The Ouroboros Strategy, Part 2: Sustainability computations for the web


Recording Gains in sustainability requires moving beyond green ingredient lists to quantification. We can do this by using a mix of Web Performance Tools, online calculators, and “back of the envelope” calculations.

As a first step, a site might list values gleaned from the design process:

  • Reductions in “vampire power” by office electronics leading to lower electricity use to create websites
  • A shift to a truly “green” web hosting service which provides documentation for their claims
In the next step, we would run tools created for Web Performance Optimization (WPO), as well as Search Engine Optimization (SEO), and post the results.
  • Web Performance Optimization (WPO) results might include
    • Reduction in downloaded bits
    • Increases in speed
    • Reduced HTTP requests
Finally, we could look as site analytics to define our users. This would help us analyze the UX and business model aspects of sustainability.
As an example, since “Mobile First” is one of the most powerful strategies for sustainability, it is useful to document shifts from desktop to mobile use, especially if it resulted from a site redesign.

While these numbers gives us a measure of overall increases in efficiency, they doesn’t tell us how much this contributed to saving the planet. To do that, we first have to compute the impact of our efforts on the energy and resource consumption of the Internet. Then, to put it in terms everyone can understand, we should also convert this change in consumption to potential reduction of greenhouse gas.

In the following, I suggest several methods that might be used to make rough calculations of this sort. Admittedly, these will be ballpark estimates, and for small sites, the gains might not seem huge. But remember that if everyone does the same with their sites, overall sustainability will be greatly improved.

Here are my proposed techniques. Each starts with a specific number of page views, which should be calculated for your website on a yearly or other basis. If you don’t have exact numbers, estimate using your site’s statistics.

Your total savings = Method1 + Method2 + (Method3 or Method 4) + Method 5

Method 1 –HTTP requests have their biggest effect on servers, and an estimate of their impact was made by Google efficiency guru Steve Souders in the following blog post: http://www.stevesouders.com/blog/2008/03/06/how-green-is-your-web-page/

  1. Calculate your number of page views per year, assuming constant traffic
  2. Calculate the total HTTP requests by multiplying the results from YSlow or a similar tool, times the number of page views
  3. Assume that one server can handle about 3300 million HTTP requests per year, and generates about 500 pounds of CO2 emissions.
  4. CO2 reduction = 500 pounds * (your HTTP request reduction/3300 million HTTP requests per server per year)
  5. The answer will be the net reduction in CO2 emissions, at the server-side, causes by your efficiency increases.

Method 2: Bandwidth reduction using Eiolca.net

This method is most appropriate if your sustainability efforts result in substantial changes in the total bytes downloaded per page. It uses estimates from an online LCA model hosted at Carnegie Melon at Eicolca.net.

  1. Calculate the number of byes saved per page download, and multiply that by your page views
  2. Calculate the cost of the saved bandwith. For example, if your ISP charges x number of dollars per gigabyte, multiply your bandwith times that value to get the amount of money, in dollars, saved
  3. Go to Carnegie Mellon’s Eiolca.net website at: http://www.eiolca.net
  4. Select “Internet Service Providers and Web Portals” as your industry sector, and plug in the number of dollars you saved
  5. The model will directly compute the total CO2 that won’t be released into the environment

Method 3: User time reduction using Eiolca.net

This method mostly applies to decreased use of desktop computers, which consume large amounts of power relative to  mobiles. It’s most appropriate if your site provides a specific service, and user testing shows that you have decreased the time needed to use the service.

  1. Determine how much faster people can use your site to accomplish typical tasks during one session
  2. Multiply by page views to get the number of seconds saved
  3. Calculate the reduced time devoted to your site. It is true that this may not decrease your user’s overall time on the computer. However, you’ve decreased your share of that time, so the calculation is still valid.
  4. Calculate the cost of that reduced time as fraction of bandwidth charges, either their cellphone carrier or ISP fees for desktop access.
  5. Finally, if the main effects of your redesign are faster user completion of tasks, you can convert decreased page view time to carbon. Calculate the number of seconds saved for a given number of users. For mobiles, the savings are small, but for desktops a value of 30mg/second carbon, following the estimates of Alexander Wissner-Gross http://www.alexwg.org/ might be appropriate.
Method 4: Green Office electricity calculations
  1. The same online LCA tool could be used to compute savings from a “Green Office”.
  2. Alternately, we could use LCA or equivalent analyses based on graphic design shops, which have a very similar organization these days compared to web design shops. There are lots of qualitative analyses, like this “setting up shop” article series – http://ecopreneurist.com/2008/05/23/finding-your-path-defining-green-for-your-business/

Method 5: Benefit of shifting users from desktop to mobile, using Eiolca.net

This method is most appropriate if the primary effect of your redesign is shifting users from desktops to mobiles. You can calculate how much energy was saved by moving them to energy-efficient computers.

  1. Calculate the amount of energy used by your users for a given number of seconds of use. If the average use of your site was 10 minutes, you would use 600 seconds. Multiply that by a typical energy drain of a desktop, about 140 watts. This gives you a starting amount of electricity used.
  2. Compute the number of users who switch from desktop to mobile access
  3. Assume that a smartphone uses about 1/15  the energy of a desktop, or about 20 watts. For tablets, you might want to use 1/5, or 20 watts. Remember that watts measure energy per second.
  4. Calculate the reduction in energy use.
  5. Convert that to the amount of greenhouse gases released by your utility when it generates said number of watts. If you can’t get this value, you can estimate it by computing the cost of the energy (convert to kilowatt-hours from watts per second), and plug this into the Eiolca.net website, using “Power Generation and Supply” as your sector.

2 Comments

  1. Hmmm, that’s a question for Steve Souder. It is true that GET is often faster, which may mean that there is less CPU overhead. But I don’t know if that is true. If you’re really interested you could look at Steven’s awesome site at http://www.stevesouders.com where he has some hard data and links to WPO slideshows. Definitely worth figuring out.

    My guess is that the reason GET is less safe is programming practices, rather than an intrinsic problem.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s