Tweet Neutral Details

What is this all about?

Everyday activities such as driving, flying, washing clothes and, yes, even tweeting create greenhouse gas emissions which contribute to global warming. Each of us has a "carbon footprint," which is the measure of these emissions. We hope to raise awareness about the impact of our lifestyles, encourage energy conservation, and encourage further carbon footprint reduction via carbon offsets. You can offset all or part of your carbon footprint by contributing to the projects we feature on this site that reduce greenhouse gas emissions. Please join us in supporting efforts to fight climate change.

While we estimate that the average footprint of an individual twitterer is a relatively small 21.5 Lbs of CO2 per year, the 10 million users of twitter (Comscore, Feb 2009) together produce almost 100,000 metric tons of CO2/year, or the equivalent of 20,000 automobiles. We have taken a "supply side" approach: The manufacturer (poster) of tweets should account for the energy used in posting them, as well as used by followers in reading them.

Learn more about the projects receiving our support here.

A First Stab - Version 1.0 Footprint

Several carbon footprint protocols have been developed, but there is no widespread agreement on the boundaries of and methodologies for measuring the true carbon footprint of products and activities. Some footprint calculations cover only direct energy use, ignoring the significant carbon impact of the inputs used in creating or using the product. Others are “cradle to grave”, and tend to overstate the impact by including downstream emissions beyond the control of the company.

The fact is, footprint modeling is an inexact science, as much as some would have you believe otherwise. In this case, we developed these estimates based on averages, baseline proxies, and best available data. We welcome and invite revised estimates and additional scope suggestions. On the measurement side, server energy consumption (including HVAC) in hosted facilities and the locations of these facilities would be welcome from Twitter, as would more exact behavioral data. Additional thoughts on cell phone vs computer energy usage are also welcome.

Methodology - Version 1.0

Scope and Assumptions

In our estimates, we included the energy used by web servers serving tweets, energy used posting tweets, and energy used in reading tweets. We did not include energy used in hardware manufacturing or energy used in administrative functions associated with Twitter.

For our calculations, we made the following assumptions:

• Average followers per Tweeter = 45 (eliminates the top 2% of tweeters, who use feeds or employ a full-time twitterer)
• Average Twitter posts per year = 750 (based on total tweets/user, factoring out top 1-2% exhibiting extreme behavior)
• Average time to post a twitter = 90 seconds (estimate)
• Average time to read a twitter = 10 seconds (estimate)

Server Footprint Estimate

Google has publicly claimed .0003 kwh per search. Using this as a very rough proxy, we estimate .0135 kwh per tweet (45 followers for each tweet). 750 tweets in a year with 45 readers each would equal 33750 total, translating to 10.1 kwh. Using the California grid estimate of .6 lbs C02/kwh = 6.1 lbs of CO2 for the servers.

Note: The unknown variables here are the efficiency of Twitter's software, and frequency of data calls for archived tweets. A separate look at blog hosting found 30,000 blogs running on a Sun T2000 would require roughly 5,500 kwh per year to operate, or a much smaller 0.2 kwh per blog, so we feel that 10 kwh per tweeter is in the high side. With 7 million users, the 10 kwh figure would imply the equivalent of over 10,000 servers. By comparison, eBay reportedly runs on 12,000 servers worldwide.

Client-Side: Posting and Reading

There are many platforms, from PCs with widescreens LCD monitors (200 watts) to laptops (40 watts). We assumed an average of 120 watts which using national grid estimates of 1.3 lbs CO2/kwh, produces .156 lbs of CO2 per hour. 750 tweets at 90 seconds each would be 18.75 hours of posting time per year, or 1.3 lbs CO2 .

750 tweets read by 45 followers, at 10 seconds/tweet, would be 93.5 hours of total computer time dedicated to reading. Assuming that 10% of reading is done via handheld devices at a much lower energy use, we estimate .15 lbs CO2/hour multiplied by 93.5 hours, which equals 14.1 pounds CO2.

Totals

Server Footprint 6.1 lbs CO2
Client Side Posting 1.3 lbs CO2
Client Side Following 14.1 lbs CO2
Total 21.5 lbs CO2

Secondary (top-down) Estimate

We made a secondary estimate using the total electrical consumption estimates of internet usage, and estimated twitter usage. This figure was surprisingly close -- 20 lbs/user -- and calculated as follows:

Total US grid: 3,500,000,000,000 kwh
Total attributed to networking and electronics use: 8% or 280,000,000,000 kwh
Web related: 25% or 70,000,000,000 kwh
Twitter users as a percent of total web users (US) 3%, accounting for 2,100,000,000 kwh of web use
Percent of online time spent twittering or following .5% or 105,000,000 kwh
Equaling 15 kwh/user or 20 lbs CO2