What is the carbon footprint of a a single spam message?

July 30, 2010 12:06 PM

A friend of mine sent me a infograph about how much CO2 is produced from a single spam email. In the infograph it said, "The average GHG emission associated with a single spam message is 0.3 grams of CO2" I'm not sure the the claim is a valid one and would like to see some similar studies or links about the topic.

Below is what the infograph was based on:


"The “Carbon Footprint of Spam” study looked at global energy expended to create, store, view and filter spam across 11 countries, including Australia, Brazil, Canada, China, France, Germany, Japan, India, Mexico, Spain, the United States and the United Kingdom. It correlated the electricity spent on spam with its carbon footprint, since fossil fuels are by far the largest source of electricity in the world today. Since emissions cannot be isolated to one country, it averaged its findings to arrive at the global impact. Key findings of the “Carbon Footprint of Spam” study included:

The average GHG emission associated with a single spam message is 0.3 grams of CO2. That's like driving three feet (one meter); but when multiplied by the yearly volume of spam, it is equivalent to driving around the earth 1.6 million times.

Much of the energy consumption associated with spam (nearly 80 percent) comes from end-users deleting spam and searching for legitimate e-mail (false positives). Spam filtering accounts for just 16 percent of spam-related energy use.

Spam filtering saves 135 TWh of electricity per year. That is equivalent to taking 13 million cars off the road.

If every inbox were protected by a state-of-the-art spam filter, organizations and individuals could reduce today’s spam energy by 75 percent or 25 TWh per year, the equivalent of taking 2.3 million cars off the road. (my note seems like they are just trying to pitch a product)

Countries with greater Internet connectivity and users, such as the United States and India, tended to have proportionately higher emissions per e-mail users. The United States for example, had emissions that were 38 times that of Spain.

While Canada, China, Brazil, India, the United States and the United Kingdom had similar energy use for spam by country, Australia, Germany, France, Mexico and Spain tended to come in about 10 percent lower. Spain came in at the lowest, with both the smallest amount of e-mail that was received as spam and the smallest amount of energy use for spam per e-mail user."

posted by wherespaul to science & nature (10 answers total)

Start here.

The long and the short of it is that this is more or less a bit of propaganda put out by McAffee to try to sell spam filtering software.
posted by FAMOUS MONSTER at 12:20 PM on July 30, 2010 [1 favorite]

The study (PDF)

Basically, none of the emissions are from sending the email, as you would expect. They are from false positives (looking for stuff in your spam folder), running spam filters, and from the end-users viewing the spam.

McAffee is arguing that better spam filtering cuts down the other two of those, so it is a carbon beneficial thing.
posted by smackfu at 12:24 PM on July 30, 2010

Much of the energy consumption associated with spam (nearly 80 percent) comes from end-users deleting spam and searching for legitimate e-mail (false positives).

This I find a little bit dubious, what would that energy be doing otherwise? Surely the saving in time trumps any carbon saving? To my mind it's like saying working from home reduces shoe leather wear. May be true but it's so tiny.
I had similar issues to this in my last job with 'time savings'.
scheme X results in Each person saving 2secs, 30,000 people a day use the road, that's 60,000 seconds or 16 hours. So the scheme saves 16 hours every day.

I just don't think tiny things make big things.
posted by 92_elements at 12:34 PM on July 30, 2010

This whole line of argument seems to spring from the notion that during the times when an office user is using their workstation to sort through their inbox, all of the electricity used by that workstation can be accounted under 'The Carbon Footprint of Spam' which I find ludicrous. Of course offices have tons of workstations and the typical office person spends a lot of time reading/writing email, but the notion that if they didn't have to deal with spam that their computers would be turned off for those time periods is simply ridiculous. The computer at each person's desk is going to be on the whole day regardless of whether there is zero or lots of spam -- except perhaps the times when the person isn't at their desk and the monitor shuts off automatically -- so the notion that blocking spam would cause less energy use is simply absurd. Any time saved by not having to filter through emails will just be time that the worker does some other computer-related activity like using Excel, writing documents, or playing solitaire.

All their argument boils down to is "workers can be more efficient if they don't have to wade through spam," but that's the same pitch that spam-filtering vendors have been making since the beginning. Some dim bulb decided that they could pervert that message into something meaningless that everyone could use to guilt people into buying their product with and ran with it.
posted by Rhomboid at 2:00 PM on July 30, 2010

Hi, I'm a PhD student and I study this sort of stuff.

I could point you to a half a dozen studies that make similar claims about the carbon footprint of downloading a music track, burning a CD, ordering a movie via Netflix, reading a journal article online, and so on. For what it's worth, 0.3g per spam message feels reasonable, in that it doesn't disagree wildly with similar studies. The way the measurements are conducted is something like this:

Client side:
1. Estimate total life-cycle carbon footprint of the client computer and divide by computer lifespan (typically 4 years) to get a footprint-per-second-of-use
2. Estimate the amount of time spent doing activity X, e.g. "We estimate it takes the average user 3 seconds to view and delete a spam message"
3. Multiply these two figures together to get carbon footprint-per-activity
4. Multiply by the number of times the activity is performed to get overall carbon footprint.

Server side:
5. Estimate carbon footprint of the internet and divide by total data traffic to get footprint-per-MB
6. Estimate data traffic per activity X
7. Multiply them together to get footprint-per-activity X
8. Multiply by the number of times the activity is performed to get overall carbon footprint.

In my opinion, there are all kinds of problems with these studies. Even though I know and respect some of the people who are putting them out I think they are missing the mark, every single one of them. Let me rattle off some of the problems:

- The estimate in part 2 is pulled out of the sky. Often it comes from surveys or questionnaires. But users multi-task and run background tasks and do all kinds of things that aren't picked up in questionnaires. What they should do is monitor users over a period of time and record the percentage of time which is actively spent on the activity, but this is rarely if ever done.

- There is an assumption that user time scales linearly. If one person gets one spam message and spends 3 seconds on it, the person who gets 1000 messages spends 3000 seconds on it. I suspect, though don't know for sure, that these relationships are not linear, and that this might undermine the validity of step 4, the aggregation.

- User time is treated as a conserved resource, where the removal of a 3-second activity means an energy savings of 300 Watt-seconds, assuming a 100W computer. But for most people, computer usage expands to fill the available or allotted time. If you were to study users who used anti-spam software and those who didn't, you should be able to see this claimed energy savings in their power bills, if the savings were real. Of course, it isn't real.

- The estimates in step 1 and 5 are completely bogus to begin with. There is huge uncertainty about carbon footprints and I would trust it only within about an order of magnitude, and that's being generous. (The problem is mainly due to the manufacturing side of the product life cycle.)

When it comes to quantifying the footprint of activities, things you do on a computer are the hardest to pin down and the exercise is probably the most useless. At the root there is an assumption that users are intentional about their tasks; that the personal tech-related footprint of a user is a function of the tasks they wish to perform, which means you can assign a percentage of their footprint to a specific task. The personal footprint is directly related to the number of hours spent in front of a computer and I suspect, but again don't know for sure, that you would never be able to establish a link between tasks and hours spent, and instead would find very strong associations with things like occupation, socioeconomic status, age, and so on.

In other words, it's pretty much a marketing piece, which is unfortunate, but not really shocking.
posted by PercussivePaul at 2:03 PM on July 30, 2010 [38 favorites]

bah. it doesnt pass my BS filter.

1. per this link: there were 90 trillion emails in 09. 81% were spam so thats 72900000000000 spam messages
2. at ".3 grams per..." thats 21870000000000grams or 21870000000000 metric tons.
3. convert that to "1000's of metric tons" and its "21870 1000's of metric tons" ( i've used the same units as wikipedia...) so now go ahead and compare that to global emmissions for that year, per wikipedia and
4. that would make spam the 81st most polluting country in the world with regard to CO2 emmissions.


bah.
posted by chasles at 2:09 PM on July 30, 2010 [1 favorite]

well said PercussivePaul.
posted by chasles at 2:10 PM on July 30, 2010

OMG sidebar! :D yay
posted by PercussivePaul at 9:07 PM on July 30, 2010 [2 favorites]

This isn't remotely appropriate for AskMe, but I'd like to apologize to wherespaul, who apparently really did want to know about this and wasn't just self-linking when he first put this issue up on the blue.
posted by yhbc at 9:34 PM on July 30, 2010

PercussivePaul your insight was very helpful and congrats on the sideblog.

yhbc The fault was mine not yours. In the past my visits to metafilter were fast visits. Was just here to search for a answer to an question, then I would leave the site once I found an answer. So when I posted the other day I had not noticed the AskMefi section. Yes a newbie move but now I know better :)
posted by wherespaul at 11:56 AM on August 1, 2010