Energy Efficiency: Email vs Letter

[337]

Hi all,

From the title of this sub-forum I understand that this is the place to have a scientific discussion that can not possibly lead to any practical outcome, so I would like to present for discussion the following question :

When I want to contact a person at a distance of 1000 km from my location - which option requires less energy, sending an email or sending a letter ?

For the sake of simplicity I would like to start the "counter" at the consumer point : buying the paper, envelope and stamp, and turning the power on an existing PC connected to an existing network.

Please let me know your thoughts on this subject.

 

[infomax]

Hi all,

From the title of this sub-forum I understand that this is the place to have a scientific discussion that can not possibly lead to any practical outcome, so I would like to present for discussion the following question :

When I want to contact a person at a distance of 1000 km from my location - which option requires less energy, sending an email or sending a letter ?

For the sake of simplicity I would like to start the "counter" at the consumer point : buying the paper, envelope and stamp, and turning the power on an existing PC connected to an existing network.

Please let me know your thoughts on this subject.

 

[DaveC426913]

That's an excellent question.

I think where you're going to run into trouble is what you count "in" and what you count "out" of the system.

For example, email is efficient in part because it relies on an existing infrastructure (servers all over the world). This infrastructure is necessary for the email to find an efficient route to the recipient. The infrastructure spends a lot of time "on" (using energy) but idle. Do you count the energy usage of these machines?

If there are a million servers around the world waiting for an email to be routed to its recipient, but the email only visits ten of them, do you count the 10 or do you count the million?

 

[337]

If we decide to count the existing infra-structure in, that includes the servers being on while idle, but it should also include the infra-structure of the postal services : offices, trucks, employees...

For the sake of simplicity - let's count only the servers on the specific route of my email to my contact 1000 km away.

I also think it is an excellent question, and I'm positive that debating this issue is sure to consume way more energy than both options together...

 

[DaveC426913]

If we decide to count the existing infra-structure in, that includes the servers being on while idle, but it should also include the infra-structure of the postal services : offices, trucks, employees...

For the sake of simplicity - let's count only the servers on the specific route of my email to my contact 1000 km away.

I also think it is an excellent question, and I'm positive that debating this issue is sure to consume way more energy than both options together...

Yeah, but it's not that simple. The efficiency of email is based on there being a large number of servers available whether or not they're used.

What you'll have to do is determine, on averge, how many servers are used to transmit an average email, then hypothetically set up a network with that many servers (though still under load as if they're part of a world-wide network).

 

[russ_watters]

Why are you guys making this so complicated? The easiest way to think about it is to take the total cost to run the all the infrastructure and divide by the number of messages sent. Fortunately, we already have one of the answers: it costs $.44 to send a letter in the US.

I'd wager that it costs a fraction of a cent to send an email. Consider that commercial hosting companies don't charge per email and have very low hosting fees.

 

[337]

Both your posts confirm that the most difficult thing here is to find the starting point of the "energy counter"...

The infra structure is already existing in both cases - the network of idle servers, and the postal services. The cost of sending a letter is not only the stamp, it is also the envelope, the paper, the ink of the pen used - all the consumables involved. Then the email which requires no consumables, but a lot of cooling (possibly long term if the recipient decides to save it) which requires energy.

Do you think we should consider all the processes involved ?

 

[Vanadium 50]

This is a very ill-defined question.

One way of defining the question is how much it would cost to send the very first message. There it's clear that email is more expensive. Another way of defining it is the incremental cost to send just one more message. There it's clear that email is cheaper. Pick some metric intermediate between the two and your answer will be intermediate between the two.

 

[russ_watters]

I don't see how there is any question that the first method you suggest is irrelevant. The second is the proper one.

 

[337]

Ok, the infra structure already exists for both, so let's consider what you termed as "just one more". However, I would like to address the following points : paper is a consumable material, a server cooling system uses energy collected from a source, the choice to save the email or the letter.

 

[DaveC426913]

This is a very ill-defined question.

The point of the question is to define the question.

It's not so much about the answer, but what logic went into getting there from here.

 

[Vanadium 50]

Fair enough - but then that should shape the discussion, right?

 

[DaveC426913]

Fair enough - but then that should shape the discussion, right?

That's what's happening. We are defining the boundaries of the problem. It would be folly to proceed without first determining what it is we're measuring.

 

[DaveC426913]

Ok, the infra structure already exists for both, so let's consider what you termed as "just one more". However, I would like to address the following points : paper is a consumable material, a server cooling system uses energy collected from a source, the choice to save the email or the letter.

Hm. The more I think about it, the more I'm thinking Russ has his finger on it.

Rather than trying to determine the cost from the ground up, we could acknowledge that a tremndous amount of work has gone into that already, in the form of those who are making a business of it.

If we deduct a few cents for profit, a letter costs 44c or so, plus consumables, which is maybe a penny.

Email, on the other hand - as witnessed by the amount of spam out there - is free. All of its expense is infrastructure.

The point here, is that, as soon as we grant that the infrastructure doesn't count toward the total, it simply becomes a matter of the actual outlay of costs for the single letter at-hand.

 

[337]

Ok, this answers for the financial costs of emails and letters over existing infra structure. But what about the energy ? Both travel to their destination, the letter by truck and the email through optic fibers. Both options consume energy...

Let's compare the energy consumption :)

 

[DaveC426913]

Ok, this answers for the financial costs of emails and letters over existing infra structure. But what about the energy ? Both travel to their destination, the letter by truck and the email through optic fibers. Both options consume energy...

Let's compare the energy consumption :)

Same thing. We're back to infrastructure.
If you discount infrastructure, the energy cost of a single snailmail versus email is nothing more than the actual energy expended in the writer writing it and sending it.

An email requires some typing and the press of a Send button.
A snailmail requires either handwriting or typing and printing, getting an envelope, writing on it, getting a stamp and affixing it, walking to the mailbox and depositing it.

 

[337]

When it comes to what a person has to do - obviously email is much easier (especially for those with an unreadable hand-writing).

What about the energy consumed by delivery from source to destination over an existing infra-structure ?

I think the email requires nothing more than cooling of the servers it goes through, while a letter has to be delivered by truck from one hub to the other - fuel against electricity... ?

 

[DaveC426913]

When it comes to what a person has to do - obviously email is much easier (especially for those with an unreadable hand-writing).

What about the energy consumed by delivery from source to destination over an existing infra-structure ?

I think the email requires nothing more than cooling of the servers it goes through, while a letter has to be delivered by truck from one hub to the other - fuel against electricity... ?

Well:
- servers run 24/7. Trucks do not.
- servers are running even if they are not used. Many, many of them.
- etc.

 

[Naty1]

You guys have apparently never worked in financial positions where cost allocation occurs. Forget cost; this question was energy. (The idea that because a consumer is charged $.44 for a stamp by the incompetent US government has darn little to do with the actual COST to pickup and deliver that letter. It has do do with government subsidies, government style cost accounting, taxes, politics, overhead and even if all that muck could be waded thru, ENERGY is only a piece of that mess. Government infrastructure, for example, pays no federal, no state,no local taxes, no fuel taxes...you know what that tally is in total?..probably 50% depending on what's assumed in any study. A plausible way an economist or financial person would look at costs is on an incremental basis: how much does the next letter or the next e-mail cost. And that does NOT circumvent all the stated issues. )

This original question IS a good one and is typical of green calculations that usually miss 2/3 of the relevant costs. It would take a year to arrive at a decent answer...then another five years to argue about the validity of the results.

 

[337]

You guys have apparently never worked in financial positions where cost allocation occurs. Forget cost; this question was energy. (The idea that because a consumer is charged $.44 for a stamp by the incompetent US government has darn little to do with the actual COST to pickup and deliver that letter. It has do do with government subsidies, government style cost accounting, taxes, politics, overhead and even if all that muck could be waded thru, ENERGY is only a piece of that mess. Government infrastructure, for example, pays no federal, no state,no local taxes, no fuel taxes...you know what that tally is in total?..probably 50% depending on what's assumed in any study. A plausible way an economist or financial person would look at costs is on an incremental basis: how much does the next letter or the next e-mail cost. And that does NOT circumvent all the stated issues. )

This original question IS a good one and is typical of green calculations that usually miss 2/3 of the relevant costs. It would take a year to arrive at a decent answer...then another five years to argue about the validity of the results.

LOL correct on all counts - and here in Europe probably 10 times more so... This is only an informative discussion in an attempt to establish the best way to "work clean" (nothing to do with the big hoax of global warming though...).

As for the energy balance of both options - please give your view.

 

[dylan123]

really it should cost the same...

 

[337]

really it should cost the same...

If there were any charges per email, it would probably help reduce overall costs of hosting and help reduce spam...

 

[russ_watters]

An article some may find interesting:

Greening the Internet: How much CO2 does this article produce?

Twenty milligrams; that's the average amount of carbon emissions generated from the time it took you to read the first two words of this article.


How green is your website? Calculating all the factors involved in a website can be tricky.

Now, depending on how quickly you read, around 80, perhaps even 100 milligrams of C02 have been released. And in the several minutes it will take you to get to the end of this story, the number of milligrams of greenhouse gas emitted could be several thousand, if not more.

Further down:

"Most people don't appreciate that the computer on your desk is contributing to global warming and that if its electricity comes from a coal power plant it produces as much C02 as a sports utility vehicle," said Bill St. Arnaud of Canarie, a Canada-based internet development organization.

I don't know who that guy is, but that claim didn't pass the smell test for me, so here are my calculations:

Good quality coal has an energy density of 30 MJ/kG. A high end desktop might use 500W while a laptop might use 50. A power plant is about 1/3 efficient. So each kG of coal burned can power a computer for 5-55 hours depending on the type. Coal produces something like 3 times the CO2 per kg of coal, so that's 1.85 to 18.5 hours per kG of CO2.

A big SUV might get 15 mpg or about 2.9 mi/kG of gas. Gas probably only produces a tenth the CO2 that coal does when burned(not exactly sure), producing more water than CO2, so that would be 9.7 mi/kG of CO2.

So if you don't drive your SUV much and you use a high end desktop for a few hours a day, those numbers might be reasonable. But for most people, no, using a computer is not like driving an SUV.

 

[mheslep]

You guys have apparently never worked in financial positions where cost allocation occurs. Forget cost; this question was energy. (The idea that because a consumer is charged $.44 for a stamp by the incompetent US government has darn little to do with the actual COST to pickup and deliver that letter. .

Actually it's worse than this. The delivery by mail costs more than $.44 per, as the post office continues to lose money every year - about a $billion or so per year now.

 

[maverick_starstrider]

I'm afraid I don't really follow the issue with requiring multiple servers. There are multiple servers to handle the world's entire volume of e-mail not a single send. If one was to include the construction and maintenance costs of all the servers then one should also include the construction and maintenance costs of all the airports in the world, even those not used to deliver the particular letter in question.

To me it seems that e-mail uses considerably less energy. If one just grossly simplifies the problem to an e-mail being a data transmission along a wire for about 1000km and a letter being truck to airplane to truck the letter clearly requires a ton more energy to deliver. The energy required to send a signal 1000km (assuming appropriate servers/boosting stations along the way) is much less then the amount of energy wasted in fuel for a truck-airplane-truck transversal of 1000km (and if someone questions the non-inclusion of the cost of boosting stations I'd say that's akin to not including the cost of gas stations and airport fuel depots).

 

[DaveC426913]

I'm afraid I don't really follow the issue with requiring multiple servers. There are multiple servers to handle the world's entire volume of e-mail not a single send. If one was to include the construction and maintenance costs of all the servers then one should also include the construction and maintenance costs of all the airports in the world, even those not used to deliver the particular letter in question.

Exactly. So, the key factor becomes: do you include the infrastructure or not?

 

[russ_watters]

You should probably include the fraction of the infrastructure dedicated to the service.

 

[maverick_starstrider]

Exactly. So, the key factor becomes: do you include the infrastructure or not?

Even if you do I'd assume airports, fuel stations, post offices, etc are for more energetically wasteful then the cost of the average number of servers through which an e-mail passes.

 

[DaveC426913]

You should probably include the fraction of the infrastructure dedicated to the service.

Yes, my key argument is that - to ensure efficient service - often the infrastructure is built such that a portion of it sits idle. Even though it is not directly in use, it is arguably required to provide the efficiency and reliability for any given single piece of mail.

For example: idling multiple servers or mutliple delivery trucks so that there is one always available, even during a peak traffic period. Does a fraction of that count toward the single email?

 

[russ_watters]

There's an error in here, corrected and highlighted:

...here are my calculations:

Good quality coal has an energy density of 30 MJ/kG. A high end desktop might use 500W while a laptop might use 50. A power plant is about 1/3 efficient. So each kG of coal burned can power a computer for 5-55 hours depending on the type. Coal produces something like 3 times the CO2 per kg of coal, so that's 1.85 to 18.5 hours per kG of CO2.

A big SUV might get 15 mpg or about 2.9 mi/kG of gas. Gas probably only produces 3/4 as much CO2 as coal per KG, producing more water than CO2, so that would be about 1.3 mi/kG of CO2.

So if you don't drive your SUV much and you use a high end desktop for a few hours a day, those numbers might be reasonable. But for most people, no, using a computer is not like driving an SUV.