What is the carbon footprint of a torrent?
March 8, 2009 9:36 AM   Subscribe

What is the carbon footprint of a torrent? (or range thereof, depending on torrent size)
posted by oonh to Grab Bag (7 answers total)
Wouldn't this be nearly impossible to [accurately] calculate, considering pieces of files may come from any number of places, all over the world, at any given time? Depending on the file, its popularity in different locations, and a number of other things, the results could vary significantly, I think.
posted by metalheart at 9:43 AM on March 8, 2009

You could make a good argument that it is zero. The incremental energy used to actually send data over a link is negligible, relative to the power required to have the link live in the first place. As long as the torrent is being sent over a path that would be live anyway, and most are, it wouldn't use much of anything.

OTOH, you could argue that a lot of computers are left on all the time, where they wouldn't be otherwise, just to keep torrents going. But it's virtually impossible to relate that to a single torrent download, especially since most of the computers may have dozens of simultaneous torrents available, and if you weren't downloading one, someone else would be downloading another.

You could also argue that torrents enable people to use much more of their allotted bandwidth, which increases the overall traffic passing through the ISPs, which requires more hardware like routers to be turned on, which requires more power. Again impossible to connect to a single torrent download.

But in truth, the biggest carbon footprint of a torrent is due to the activities by those who are trying to shut them down. Or those defending the Pirate Bay. People doing stuff uses a lot more carbon than computers doing stuff.
posted by smackfu at 10:26 AM on March 8, 2009

The problem with calculating something like this is that torrents are just so variable that, like metalheart and smackfu say, it's nearly impossible to get a fix on this. If you wanted to come up with some sort of formula to approximate this, you'd want to figure out:
  • size of the torrent.
  • how many nodes are seeding the torrent.
  • how many nodes have leeched the torrent.
  • average bandwidth of a node, both up and down (so you can figure out how long it takes a node to seed the torrent as well as how long it takes to leech; this is probably going to be the most complicated part).
  • related to above, the average seed:leech ratio, so you can figure out how much time each node spends seeding and each node spends leeching.
  • some sort of an assumption on the average electricity use per-node, taking into account that this can vary wildly depending on whether the node is somebody's laptop or a server in an actively cooled and staffed datacenter somewhere.
  • some sort of assumption of how much energy is used by routers and such in transmitting the data over the Internet, if this is even possible to find.
And it goes on. You're going to have to make assumptions about this: do you include the entire carbon footprint of the manufacture of the computer hardware? Probably not, but how much of that should you consider? How about the networking gear? It's a complicated problem, so however you do choose to calculate this you're going to want to be upfront about your assumptions and method.
posted by kdar at 10:33 AM on March 8, 2009

There's also the question of how the electricity is generated. If all the nodes are powered by hydroelectricity, the carbon footprint will be considerably less than if they're powered by coal.
posted by Sys Rq at 12:17 PM on March 8, 2009

At the lower bounds the answer is: negligible. A great many torrenting computers are on for other purposes, as are all of the network connections between them.

The top range is, as others have said, impossible to calculate. But it's Sunday, so what the hell. I'll at least run through the big numbers so you can fudge the math in whatever way makes you feel better (if you're pro or anti pirating).

Lets say we pirate a DVD and the torrent takes an hour to run. (I'm pretty sure the amount of data doesn't change the calculations much if at all, all we're concerned about is the time the computers are left running.)

Over this hour we have, say 99* other computers connected for a total of 100 computers on for one hour.

Computers electricity use, say between 60 and 250 watts when on. Bit torrenting isn't high demand, so lets say that between the low power laptops on the network and the idling computers they pull an average of 120 watts each.*

However many of these computers are on for other reasons besides file sharing, so lets break it down. (math= # of computers * watts per hour * % of use the torrent takes up)

10* of the computers are used for bit torrent exclusively. 10*120*1.0 = 1200 watts.
60* are used 50% for torrenting, half for unrelated stuff. 60*120*0.5 = 3600 watts.
The rest (30*) are multitasking crasy and only 10% torrenty. 30*120*0.1 = 360 watts.

Total: 5,1600 watts = 5.16 kilowatts per hour of torrenting.

(I plugged in some different, but also believable numbers and got a range from less than 1kwh to 18 kwh with the number of connected computers being the big variable.)

I'm not going to count the power used by the network because a) I can't find any useful numbers, and b) they're all on anyway and the torrent traffic you're adding is usually small.

I'm also not adding power use for thinks like monitors (since you don't need a monitor for torrents to work) or wireless routers or external hard drives. Add these in if you think they're important.

So, now how is that power being generated? Well those computers could be (And probably are) anywhere in the world. Wikipedia tells me that in general world power generation is 37% oil, 25% coal, and 23% natural gas and the rest misc. So we throw out the misc, and look at the footprint of these three sources.

Oil: 900 grams/kwh. 900*0.37 = 333 g/kwh
Coal: 950 g/kwh. 950 * 0.25 = 237.5 g/kwh
Gas: 600 g/kwh. 600 * 0.23 = 138 g/kwh

Total: 708.5 g/kwh.
(These numbers came from here. However if all the computers were in, say, Japan, it would run you around 500 g/kwh, and in China it would get you over 1000g/kwh.)

So we multiply the two numbers together, 5.16 kilowatt hours * 708.5 grams per kilowatt hour and we get 3.65 kilograms of carbon produced for one hour of torrent. Putting in different but possible numbers above gets from 18 kg to 0.2 kg per hour.

So there you go. Go forth and warp these numbers to support whatever you're trying to prove. Either it's destroying the planet or it's a greener alternative to buying DVDs.

* made up number.
posted by Ookseer at 2:29 PM on March 8, 2009

Whoops, forgot a big modifier. Many of those computers are serving more than one torrent. So divide the answer above by how many torrents the average computer is serving. A reasonable number is 5, so divide my numbers by 5, which give you:

730 grams per hour with a range of:

- Worst case, dirty and everyone only has two active torrents: 9000 grams/hr (9kh/hr)
- Best case, clean and everyone has 8 active torrents: 0.25 grams per hour.
posted by Ookseer at 2:38 PM on March 8, 2009

All of these answers have ignored the material storage media, which has a much higher carbon footprint owing to manufacturing costs.

For small files and a small number of users, the answer is still "negligible". Once either variable increases the footprint increases dramatically.

A DVDr has a footprint of about one pound. For every torrent that serves a bootable ISO, we can safely say the number of seeds and completed leeches = lbs CO2. Perhaps minus some small percentage of users who complete the download but never burn the image.

That small percentage increases if the ISO isn't bootable: we can assume that between 5 and 20% are mounting them as images, so for application ISOs the carbon from plastic Discs is lower.

But that data too, has a cost, in the form of platter space. In fact the cost could arguably be higher. If 250 users are storing a 4gb torrent of, say, a Photoshop 10 Disc, or a DVD edition of Brazil, between them they have devoted a 1tb HDD to the the image.

It's not easy to find the total manufacturing footprint of a HDD, but the just the shipping costs of components to the assembly plants are estimated to be about 50 lbs per unit. Add to that component manufacture costs, assembly costs, unit distribution and packing costs, and the carbon price of a moderately popular .trorrent begins to tower.

All the carbon emitted by serving the .torrent is indeed negligible by comparison.
posted by clarknova at 11:58 AM on October 21, 2009

« Older Where to get local China news, in China   |   Ideas needed for back side of business cards Newer »
This thread is closed to new comments.