Greening Digital #1 - Amazon's Climate Pledge and Eco-costs
The one written on a train from Paris to London, after the 1st climate strike day
I’m writing this on the Eurostar from Paris to London, as part of my vaguely comical rail oddysey, trying out the use of interrail passes for business travel. So, hello from a train!
Here’s the menu for today:
1.1 Analysis of Amazon’s new climate commitments, and a correction
1.2 How you might respond if you want to do something similar
1.3 Link Grab Bag - developments that caught my attention, and are worth monitoring in future.
1.4 A request for notes, and clarifying that it’s okay to share this letter
A correction and some analysis of Amazon’s new climate commitments
Amazon released a new sustainability website, just before the big climate strikes. There’s a lot of interesting stuff in it.
In particular, the total carbon footprint of Amazon was released for 2018.
It’s about 44 million metric tonnes. That’s hard for most of us to get our head around, but the mental reference point I use is “about the same as Hong Kong, or Finland“.
That’s roughly 20 million tonnes more than Apple’s footprint for it’s products, at 25 million tonnes in 2018. I use "about the same as Mongolia".
Google’s carbon footprint by comparison, before they offset the last unavoidable emissions, was, according to their own report, was about 1.2m tonnes. If you prefer, “pretty close to Liberia”.
What I find more interesting though is, what Amazon is doing to quantify this figure, and how they’re set up to make carbon reductions.
The methodology is outlined here:
Our approach to quantifying our carbon footprint reflects the complexity of our business.
Our team of researchers and scientists have combined cutting-edge life cycle assessment (LCA) science and Amazon Web Services (AWS) big data technology to develop a robust software solution that processes billions of operational and financial records from Amazon’s operations across the world to calculate our carbon footprint.
The software estimates carbon emissions for all activities within our system boundary using a dollar-based environmental assessment model, then enhances the accuracy of carbon-intensive activities with detailed, process-based LCA models.
They give quite a detailed methodology in a PDF linked from the new sustainability site, explaining how they do it.
And in a promising note, they now have independent audit of their figures and verification of their approach, just like Apple. They even have the same guy called Trevor signing off on the reports! They now pass 3 of the 4 criteria I listed in the last issue. I was wrong about this in the last issue.
Note: after looking into the numbers a bit more and asking People Who Know, it looks like Amazon is doing something a bit tricky on reporting with their carbon numbers for electricity use.
Explaining it is pretty technical, but the short version is that they’re doing the equivalent to the "we pay the legally required amount of tax" line large companies, use, but for emissions.
They might meet the requirements on reporting, but it doesn't really follow the spirit of the reporting. Climate emergency, remember?
If there's interest, reply to this email, and I'll explain it in a future edition. It’ll take a few paras to explain, and this already a long newsletter.
Tracking environmental impact - top down versus bottom up
If you want to work out the carbon footprint of an activity or thing, you generally want to look at it’s entire life cycle. Indeed there’s a whole discipline around life cycle analysis (LCA), for this, and generally speaking there are two main approaches.
One approach is the top down approach. You take the total emissions figures for an country, or organisation, and economic output for a country, adjust for flows of material and capital in and out, to work out the carbon intensity of different sectors per unit of money (i.e. how much carbon is emitted per unit of money).
So the steel sector, which uses more energy, will have a higher carbon intensity per dollar/euro/pound than services sector, which uses less.
Then, based on the amount you’re spending in that sector, you can arrive at a carbon figure by multiplying the money amount by the carbon intensity.
This has the benefit, of being in wide use anyway, and is fast, at the expense of accuracy. More specifically, your accuracy is determined by:
how well you work out a carbon intensity figure per sector, and assuming you have spending data…
how well the categorisation of your spending matches the categorisation of this per sector
This is pretty close to the approach that CO2 Analytics use, and what I understand Normative, and Ecochain in Europe to use too.
The other approach is to go bottom up. If you're looking at a product, you look at every component that goes into it, and for each component, work out the carbon emissions that result from making it.
To do this, you need to know the carbon emissions for all the materials you're using, along with how emissions intensive the different processes you use are.
This is more accurate, but it's very expensive. You need specialised software, datasets that are often proprietary, and expensive specialists. We're talking 5 or 6 figures per product easily.
If you're doing it for a company, there is guidance on how to do this too.
The guidance largely says measure every activity you do, then apply some kind of conversation factor (we call them emissions factors) for all the energy and materials used, to get CO2 figures for all of them.
Again, this is accurate, but expensive and tedious.
So: top down is fast, and cheap, but can be imprecise, and bottom up is usually more precise, but often slow and expensive, and needs to be done for every product you might make.
Amazon's approach appears to be proprietary, but they do at least link to a bunch of academic literature in their report. This is good for a number of reasons, and as far as I can tell, their approach could be summarised as:
combine use of appropriate measurement techniques, to make the most of the resources they have to inform decisions
use a single synthetic, "full cost" monetary value to inform decisions for working out where to make meaningful reductions for the cash being spent
provide free hosting for sustainability data, so sustainability wonks flock there, and use this proximity to a community to maintain a lead over other companies in this field, by watching which datasets are used, and what practices emerge. This is looks a lot like the ILC strategy that i wrote about in this piece about Microsoft buying github in 2018.
What affect this might have
We already have binding legal carbon reduction targets in the UK. It looks like we'll have them soon across all of Europe too.
This is good, but I think there's a chance this push for sustainability might end up having some similar effects on the tech industry as GDPR did.
The introduction of GDPR had good intentions, but in many ways helped entrench existing, well resourced companies who could afford to sort out compliance early.
Smaller players (who - let’s be fair - often weren’t being careful enough with data anyway), had a hard time dealing with GDPR, and the bigger ones were able to make hay while much of their time was spent understanding how to comply.
I think the same might happen with climate and sustainability in tech.
On the first read, I am impressed by Amazon's approach for quantifying emissions, and sufficiently underwhelmed by the capacity of the rest of the industry to do this, that if climate is finally taken seriously, I think it might entrench their dominance in the sector even further.
1.2 Doing something like this if you're not Amazon
That said, you don’t need to be Amazon to be able to use this approach of combining top down methods fora high level overview and bottom up models for specific processes.
In fact, I was going to suggest exploring it as a strategy in a workshop later this month with a new client, to start out with working out the emissions for delivering a digital service. Seeing a trillion dollar company essentially say the same thing was some welcome validation.
How it might work
The input output data for the top down figures is published by public sector sources in many cases. Here’s EXIOBASE, a common, publicly available database of the kind of source data used by Amazon’s own top down metrics. The industrial ecology community uses it a lot.
The bottom-up data figures are also published, often by governments. Here’s DEFRA’s latest one for cars in the UK.
If you wanted that single dollar value thing, to help prioritise, you’d need some defensible approach, that ideally isn’t proprietary, and hard to challenge.
Last year, I think I found one.
I was introduced to it by Dr Isabel Ordonez, who told me about the work at TU Delft on EcoCosts.
EcoCosts is a similar idea to the single dollar amount mentioned by Amazon:
A synthetic number to take into account the environmental impact associated with different materials and activities, that reduces down to a single numerical value.
Here's the mental model I use for regular costs of a service. If the value people perceive is greater than the price they pay for it, people are happy. If your costs are lower than the price they pay, you’re happy too.
And here's the mental model I use for Eco-Costs. It’s designed to capture the other costs we have too. In the case below, the ecocosts are now higher than the price, so we’re much less happy.
It's typically used in the bottom up product design process, and there are curated datasets showing the ecocost for all kinds of materials, and even a smartphone app for quick and dirty LCA analysis to settle an argument in a pub.
Isn’t this reductive?
Yes, this approach is massively reductive. But you can make the argument that at least it's explicit, compared to the implicit approach we typically take when making these calls.
Right now, we typically incentivise people to make the financial numbers add up on delivery of a service above everything else, with any other costs as secondary.
For example, in the UK, we literally use cost of transaction as performance criteria in digital service standard.
I can imagine a scenario where you might use something like an EcoCosts value for top down, broad strokes analysis of most of your organisation's activities. You'd combine that with bottom-up figures measured in some EcoCosts-like unit, for the key things you do or make. This would at least give you a single dimension to work with in a similar manner to how cost of delay works, to help inform decisions reducing the emissions from running a service, or making a thing.
If you’re an industrial designers you’ll have tools like materials, energy, toxicity matrices to help to this, and they’re taught to undergrads. I’m not aware of similar things for service design though, and when I ask professional service designers, the ones I’ve asked haven’t come across any either.
Summary
We have the data and ideas already out there. But not in a nice package. I can imagine us shifting from cost per transaction to “some value to account for carbon” per transaction, in the light of binding legal targets on carbon emissions in the public sector.
The approach from Amazon is likely reductive, and any similar single number approach will be unavoidably lossy anyway, but it may be better than what lots of orgs use right now, and I think is worth exploring.
1.3 Link Grab Bag
I learn a huge amount in the ClimateAction.tech slack, and there's a good brain trust of nice, people building here. It’s free to join, and I'm biased as help organise it, but I'd recommend if this newsletter was interesting.
Amazon's Pledge. It's worth reading around the site, as there's more detail than I expected.
Gizmodo's take. It rightly calls out the 2040 deadline, 10 years after what Amazon Climate was calling for as weak sauce.
The Cleanweb London and Climate Action.tech meetup is happening this week in London.
There’s an unconference, OMGLIMATE, about tech and climate change happening on Oct 18th in London. Register interest here.
ShareAction in the UK released a web site this week for seeing how your pension props up fossil fuels. Fantastic copy: What World is Your Money Building?
1.4 Shouting into the void
How was this? Let me know reply to this email. I'm enjoying the feedback.
Also, do please forward this to others who might be interested.
There's an uptick in interest since last Friday, but there are still too few of us talking about what to do next, in the face of a changing climate.
Chris