The Noisy Channel


Is Google destroying the planet?

January 11th, 2009 · 11 Comments · Uncategorized

In Lewis Caroll’s Through the Looking Glass, the White Queen tells Alice that she’s “believed as many as six impossible things before breakfast”. Well, it’s a good thing I read this post about the environmental impact of Google searches before breakfast. It cites physicist Alex Wissner-Gross as saying that the average Google search generates 7g of CO2, using up half as much energy as boiling a kettle for a cup of tea.

I ate some breakfast after reading it, and now I’m feeling appropriately skeptical again. If nothing else, I doubt Google reveals enough about its internals for anyone to come to such a precise calculation.

[Note: Google explains here that the calculation is off by a fair amount–the average search generated 0.2g of CO2. Thanks to Jeremy for the heads up. Also, Jason Kincaid criticizes the Times Of London’s reporting here.]

Nonetheless, there may be something worth exploring in this argument. Even if the environmental cost of Google is far less than this research claims, it’s still a cost. Perhaps we should think about information seeking systems not only in terms of efficient use of human attention, but also in terms of other non-renewable natural resources.

In the automobile industry, we (at least in the United States) made the mistake of assuming the customer was always right, thus favoring SUVs over more economical and energy-efficient alternatives. Hopefully we’ve learned our lesson, though that’s still to be determined. In any case, perhaps there’s a similar lesson to be learned in the world of search engines.

11 responses so far ↓

  • 1 jeremy // Jan 11, 2009 at 8:38 pm

    Forget for a moment the cost of the searches themselves. The larger question that I’ve long had is whether, by relying on advertising and their main source of revenue, Google’s fundamental business model is a “consumption acceleration engine” or not.

    Are citizens more consumption-oriented when they are presented with additional, relevance-based advertising alongside their query, than they are without such advertising? Or are citizens less consumption-oriented, because access to more information on the web allows them to make wiser choices? Or are they equally consumption-oriented, and would buy and consume just as much junk without Google as they do with?

    That’s the bigger question that I have. Because if Google’s fundamental relevance-based advertising model is driving or accelerating worldwide consumption, then that is more globally devastating than the energy required to satisfy a few queries. Capish?

    I don’t know the answer, but I would be very interested in finding out. The only example that I have to go off, which is, is that relevance/related information presentation does indeed accelerate consumption. The “people who bought this also bought..” and “similar item” features are really a form of contextual advertising, just like AdWords. And folks at Amazon have told me how much of a boon to profits (and therefore an increase in consumption) those contextually-relevant ads truly are.

  • 2 Daniel Tunkelang // Jan 11, 2009 at 9:07 pm

    An interesting line of reasoning. But I don’t know how anyone could determine the extent to which Google’s is accelerating consumption vs. taking a share of existing consumption. I have a hard time believing that online ads spur impulse buys, but perhaps that’s because I block all of them.

    Even in the Amazon case, is there any evidence that they’ve increased global consumption, rather than simply claiming an impressive share of it?

  • 3 jeremy // Jan 12, 2009 at 2:20 am

    Yeah, I really don’t know the answer as to either how one can determine the extent of acceleration. I also do not know how to determine whether consumption has increased, decreased, or simply shifted from one company to another, from one communications medium to another. That is my ongoing question.

  • 4 Tech Breaking » Destructive Google! // Jan 12, 2009 at 7:48 am

    […] count of the searches crosses millions. Without rejiggering of Daniel Tunkelang’s words from The Noisy Channel, this is what we read: “[w]ell, itโ€™s a good thing I read this post about the environmental […]

  • 5 Albert Sheu // Jan 12, 2009 at 11:14 am

    If we’re attempting to find the environmental impact that Google has, it seems like it’s much easier to profile Google as a whole, rather than try to calculate the marginal cost per search and extrapolate, from that, the amount of CO2 generated total. I mean, we can just look at an average datacenter and try to estimate its power consumption, which we have very solid numbers for, rather than tracing the average query through its route.

    Starting with 7g and straight up multiplying by the volume of Google searches can lead to huge margins of error, not to mention that we’re either ignoring or misusing fixed costs in setting up the datacenter, cooling, engineering time, production costs, which in the end all of which combined probably doesn’t even come close to a fraction of a percent of the power used by the Internet as a whole.

    I remember reading before that Google’s power consumption is one of its highest costs, which I definitely believe. It’s in their best interest to be environmental, which has the nice side effect of saving the whales.

  • 6 Daniel Tunkelang // Jan 12, 2009 at 11:25 am

    I assume that’s what Alex Wissner-Gross tried to do. I’m still skeptical that the cost can be that high, and that Google discloses enough information to calculate that cost reliably.

    It might be more useful to run a Google Search Appliance at maximum load and see how much energy it expends per query. Maybe two of them, one acquiring content and the other processing queries. I know that isn’t the same as, but it at least would allow for controlled, repeatable experiment, and I’d think it would be in the same ballpark.

    Please send me a thank-you note if you manage to get funding for this. ๐Ÿ™‚

  • 7 jeremy // Jan 12, 2009 at 11:28 am

    Google’s response:

  • 8 Daniel Tunkelang // Jan 12, 2009 at 11:38 am

    0.2g of CO2 per search it is. A lot lower than 7g, but still much higher than I would have expected.

    An interesting point from their response:

    “In fact, in the time it takes to do a Google search, your own personal computer will use more energy than Google uses to answer your query.”

    There’s a lot to unpack in that observation, not the least of which is may be that pushing computation to the cloud in order to minimize the energy consumption of inefficient client hardware may be the best way to make computing energy efficient.

    In any case, props to Google for addressing this heads on. I often chide them for their lack of transparency, but here they went beyond the call of duty in their explanation.

  • 9 jeremy // Jan 12, 2009 at 1:13 pm

    I found this statement interesting: “so a Google search uses just about the same amount of energy that your body burns in ten seconds.”

    What this suggests to me is that if you are interested in energy efficiency, you should consider asking another human, rather than searching on Google.

    Consider that an average searcher often reformulates his or her query three or four times, and that each query costs 10 seconds of human energy expenditure, that means that if you can get the answer from a coworker or spouse within 30-40 seconds, you should probably just hold off and do that, rather than trying to get the answer on your own.

    Maybe I should also start reaching for that dictionary on my bookshelf more often, too, rather than doing Google lookups. If I can find the word in less than 10 seconds, there is a net gain for the planet. Same for units of measurement/conversion and Google calculator.

  • 10 Daniel Tunkelang // Jan 12, 2009 at 4:10 pm

    There are any number of reasons to consider alternatives to Googling everything. But somehow I’m guessing it’s going to be hard to save rain forests solely through more energy-efficient information seeking behavior.

  • 11 Daniel Tunkelang // Jan 12, 2009 at 4:45 pm

    A colleague of mine who prefers to retain his or her anonymity sent me this link as the final word on the subject:

Clicky Web Analytics