Green software is greenwashing

Decarbonizing the economy is tough. But I reckon the main reason we'd fail to meet our climate targets is because we can't distinguish what actually helps and what looks like it will help. Added to that, someone making false claims can misdirect the money (and the good will) to make a profit.

Since the start of COVID, I have been in conversations where out of the blue the subject of our "digital environmental footprint" comes up. During COVID I heard my friends feeling guilty about their video calls and now, in the AI era, I hear people around me feeling bad about using a chatbot.

When these topics come up, I am often surprised by how everyone in the group nods, as though they are indisputable facts. Without doubt, tech has a big environmental footprint, but (and this is without sounding conspiratorial) the topic is a red herring.

Greenwashing and how to be good at it

Greenwashing is often known as advertising bogus environmental benefits for a product. Take a step back and you realize it's not just about each product in isolation. The feature of the carbon markets is that they are interconnected, i.e. when a decarbonization path is easy, capital flows to it. This amplifies the damage greenwashing causes. On top of introducing unreliable emission reductions, it takes away the attention from actual areas of potential.

If all these sound attractive to you, here are a few hallmarks that I suggest you include in your own greenwashing campaigns:

1. Find a sector that's already used to certification regimes.
2. Pick a well-reported environmental issue and tether yourself to it.
3. Make everyone feel good by boasting improvements even though you had no role in them.
4. Shift the blame to the consumer for anything else.

Let's see why each of these give a boon to your campaign.

Sector selection

There is a lot of money in creating certifications and standards. From the access to the standard, to the ecosystem of the compliance consultants that they create, to the auditing and renewal of the certificates. Companies pay a lot of money to get certain security or supply chain certificates.

Enter the field of software. Just go to the Google Cloud's compliance page to see over 175 certifications, attestations, frameworks, and laws that they comply with. Not only that, Google Cloud itself offers career certifications for professionals who put stuff on Google Cloud.

This is the perfect sector for a new certification. It's much easier to convince Google Cloud that they need a 176th certificate than to go to an unregulated sector.

Visibility

There are many news articles sounding the alarm about datacenters popping up everywhere. They rightly point out that such rapid changes clash with the bureaucratic commissions in charge of electricity and water usage. In many places, outdated regulation means that these datacenters are getting very good deals for their usage while the residents of the community see their fees rise (update: on top of all these, there is interesting discussion about infrasound being emitted from the datacenters).

These issues have every reason to be in the spotlight; however, these are local issues. In a broader context, the datacenters could have environmental benefits. The emissions caused by a professional video calling someone in Europe are nil compared to taking a transatlantic flight for that meeting.

The nuance between the local detriment and the broad offsetting effect is often lost. The reader of the news articles concludes that the datacenters are bad for the environment. The broadness of this conclusion leaves the door wide open for greenwashing. Resolving the local issues is extremely difficult; solving a non-existent problem and touting its success a "datacenters are green now" is so much easier.

Automatic improvement

So far we found a massive sector and favourable public sentiment. That was the easy part; the hard part is being able to improve on what we are targeting.

Here is where the magical always-improving nature of tech comes to the rescue. There is no need to encourage any efficiency increase when those happen on their own. Moore's Law says that consumer tech roughly doubles its spec every two years. The battery technology has not seen a leapfrog innovation in many years but the battery lives of laptops and phones keep increasing when tech companies manage to fit more transistors in the same space and use them more effectively.

No environmental cause prompted such improvements -- it was only competition and market forces over many decades. A savvy greenwashing campaign can take credit regardless. Simply compare a nanometer-scale transistor today to a vacuum tube of the 1950s and claim the environmental victory.

Shifting the blame

Since the invention of the automobile, car companies have done a lot of work to encourage highway building while squandering modes of public transportation (like buying tram companies and shutting them down). To their success, most places in North America don't have any viable mode of transportation other than the car; however, it's the consumer that's expected to use more sustainable modes of transportation.

Car companies and the public sector avoid the blame by framing the use of cars as a personal choice. North American cars have trended towards being more fuel-hungry, larger, heavier, and more lethal to the pedestrians. These all get explained away as market choices.

Software is in the same boat. With most public and private services being online-first, contactless payments becoming the norm, and the ubiquity of the smart phones, being always online is a fact of life. For anything that's not an automatic improvement, like the aforementioned local harms from the datacenters, the consumer is to be blamed.

Comparing to home retrofits

Another thing that cloud providers comply with is reporting their emissions all the way to Scope 3. So when a Properate client asked about our cloud computing emissions, I didn't have to look too far.

Having many years of experience in the environmental field, I knew just reporting a number would be too vague. Instead, I put it in context. The tools we have on the cloud are for advising people on home energy retrofits. Let's consider one user who follows the advice. Are we saving enough emissions with their action to justify the emissions our tool has caused? If so, by how much? Or, in other words, how long does it take for our operations to emit as much carbon as that retrofit has saved?

We first looked at our operational emissions. Emails, video calls, online collaboration, etc. Turns out, it would take 600 years for our operations to emit as much carbon. Including all our cloud activity such as running data pipelines, content delivery, and compute cores, one retrofit still offsets those emissions for roughly 34 years.

Let me reiterate. This is not a city or a neighbourhood -- it's just one home. A house with retrofit potential is emitting so much excess carbon that it won't even compare with the emissions from software. Same goes for AI tools. The emissions caused by getting an AI response are negligible compared to what heating and cooling a home would emit.

It is unfortunate that people are nudged to feel bad about their internet usage, and yet the elephant in the room is the room itself.