How Green is your Email?

April 22, 2009 8:17 AM

Can one quantify the carbon footprint of email?

Perhaps I'm grasping at straws, but I'd love it if there was a simple way to quantify the carbon footprint of emails, given the size of the email and the number of recipients. There's been some discussion of this recently, but it's either been too general or focused on spam.

Full disclosure: I've found that my company has a nasty habit of sending internal memos as PDFs or .docs or images, when the information could easily be conveyed through plain text. Today, after seeing a 591K email going out to everyone in the building to celebrate Earth Day, I was hoping to politely point out the impact this has, and suggest that there may be a better way.

I'm thinking this would be most easily approached in terms of electricity used by the server to send these out, but would be curious if there are any other ideas. What little I know about email servers is irrelevant as soon as electricity use comes into play, so I appreciate baby steps where necessary. Thanks in advance.

posted by SpiffyRob to computers & internet (10 comments total)

I would guess that the marginal electricity use of sending an email is vanishingly small.
posted by sinfony at 8:33 AM on April 22

Not exackly relevant, but tangential - the carbon footprint of a google search.
posted by jourman2 at 8:33 AM on April 22

I'd say this is grasping at straws, yes. Energy is hot. When you make toast, how hot does your countertop get? When you receive new email, how much hotter does your computer desk become?

There's also an economic way to think of this: The price of things is a good estimation of their energy use. How much do you spend to drive and how much would interent cost if everyone stopped the youtubing and just sent emails to each other?
posted by sleslie at 9:09 AM on April 22

Not trying to moderate the thread, just want to make sure what I'm asking is clear.

It's not about not sending email at all, and it's not about "being that guy." (I understand what you mean, but I assure you, I have the ability to present ideas without being a dickhole if I feel so inclined. Professionalism and all.)

The main question, which in reading now I realize I was slightly vague on, is how much would be saved by using regular ol' email (even HTML email) in lieu of weightier attachments. For instance, say an email goes out with a 600KB attachment wherein that info could be instead presented as text. We've got roughly 400 people in the office, so that's 240MB vs. (assuming 10k for plain text and headers) 4MB. Say one of those goes out every day (let's call it 250 days/year) that's 60gigs versus 1gig.

I'm not so much concerned for the computers receiving as the servers sending, as that seems to be the approach investigated by the articles I linked to. (And the one jourman linked to, which is helpful.)

I'm not suggesting that sending 60 gigs is SIXTY TIMES WORSE than sending one, I was just curious if there's a semi-meaningful answer out there for what the difference would be. If, in fact, the distinction is as small as zpousman is alluding to, so be it, and thanks to everyone for there help. I honestly just didn't/don't know.
posted by SpiffyRob at 10:22 AM on April 22

I'd say that an argument for increased productivity is better than one for electricity usage:

- employees have to wait for said large email to download, even those who won't read it
- attachments can't be searched using readily-available tools, which would be helpful if someone was trying to find something specific
- email has to be stored on all receiving computers, thus increasing storage requirements for everyone in the company

Don't use "be green" or "carbon footprint" as the main argument. Businesses don't care about these things. Talk in dollars only (time can be expressed by multiplying by everyone's salary).
posted by meowzilla at 10:55 AM on April 22

I'm not suggesting that sending 60 gigs is SIXTY TIMES WORSE than sending one
How is sending 60 gigs not 'worse than sending one' (where 'worse' -> more expensive, uses more resources, or pick another measure)?
That said, (60*trivial) is only marginally less trivial. How many of the e-mailers leave work at 5 o'clock in a low-MPG vehicle vs. by a more efficient car, bike, or on foot? Estimate the carbon footprint improvement of just one of those people improving their travel efficiency -- heck, estimate it for just one of those people improving it for just one workday a year! How does that stack up to (60*trivial)?
posted by TruncatedTiller at 10:55 AM on April 22

OK, let's try to work this out.

Coal-fired powerplants produce about 2 lb CO2 per kWh.

Assume you have 1 server that runs at 750 W and 60 desktops that run at 200 W (I'm using Apple computer as an example because the info is easy to find on their website). Lets say that at idle, each runs at 20% maximum power (150 W and 40 W).

So how many kWh does an e-mail account for? Lets say, with complete arbitrariness and just so we have some numbers to toss around, that for every 1 MB of e-mail, the sending and receiving computers will go from idle to full capacity for 1 second.

So a 1 MB e-mail message will result in an uptick of (600 watt seconds * 1) + (160 watt seconds * 61) = 10360 watt seconds or 2.87 watt hours, or .00287 kWh, which would be responsible for about .092 ounces of CO2. We'll use this as our base unit for the CO2 footprint of an e-mail.

So now we have something to play around with. If you send out a 6 MB e-mail every work day for a year, that's about 250*6 units, or about 8.625 lb of CO2 per year. If your messages were 1/6th as large, that'd be about 2 lb of CO2.

Note that these numbers are all ridiculously inflated—probably by an order of magnitude—because of my initial assumptions.

You might be able to build a more persuasive case in terms of hogging company resources that would eventually require fancier equipment, but even with that, bloated e-mail is not going to prove the low-hanging fruit.
posted by adamrice at 10:56 AM on April 22

The general method for this sort of calculation is to tally up all of the computing and networking infrastructure involved in the email, and figure out what percentage of this infrastructure was dedicated strictly to this email. It will be a very very small percentage, but non-zero.

You would need to know what is the marginal energy usage for one email; it will involve some memory and disk access. I would guess the cost of this is near zero on a server, since the server is always on and always drawing power. For the client computers, likewise it's a few seconds of thinking; almost nothing in the aggregate, and even if you could shrink the emails, these computers are still going to get used during those few seconds you've saved. (on preview, pretty much agree with adamrice.)

If you want to save energy, make sure people's computers are off at night. I saw a paper that did a night-time audit of several companies and they found that nearly half of all PC's were powered on overnight. Talk to your IT staff about this; several companies offer software to automatically power down computers at certain times of the day. Also, make sure the lighting and heating is just what you need and no more. The best way to reduce your footprint is to turn down the heater a few degrees.

By the way, I think there is a strong economic argument to make about email attachments. The email pipes have a limited capacity and can get clogged up by too many big attachments. Also, presumably your users have limited hard drive space, and the servers do too, especially if the servers keep a copy of the message in a redundant backup system, which is expensive.
posted by PercussivePaul at 11:12 AM on April 22

I'd say little to none. There is very little CPU usage for sending e-mails, its just network traffic and that overhead is mostly absorbed into the idle cost of running the e-mail client.

I tested and sent a 1mb mail to myself - i have a my desktop plugged into a kill-a-watt. CPU usage went from 4% to 6% for ~2 sec and there was no change in the amount of power indicated by the kill-a-watt. The CPU usage & time was similar for sending a 1k e-mail.
posted by wongcorgi at 11:20 AM on April 22

I don't think that the client is going to see a large increase in energy use, but checking the energy use on the server for handling (say) 500x 1MB emails versus 500x 1KB emails might matter.

Then there's also the fact that if this happens often like the OP says, it's possible that such a beefy mail server isn't needed, which would save money in both initial expense (though it's been spent now, it would be good for the future when the company is at capacity) and in electrical use.
posted by Brian Puccio at 9:18 PM on April 22