internet Index. In others, like the United States and Western Europe, it is very high. To get a quick “back of the envelope” calculation, we assume that:
- In 2012, the USA uses 20% of the world’s electricity – http://www.eia.gov/tools/faqs/faq.cfm?id=87&t=1
- In 2012, the percent of Internet users in the US and advanced countries was 2-2.5 times greater than in Asia and Africa – http://www.internetworldstats.com/stats.htm
Now, its is possible that US users are hyper-efficient, and the web elsewhere is more energy-intensive. But this is unlikely. The structure of data centers worldwide is mostly the same, and the data center model was developed by the US. Alternately, we could assume that US users are just using more of the “idle time” on the Internet – the devices have to be on all the time. But this is also unlikely – people in the US spend lots more time surfing the web than someone checking their cellphone service in Asia.
So, its safe to assume that if we have twice as many US users on the Internet, they result in twice as much power being used.
This implies that we should double the 2% rate to something closer to 4%. This number is closer, but still not the 7-10% numbers reported earlier.
However, to really compute the impact of the Internet, we need to think of the near future, rather than the present. The Internet is growing much faster than electricity use worldwide.
- According to the study I mentioned above by Cisco, Internet use is expected to go up 4 times between now and 2016 – http://newsroom.cisco.com/press-release-content?type=webcontent&articleId=888280
- More detail on its growth: http://www.cisco.com/en/US/solutions/collateral/ns341/ns525/ns537/ns705/ns827/VNI_Hyperconnectivity_WP.html
- In the same time, US power use will only go up 4% – http://www.ecmag.com/index.cfm?fa=article&articleID=11176
So, unless there is a massive increase in the efficiency of the web, by 2016 we will suck 4 times as much power for our ongoing “virtualization” of physical products and services, for a number around 15%. Right away, our “sustainability red flags” go up. This isn’t a stable situation, but one that is rapidly changing.
The results shouldn’t be strange. We’re used to the idea that even though the US has a small component of the world’s population, we produce more trash and use more fossil fuels. It’s the same with the Internet – we devote a greater percent of our energy to the network. But the will to believe that “technology will save us,” especially computer tech, is so ingrained that there is a will not to believe.
But this in turn implies a way to compute sustainability of web efforts. To properly weigh the value of the electricity used by the Internet, we should multiply by its importance to a particular country or region. Enter The Web Index, which does exactly that. Converting by the Web Index will mean the following:
In countries with a very low Web Index, the chance of an Internet service adding, or replacing a physical service is quite high. Since this will usually replace a high-energy physical process, the new service makes that society more sustainable. The designed Internet service “blue,” meaning it actually improves, rather than simply sustaining current practice.
However, in societies where the Internet is in widespread use, most of the low-hanging fruit replacing physical transportation may have already happened. The percentage of power devoted to the Internet, relative to total consumption, is much higher. In addition, the low-hanging fruit has been taken. The new service either overlaps or competes with another virtual service, or creates a brand-new, energy-using service that does not exist in the physical world.
As an example of the Internet replacing a service that was more green, consider that a decade ago lots of graphic design was done on paper. The designer would sit for hours sketching and erasing. The paper wasn’t draining energy, and the designer’s brain was pulling only about 20 watts of power. Contrast that with a graphic designer today that jumps onto their widescreen workstation first thing in the morning and does “discovery” – experimentation using a design tool like Adobe CS. During the time their logo or other identity is being figured out, the computer is sucking around 200 watts of power, more if it is a big fancy workstation or if 3D is being use. The resulting “embodied energy” of the design logo is much higher, and the net energy consumption of design itself has risen.
The other thing we have to consider is rising Internet use. Extending the Cisco study, the Internet would consume 100% of the world’s power in in less than two decades. Obviously, this won’t happen, and physical services might rise as well.
However, in all those countries with a low Internet Index, current physical services are often pretty terrible. That is, people in these countries would be doing a lot more physical things, but are prevented by the lousy state of roads and distribution. Put in the Internet (which can often bypass the current infrastructure), and you’ll increase consumption by the population to a greater extent than you would in a country with great roads. The Internet will start from the beginning pulling a large part of generated energy. This is essentially Jevons’ Paradox at work. This is also happening for some Internet services, which, like e-commerce, almost certainly increase consumption of physical products. E-Commerce is likely fueling more mail-order purchasing, since people can find and shop for stuff they would never drive to find. Plane traffic has risen, despite the rise of teleconferencing.
The take-home: You do have to worry about the Internet’s impact on the environment.