Fake CO2

My guilty pleasure is listening to DLF Radio - and DLF’s guilty pleasure is inviting people who come up with fake CO2 emissions figures.

My absolute favorite: Emails. You can see these fake emission numbers in business signatures as well (“think before replying…each email emits 100g CO2”).

It’s absolutely insane that these numbers get published unchecked in scientific radio formats like DLF. For example, last November they promoted a “CO2 Diary App” that claimed: 7 received emails with attachments = 350g CO2 = 50g CO2 per email.

Is this true? Absolutely not. Anyone who has worked with email servers would never think that sending 7 emails requires a whole kWh of energy.

It’s really simple to get a rough estimate showing this is way off. But first, let me be clear: these circulating numbers are made up. Just like the total number of emails sent per day. If you multiply the fabricated 50g CO2 per email by the fabricated total number of emails, you quickly realize that emails may not be emitting 25% of global CO2.

Nevertheless, let’s check if these numbers are way off or not:

• We rent one server in the US for around €4 per month at Hetzner (2 cores, 4GB RAM). We rent the same server in the EU to receive the emails.
• This can easily send 500 emails with small attachments per minute - around 750,000 per day.
• Yes, there’s plenty of internet infrastructure between them, but access to this infrastructure is included in the €4 monthly cost (Hetzner doesn’t operate at a loss).
• So for around €8 total, we can send approximately 22 million emails per month.

Using the claimed 50g CO2 per email, 22,000,000 emails would result in 1,100 tonnes of CO2 emissions.

How much energy would produce 1,100 tonnes of CO2 on the German grid?

Assuming a dirty German grid (though server farms usually use renewable energy) with 360g/kWh, you would need around 3 million kWh! German energy is expensive, but even at 10 cents per kWh, that would cost roughly €300,000 in energy alone. Even if you think these 1,100 tonnes are solely for producing the hardware, try to buy the coal you need to burn to generate 1,100 tonnes of CO2 for €8.

Here’s my FAQ summary:

FAQ:

• Can you get €300,000 worth of energy at Hetzner for €8?
  No.
• Are emails likely to emit hundredths of a gram - easily 10,000 times less than usual estimates?
  Yes.
• So switching from E-Mail to Whatsapp will not help reach Paris climate targets?
  No.
• So it’s not more climate-friendly to drive your documents around town in a diesel car?
  No.
• Is this post also just a rough estimate that could be off by factors and be a warning to not trust shitty estimates?
  Yes.
• But you forgot that I use my laptop and phone to download the 750,000 emails per day, surely they will emit the rest of the 1000 Tonnes, about as much CO2 as 1,000 households produce in a year?
  No.
• Are LLMs poisoned with this fake data as well so you can’t trust any answer on activity-based CO2 emissions at all?
  Yes.
• Are CO2 emissions counted several times when we divide emissions again and again to all kinds of IT activities?
  Great question, if you divide all worldwide IT emissions through the 3 posts per month on my blog you really start to get a feeling of the dramatic climate effect of my personal internet usage.
• But at least the 15,000 liters of wasted drinking water per kg of beef is accurate, right - not just rainwater that would fall on that pasture anyway?
  I have bad news for you.
• Will spreading guilt & fakenews about everyday activities hurt or benefit the climate change discourse?
  See above.