Download this PDF slide deck (~ 60mb) | Download the video of this this talk (~55 mins, ~120mb) | Link to navigable, commentable linkable Miro board, with all the references | Visit the synchronised video transcript using Descript
Talk transcript
Below you’ll find the transcript from this talk delivered for DjangoCon Europe 2021. It has a few ‘ums’ and ‘ahs’ cleaned up, but for better or worse, it should be fairly close to the actual spoken words in the talk. It’s nearly 10,000 words, so use the find in page feature if you are looking for specific subject, or phrase. You can also use the descript link above, and highlighting the text will move the the video to the corresponding part.
How to be a Djangonaut in a climate emergency. Hi there. I’m Chris. I’m a director of the Green Web foundation, a non-profit set up to track the transition of the internet away from fossil fuels. And I’ve spoken at Djangocon conferences around climate and tech a fair few times now. This is my first remote keynote. And it’s worth knowing that there’s a full time synced transcript of this talk, at the link on the website you see here – just in case you need to pause me or what I said at one point, wasn’t too clear.
Did you know that the first ever mainstream carbon calculator was a Django project? It was made back in 2007 for the world wildlife fund. I learned this by doing some research for this talk and along the way, I realized that if you know where to look, there’s a hidden tradition of working on climate all through the Django community.
Here’s an example.
In addition to the first carbon footprint app being a Django app, the same year, the carbon account was created – an application designed as the next step you might follow once you’ve worked out your carbon footprint. In groups, you track the footprint of your activities and support each other as they do the same. And in some cases, take part in something a bit like what we might recognize a kind of tradable carbon offset scheme. This is the kind of stuff that today you might see you getting a load of funding and probably ends up using a blockchain somewhere. And this was back in 2007!
And what’s more, when I looked at who worked on this project, I saw some young looking versions of faces I recognize as being active members of the Django community. There was Simon Willison, one of the original co-creators of Django. And here is Andrew Godwin, who you might know from work on Django channels, on migrations, and most recently the new async features being built into Django.
Now climate is an absolutely massive domain, and if you ask them, I’m sure they’ll be able to tell you about a bunch of stuff they learned along the way and how complex some parts of this can get.
It’s not like climate as a subject is going to become less important in the coming years, so in the time that I have with you, I’m going to cover four main bases that you can see here. First, we’ll start with a brief primer, on the domain of climate, where we’ll spend about 10 to 15 minutes getting up to speed with what you need to know. And then we’ll use the same amount of time talking about the levers we have for change as technologists.
For the second half, I’ll share a mental model with you for designing climate friendly applications. And then finally we’ll wrap up with some specific, next things you can do as a Djangonaut.
Ready?
Off we go.
So normally when I do talks about climate and tech, it’s common for me to start with a bunch of scary statistics and photos to get across the idea that we really need to act now. If you’re watching this, I figure, you know that we need to act. But you might not know how many other people feel the same way now. There’s a large and growing body of governing bodies, like at the local and city level. And also at the national level, like the UK and other countries, and even at the supernational level, like the European union, who have now declared climate emergencies. This covers nearly a billion people. So if you think we need to be acting faster, you’re in good company.
And it’s worth remembering why we are in a climate emergency. I mean, it’s literally spelled out in front of you on this screen. And while as engineers, we tend to really like jumping straight into making innovative solutions to problems, it’s really, really important that when we talk about climate action, there is a need for deliberate decisions as well as innovation. And this is the decision we keep putting off- getting off fossil fuels.
So in high school, the chances are you were taught about the water cycle when explaining where rain come from. We know intuitively that things like rain or snow is the result of water evaporating somewhere in the world, then condensing in the sky and into clouds, then coming down as precipitation. Knowing that the water cycle exists is helpful for managing supplies of water.
But there are other cycles too, like the carbon cycle.
Anything that is, or was alive, has carbon in it. And it’s useful to know the carbon moves around the world in cycles too, albeit at a much, much slower pace. These white arrows demonstrate this or show this.
Now the green stuff here, that’s us. I’m made of carbon. You are made of carbon. Trees are made of carbon. When living things die and decompose, some of their carbon ends up in the atmosphere. The purple-y bit is the carbon in the oceans, fish are made of carbon and so are plants in the sea. Some of these when they die, they sink to the depths and eventually become sediment and form rock, forming that black bit in the bottom right. Up in the top in the blue, we have carbon in the atmosphere.
There’s a bunch of gasses, but it’s carbon dioxide that we care the most about. In addition to this, there’s a whole range of gas exchange, from plants, how they breathe and so on, but that’s not the only thing that puts carbon in the sky.
Around about 1850 as the industrial revolution picked up steam, we started using much more energy. And to meet this demand for energy, we started burning things to access that energy. We started mostly with wood from trees, but things really picked up when we got into fossil fuels depicted here as this bunch of red dots. All this energy has driven progress, and thismeans we now have wifi, i-phones and planes, and vaccines and conferences and the rest, but it’s also taken a bunch of carbon that was in the earth and put it somewhere else. So where’s it been going? Well, there’s a clue in the arrows on the screen.
Let’s fast forward to 2017. Over the last 200 years, we’ve taken a load of carbon that was trapped underground as those red dots. And by burning them, we’ve put them into circulation with the other black dots. Now, if you’re close to the screen, you’ll see the red dots are getting everywhere. But the key thing I want you to take notice of is that we have fewer red dots underground, and we’ve ended up with a bunch more than in the sky. That’s bad news for us and while I could talk about the heat waves and the forest fires and the flooding, and so on, I’m not sure I need to. If you’re watching this talk, you’ve probably already seen new stories around this stuff already.
If you want some more detail around this, or if you want to see the underlying data, I’d recommend checking out this website from the Potsdam Institute of Climate Research. This chart here tells pretty much the same story I just did, on a global scale, but if you visit the link in the Miro board, linked here, you’ll see that you can see the data for your specific country too. And there’s even a bunch of cool Python notebooks to explore.
Now, this is the data that goes into the big research reports that scientists make, when the IPCC – the intergovernmental panel on climate change – issues reports on how bad things are, and how we might avoid the huge amounts of mass death, economic damage, and general violence visited upon not just future generations, but people who are exposed to climate risks right now. And one recurring thing about these reports, and the data put forward by scientists is that every year we put off meaningful action, we make things harder for ourselves. Instead of making small changes over a longer time and starting early. When we delay, we build up more and more and more needed action. This chart was made in 2017 with the general direction of travel is about the same. Just like a code base where we put off paying down technical debt, we’ve done the same thing, but on a planetary scale.
Now the scale of there and rapidity of changes we need to avoid the worst impacts of the climate crisis is leading to a new kind of language being used by scientists when they tell us what we need to do.
Back in 2015 representatives from almost every single country in the world, met in Paris and agreed that we need to act meaningfully on climate. You might’ve heard of this as the Paris Agreement, and just like we have DjangoCon 2021, their conference is called COP 21.
It’s taken 21 goes for them to agree on something substantial, but one thing that came out of the conference was the decision to commission scientists to see what kind of changes we would need to make globally, to avoid this path of increasing disasters, mass death, and so on.
They released another massive report in 2018, and these three words were used in the message to policy makers, to describe the changes we now need to make in the face of what the delay that we’ve seen.
Rapid. Far-reaching. Unprecedented.
I’ve used the word delay a few times and it’s important. When we’re talking about climate. Predatory delay, is a term coined to describe the behavior we’ve seen by certain groups who do well out of the current system. And the best example I can think of would be action by a small number of fossil fuel companies, who have worked to spread fear, uncertainty and doubt so they can keep making money. To make this more concrete. I’m 40 years old next year. For longer than I’ve been alive companies like Exxon have had internal research that shows they understand the dangers of a changing climate, but have worked to introduce delay on this action by influencing the public discourse.
And in academic circles, these are now referred to as discourses of delay. Because they come in many forms. But now there’s a growing body of knowledge to document the strategies used to bring about this delay. If you look at some of the strategies here, you might recognize patterns in how we speak. When we talk about personal carbon footprint, for example, it’s worth knowing that it’s an example of redirecting responsibility on individuals. And it’s worth knowing that this was something that oil companies spent a lot of money on to popularize in the mid 2000’s as a way to avoid talking about more transformational change that was necessary.
In truth, these ideas are useful, but the presentation is somewhat dry and not everyone enjoys reading academic papers. So these ideas are also available, in a number of languages as these cool cartoons, with shareable images, that cover the same themes, but are much, much more memorable- it’s useful to be able to label these kinds of discourses when you hear them.
When we talk about the way we talk about climate change, it’s also useful to understand the demands initially made by Extinction Rebellion, a knew pressure group that rose to prominence in recent years. They had three very specific demands.
Tell the truth. So people understand how urgent the issue is.
Act now, because we still need to actually do something.
But the final one, go beyond politics, can be seen as a tacit admission that our public discourse is now too polluted by powerful figures to be fit for purpose. They demanded citizen assemblies would be set up to allow regular people to discuss the climate response we now need.
I think citizen assemblies are an interesting idea and that more people should know about them. They’re designed to mitigate against bias and power by using a random selection of a population, to meet and deliberate on a subject rather than having an appointed set of insiders. This group of people are then presented with evidence by a wide range of experts and advocacy groups, and given time to think about the subject and then finally make recommendations for our government to act on.
These recommendations about what kind of climate response we need tend to be more nuanced and well thought through the more you might have otherwise. And have historically gone much further than typical government policy, introducing ideas into the discourse that simply weren’t there before.
I’ve used the term climate response a few times now in this talk. And I use it to evoke the idea of response to an emergency. If we’re in a climate emergency, then talking about personal carbon footprints, won’t cut it. We need a meaningful response to the emergency, given how late in the day it is. It also shifts the framing -rather than talking about how small a person company or government footprint is. I think it’s better to talk about how big or ambitious their response is, because as the scientists say, we now need rapid far-reaching unprecedented changes.
Let’s try this with Fridays for the future movement.
Every Friday, because they’re too young to be running companies or nonprofits or parts of the government themselves, they organize and protest every Friday to remind the rest of us that we need to do something. So, this is already a pretty good climate response. But it’s useful to know that they’re not stopping there. They’re using every lever they have, like their lives depend on it.
If the science is spelling out the need for more ambitious climate response from governments, but governments are ignoring protests, there are other ways to achieve change.
If the rules are the system, we’re part of aren’t working, one thing you can do is work to change the rules of the system itself. Now, this here is a photo of Luisa Neubauer; one of the nine young people who brought a court case against the German government, for its lack of action last year.
And in April, the courts ruled in their favor! Changing the system is a pretty awesome climate response, in my opinion. The highest court in Germany, has compelled the German government to increase the scale of its climate ambitions.
And one week later after this this case, the German government came up with new commitments- it brought forward its targets to get to net zero and across the board is doing much, much more. Germany is the biggest carbon emitter in Europe, so this is a huge deal and it sets the precedent for future cases to be brought forward by the same kids who go to Fridays for the future protests in other parts of the world.
Seriously, the kids are going for the win here and I am here for it.
And if you follow climate news, you will know that it could be an emotionally exhausting pastime. But just like the German example, some days can be pretty special. And just last week was one that I want to share with you. In a single day, we saw a huge ruling in the Netherlands over Royal Dutch shell, who you might know as Shell Oil.
For the first time ever, courts pretty much said sudo cut emissions to a huge oil company, forcing them to nearly half of their emissions in 10 years. We didn’t know this was even possible with the law. And again, this sets a huge precedent.
Elsewhere on the other side of the world, in Australia, eight teenagers and an 86 year old nun have been able to make change, get a change in the law there too. That looks a little bit somewhat similar to what we see in Germany.
Even inside oil companies, you’re seeing change this same day, two of the oil, super majors, the massive oil companies, I have seen shareholder rebellions to get them to change course.
If this wasn’t enough May was special for another reason. At one level down the tech stack from us in the energy sector, there is an organization called the International Energy Agency.
It was initially set up by Henry Kissinger in the 1970s to ensure a steady supply of cheap crude oil, as a response to a number of oil producing countries in the middle east restricting how much oil they produce, during the oil crisis of that decade. It’s since become incredibly influential over the last 50 years. Countries and other large economies base their entire industrial strategy around the reports that it publishes. It’s really powerful.
And in May for the first time ever. They came up with a roadmap to get the entire energy sector on the same 1.5 degree pathway that scientists were saying would need rapid, unprecedented and far reaching changes to meet. This report is nothing like. what’s ever been published by them. For a start, and remember, this is an organization that was initially set up to ensure a steady supply of oil for rich countries. They literally say no more exploration for fossil fuels from now on. You can see it in the bottom left on. In this diagram. Then, but then they go on to say,
yeah, Not only do we have to get our fossil fuels, but almost all of the new energy we use has to come from wind and solar. And yeah, this has to happen really, really fast. But it’s totally doable. We just need to get on with it.
Okay. Let’s have a quick recap before we finish up.
climate emergency. It’s a thing. We can’t wait anymore.
Predatory delay. This has worked for longer than most of us have been alive, but it’s no longer as effective as it was.
Climate response. Our response to the climate emergency. We’re finally seeing mobilization on all levels of society. And it looks different in each case. But we need more and bigger responses.
Finally we need more and bigger climate responses because even though we’re starting to win, winning too slowly, now is the same as losing.
So that’s what’s happening. in the rest of the world. What about our sector? I mean, this is a Django conference after all. While this isn’t exhaustive, one way to look at categorizing the kinds of climate responses we might have in our work is below. You can look into the consumption, reducing the inputs we require to meet a specific need. A common example might be energy efficiency.
Or intensity. That’s reducing the harmful byproducts of meeting a particular need. The common example here might be using cleaner energy. Or direction working to accelerate the adoption of a specific activity, to push society and their given direction. An example here might be making it easier for people to take public transport or cycle. Instead of driving around in cities.
If we look for concrete, examples of consumption. A good example would be what we’ve seen in the tech sector itself for the last 10 years. Now, let’s look at this in a bit more detail because it’s often really poorly reported in the media. If you were to ask this question, Of the environmental footprint of the tech sector you’ll get a wide range of replies, depending on who you speak to.
And because most of us have a hard time picturing what a gigaton of carbon is. When I answer this question, I tend to talk in relative terms, comparing it to stuff you might’ve heard of. So for the purposes of this talk, you can think of the tech sector as having a carbon footprint of around 1 to 3% of global carbon emissions.
This is the same ballpark, as the entire shipping industry. Or the entire aviation industry. Or the carbon footprint of Germany or Canada. Or all of the combined emissions from all the commercial and heavy industry of Europe. So it’s not small. But it’s not the biggest thing in the world either.
Once you have an idea of the footprint. You might want to know what different parts of tech might actually have. I found this chart from Ericsson’s recent report on digital carbon footprints. Useful. As you can see in this chart. Making computers causes emissions because remember we need to melt sand before we can start turning it into microchips and unsurprisingly running computers uses loads of energy too, which causes further emissions.
This chart shows us that because networks and data centers are used 24 / 7, energy makes up a larger part of the footprint than gadgets. This means energy efficiency is a bigger lever there than it might be with user devices. In fact energy efficiency, has been something of a success story for computing.
Over the last 10 years, we’ve been able to benefit from a loads of valuable digital services. And its use has exploded. But when you look at the total energy use by data centers and internet infrastructure. It hasn’t really changed all that much. Imagine what this pandemic would have been like without all the digital services, we rely on, for example. It’s pretty good.
If we look in more detail at these numbers, we’ll see that this has been made possible by us, very much switching from a running code in relatively inefficient, smaller scale data centers with relatively low levels of utilization to running code in hyperscale cloud data centers, which can have 10 times as many applications using the same hardware. Also because they’re so much larger, they can afford to replace older inefficient kit more quickly, making better use of Moore’s law. This has also come at the cost of a much less diverse tech ecosystem though. But from an energy systems, point of view, things could be worse.
It’s worth noting though, that Moore’s law has been slowing down and increasingly gains will need to come from closer up at the top of the stack where we live, rather than being able to rely on changes further down the stack at the hardware level. This means that as a developer, understanding performance and efficiency is likely to become more important over time.
So that’s an example of consumption. How about intensity? Well, the best example I can think of, would be the simple fact that the internet is the biggest machine in the world, and it runs mostly on fossil fuels right now.
We might not see the impact directly, and when we talk about cloud computing, we tend not to think about burning coal or fracking. But there is avoidable harm being caused and it’s something we do have control over. To be fair, this is largely because our entire energy system still mostly runs on fossil fuels, as you can see here. But there there is avoidable harm being caused because burning fossil fuels not only accelerates climate change, but it also leads to loads of early deaths of literally millions of people around the world, from breathing in poor quality air.
Now this chart shows how it’s changed over the last 10 years. We are seeing changes and that green wage of wind and solar is growing. But remember, winning too slowly is the same as losing now. And remember how we spoke about climate response before?
This chart shows in a bit more detail, how we’d get away from this sad scenario. We can decouple the harm caused by the underlying energy systems from our use of digital services. To do this, most scenarios, point to wind and solar doing most of the work because it’s , because it’s the cheapest form of energy available now. Moreover, as we scale up the quantities that we buy, things like wind and solar in the cheaper they get, creating a kind of virtuous cycle. This virtuous cycle is something that tech firms have played an admirable role in. And is another success story.
What you see here is a chart from the IEA, the international energy agency, showing power purchase agreements, where companies agree in advance to pay for all the power generated by a wind farm or a solar farm, which means that it can actually get built. These agreements, are one of the most effective ways to get new wind and solar built. And the blue bits represent tech companies like Google and Microsoft and so on. So we’re doing pretty good here. Another reason that tech companies are piling into green energy, now, though, is because it’s increasingly being used as criteria for deciding who to buy from by decision makers.
Here’s an example from the UK government. When they released a new paper about greening their own use of technology. This guidance applies to the entire UK government and set some minimum standards for spending public money. And, yes, this is nerdy as hell, but I want to draw your attention to how explicit the standards are in appendix A, strategy deliverables. If you want to sell to public sector, you will increasingly you need to be able to tell a good story about your supply chain, and the use of renewable energy in it.
Now the UK is not alone in this and other countries are following suit as well.
There’s even more going on in terms of intensity here.
Last year, Google announced a new plan to run all of their infrastructure on carbon free I E fossil free energy by 2030. But to do this on a 24 / 7 hourly basis. Now renewable energy is getting cheaper incredibly quickly, but even taking that into account, the only realistic way that you can do this by 2030 is to make the entire grid greener for everyone. Because even a trillion dollar company, like Google can’t afford to do it by itself.
And, this is what you’re seeing, happen in real time now. We saw how kids basically said we can’t change physics of climate change. So let’s change the laws around climate instead. And now we have courts ordering oil and gas companies to halve their missions in 10 years. Now you see companies like Google saying we can’t change the physics of climate change, so let’s change the economics around climate change.
The us federal government is the biggest buyer of electricity in the world. And now they are going for the same 24 / 7 moonshot, that Google and others have announced, to basically make this cheaper by buying in scale.
There is quite a profound implication in this, that’s easy to miss. If you’re watching this in 2021, it’s almost certain that during your working life, you will see us achieve an entirely carbon free fossil free clean internet. The energy sector says it’s possible. The courts are saying it’s not only possible, but it’s necessary and ruling in favor of future generations. Now, the question is now when, and because speed is justice, I think this should be part of what drives us as technologists now. It’s common to talk about climate in terms related to wartime mobilization. But when I speak to other developers, I think Achieving Warp in the Star Trek sense might be a better way to frame this. Getting to a fossil free internet would indeed be a sign that we’re becoming a more advanced, humane society. And it’s going to happen. It’s just a matter of when, and that’s something that we can influence as technologists. Speed is justice, my friends.
The final lever. I want to draw your attention to its direction. And is related somewhat to this idea of when, not if. I
t’s common to talk about technology in terms of speed and the phrase move fast and break things is now an infamous term. I found the framing by Cade Diehm, of New Design Congress useful when thinking about what we work on as technologists. And his phrase has stuck with me since I heard it.
Technology is a social political and environmental accelerant.
You can think of an accelerant as something that speeds up a given reaction. You might use it to describe, say throwing a can of petrol onto an open fire. I think it’s a useful phrase because it emphasizes the harm that can be caused when used carelessly. And it’s speeding up, what’s already there.
This might be a stretch, but I talk about direction because an earlier Django con conference, I’ve heard developers respond to questions about tech ethics by saying, I am just a developer. I have a hard time with this, because I feel like it’s used where it can be used to avoid thinking about the role we play in society. And about what kind of society we’re working towards. If you’re a professional Django developer, you likely get paid multiple times of the median salary in your country and have a degree of professional mobility that other people in other fields only dream about. There are absolutely places you can choose to work, that accelerate society in one direction over another.
One example that I really like is the work by my society and what they’re doing with climate emergency groups in the UK. They’ve recently released a new service where they’re deliberately choosing to accelerate the process of civic engagement on climate, by the public with their local governments. They make it easier for people who were interested in seeing what the local government is doing and then compare it to other ones. what they do is make all the plans from all of the counselors searchable so that good ideas are really easy to surface. Now, this sounds simple, but there’s information was surprisingly hard to find before.
This compliments existing systems and levers for change and accelerates what’s already there.
I also use direction here to indicate that you might need to change direction in future. Based on what you learned along the way. Let’s say you work for a company that wants to reduce the carbon footprint of transport. And you’ve decided that a good way to do this, is remove the need to own a car. So make it just as easy to hand a ride as possible some way. You know that most cars spend the majority of their lives parked, so having fewer cars on the road sounds like a win.
And on a per trip basis, you might represent this new way of getting around like, so with this new footprint compared to the old footprint of getting around.
The old way might be us all buying cars so we can drive somewhere. And the new one might be fewer cars being around that we just hail. This sounds like a win, right? Now it’s important to understand that in many cases you can end up with a rebound effect for your new activity. If it’s much easier and convenient to book a ride, then we can see more usage in absolute terms, even if the new activity is more efficient.
This is not a new idea. It’s known as the rebound effect or sometimes known as Jevons paradox. Named after William Stanley Jevons who first wrote about this, in 1845 talking about coal powered steam engines.
This is pretty much what we’ve seen with ride hailing services like Uber and Lyft. When they first came out, they were seen as a greener alternative to car ownership and were used as examples of a sustainability success stories. What we are learning more now, though, is that ride hailing, because it’s so convenient, can end up displacing walking and public transport too, especially when providers are prepared to lose money on each ride so they can get market share.
Accelerating in this one direction might cause more harm than good. But it might be hard to course- correct, especially if you’ve promised VC style returns to investors.
The good news is, is that it doesn’t need to be this way. Some of these same ride hailing companies have invested in smaller green travel startups as a backup plan now. And it’s worth knowing that that we are entering a climate tech boom period. This means there are lots of competing visions of the future that you can choose to accelerate now. And as a technologist your skills are in demand.
So quick recap.
Consumption. Can I change how much we need?
Intensity. Can I change how much we pollute?
Direction, can I change where we are heading?
All of these are useful levers for climate response as a technologist.
So far, this has been pretty high level. I want to share with you now a mental model that you might use in your work. This is a work in progress, and it’s inspired by ideas that I’ve seen have some success in accessibility and inclusive design circles.
If you have ever worked with anyone interested in web accessibility, you might’ve come across the idea of POUR, as a way to remember qualities that make digital services accessible.
Perceivable covers content being perceivable by more than one sense. If I am blind. I can still perceive the content by hearing the words with the screen reader.
Operable covers being able to operate a site or application with more than one input, being able to use a keyboard as well as a mouse, for example.
Understandable, covers using language that an audience is able to understand, so they know what’s going on instead of having to wade through unintelligible jargon.
Robust covers working in the face of hostile conditions, even if some parts don’t work as expected. Working without JavaScript is a good example. If you’ve ever tried to use a react only website on a crappy phone connection without JavaScript. Not much fun.
So it might seem contrived. But as you can see, having something like POUR is useful because it makes talking about accessibility easier. Sure, you could talk about the W3C web content accessibility guidelines, but having something like POUR makes it easy to remember, and it makes it easy to talk about to both developers and to stakeholders.
It’s also useful when speaking to stakeholders, because you’re able to communicate the benefits of accessible design to product managers, and people who commission sites, so they know to include this, when they’re asking for a site to get built.
This means that you can turn something which was treated as a nice to have into something, that’s non-negotiable.
And we can see a good example of this in the UK government’s own website guidance now. Let’s look in more detail at these. Yes. We see POUR. We see perceivable operable, understandable and robust written in there.
Explicit guidance. And they also say: you might be breaking the law, if your website does not meet these requirements. Because government websites are supposed to serve the entire population of a country.
There’s obviously much more of a focus on accessibility in the public sector than the private sector. But this is starting to change. While we might agree here that making websites more inclusive and accessible is a good thing, typically, it’s been a nice to have rather than being a requirement.
But increasingly this is becoming a requirement and you can see cases like this, where having an inaccessible site means that you are bearing a risk of being sued, just like how Domino’s was.
So, inspired by POUR. I want to share GOLD with you.
It’s a mental model I use now to talk about desirable qualities, if you want to build digital services with sustainability in mind.
And as you can see, GOLD stands for Green, Open,lean and distributed.
And I’m sharing these principles, because they’re still a work in progress, but I think there’s enough that you might use when thinking through the changes you might make, if you wanted to green your stack.
Unsurprisingly green refers to green energy and using green materials. As we’ve seen before. When you think about this when we’re running computers, as well as when we make computers, but for this talk, we only really have time to talk about green energy. And as you’ll see, even that by itself is a pretty huge topic.
When we talk about green energy, it’s important to know that most of the time we’re talking about the servers that draw power from the electricity grid. Because you’re drawing power from the grid rather than say your own solar panels, you don’t have direct control over what others put into that grid. So even when companies say they run on green power, They are using electrons that might have come from fossil fuel sources. This might feel like cheating, but if you know how much of your power is coming from fossil fuels, you can at least account for this. And for the most part, this is how green power works.
Generally speaking. If you have put enough green energy into the grid to balance the fossil fuel energy you draw from the grid over a given time period, it’s considered green. This time period is usually one year, and this is considered green because over time, the grid becomes greener as renewable energy makes up a high proportion of the power that everyone uses.
It’s important that we are clear about the time period, because these figures are averages over a given time. When people say 30% of their power comes from wind, for example, it doesn’t mean they’re getting a solid of 30% all yeah. It fluctuates because the wind fluctuates just like how solar panels obviously generate less power when there’s no sun. This means that if we want to continuous, 24 / 7 green power, on and say, an hourly basis, it’s much harder to do.
By the way that always solar panels can generate electricity at night. And I shared a bonus link explaining how in Miro board, if you want to check it out,
Anyway, back to 24 / 7 green power.
Running only on green energy on an hourly basis isn’t impossible. If it was Google, wouldn’t have made such a big public bet on achieving it by 2030. But it does require a lot more ways to generate green energy, as well as ways to store green energy. You need more green energy so that as much of the time as possible, you know, you’re feeding in more green energy than you’re using than fossil energy than you’re using. And you need storage so that when you need more energy than you are generating, you can draw power from energy you’ve stored up elsewhere. Or previously during times of plenty.
In both cases, moving to hourly green energy gets loads of new green energy generation deployed like wind and solar and geothermal. In addition to being safer,. They keep getting cheaper, so this helps in the push to get fossil fuels off the grid.But it will take a while to get there, and in the meantime, if you want to run on the greenest energy possible, it’s worth knowing that even if a data center is using green energy on an annual basis, you can choose to run workloads in data centers that have different amounts of green energy matched on an hourly basis.
To make this more concrete. You can see this nifty region picker. That was created by some engineers inside Google. Unless you define criteria like cost, how green the power is you want, as in how much energy is matched on an hourly basis. And latency. Then it will tell you which regions are the best fit for your compute workload. And as you can see green, isn’t a binary thing. It’s something you can include in your own technology strategy when coming up with your own climate response.
Open in this model refers to an approach, not just open source. We’ll look at the use of open data as an example, but the general rule is that the more open data exists about the supply chains you use, the greater the scope for making digital services more sustainable. This applies to hardware too, but because energy use is the greatest source of emissions with servers right now, we’ll look at the energy example.
In Europe, data is published at an hourly resolution listing. How electricity all around, all the countries is generated. This is made available on the ENTSOE transparency platform that you can see here, and this data is published for anyone to use as they wish.
A good example of use of this data is electricity map. They take this data, apply a bunch of machine learning and other clever ideas to it, and make the results available over an API so that you can have a live feed of how green energy is, but also, so you can make predictions about how green it’s going to be, in the coming
hours. Feeds like this are really useful because they compliment projects like code carbon an open source tool you can use in Python projects to get an idea of how much energy, your code is likely to use in production. This was initially aimed at AI practitioners because machine learning can be such a massive energy hog, but because it’s python, it can pretty much instrument any Python project, including a Django one.
What’s really cool though, is that it will tell you not only the emissions of your job or what you’re using, but it will also show you what those emissions would have been, if you run the same job in different data centers around the world. Because it uses the data and APIs, like the ones we saw with electricity map.
The L in gold, it stands for lean. We’re in the digital realm and because pretty much anything that uses electricity also emits carbon, it means that we really should make the carbon count. It also means that we can count carbon emissions largely by looking at usage data. Let’s take web usage, for example. We know that it takes energy to shift data over the wire to our users. And in many cases, we probably track this using tools like Google analytics or Plausible and so on. One service that also makes it easy to track carbon usage is Cabin. Just like Google analytics, Cabin tracks, the usage of your site. But it also applies carbon calculations to give you figures on a per page load basis. Or highlight hotspots.
If you spend more time, server-side, I’ve already spoken about code carbon, but if you want to see something simpler, you can literally pip install energy usage to start play playing with stats like this.It’s made by the same people.
And you can use it to wrap any python callable. When the callable returns it will return figures for the emissions caused by that piece of code.
And if you enable it, a full on report, going into all kinds of nerdy details are generated for you as well.
Finally, if you use AWS or Microsoft Azure or Google cloud platform, you can also hook into the cost explorers and billing APIs to generate these kinds of stats. And it means that you can budget for carbon do where you manage budgets for actual cash, when thinking about your cloud bills. ThoughtWorks announced the cloud carbon footprint project just last month. It’s entirely open source, and it’s a TypeScript project, but there are a bunch of handy Docker containers you can use if you don’t want to mess around with node JS or react by yourself.
Finally the D stands for distributed.
We’re moving from a model where power generation is centralized around a few massive fossil fuel power stations, to one where power comes from a more diverse set of distributed forms of energy generation. Like wind farms and solar and so on. This means that the underlying economics are different. And if you build with this in mind, it’s possible to take advantage of this change.
The world we’ve been in so far is one where if we want to generate electricity, we have to pay for the fuel, we burn to generate that electricity. We are moving to a world where once we have infrastructure built, generating power is free.
This means that at certain times, like around lunchtime, when the sun is high in the sky, we have more power being generated for free than we need. This oversupply can end up pushing the cost of electricity right down to the point where the price can actually go negative.
If you plot the cost of electricity over time like this, when this happens, you’ll see that it looks a little bit like a duck. So this phenomenon is sometimes known as the duck curve.
Why does this even happen? The short version is that it’s really important to keep the electricity grid balanced with as much power going in as is coming out. And when there is an oversupply of power, It can be cheaper to just pay people to use the spare power, rather than pay someone running a massive coal fired power station to turn down their power station.
Octopus energy, another Django shop offer an agile tariff that lets you take advantage of this yourself in your own home. But if you are able to control when you run jobs, you can essentially move compute through time to take advantage of when energy is cheap and green.
In addition to moving compute jobs through time, you can also move them through space. Lancium is a US-based company that takes shipping containers full of servers, that have already had one useful life in hyperscale data centers and they plunk these shipping containers full of computers right next to wind and solar farms. They then take advantage of when there is an oversupply of power to use it to power these compute jobs, like machine learning and so on.
Because their cost is so much lower than normal data centers, they can sell cloud compute at a much lower price than traditional providers, like you can see here.
This is already kind of cool. But you can go further. Renewable energy is intermittent. You know, sometimes winds die down and clouds block out our sun. If you design a system that can tolerate individual failures and allow that work to be rescheduled, then the cost can come down further still, and this is because you’re taking advantage of the new properties of the underlying energy system.
This is also kind of good practice for reliable distributed systems anyway, but this is a really interesting new application of some skills you might already have. These ideas can be applied at a protocol level on the network as well. Not just compute. Let’s take the example of me in London, trying to connect to a website in servers, running in Poland. If I wanted to do this, I might need to make a series of hops to get there, going through infrastructure in France and then the Netherlands and then Germany, then finally, Poland.
Every hop uses energy and the greenness of each hop affects the final carbon footprint and for each , the greenness of the electricity effects, what kind of impact it’s going to have. There are a few things I could do to reduce the impact for this.
One thing I could do is use a CDN. Django already encourages me to split out static files so they can be served by Nginx or a content delivery network, like CloudFlare. So, this is good because these files are distributed all around the world to be closer to the users, and CloudFlare servers are probably going to be much more efficient than my servers, for serving that particular file.
Having a CDN also cuts the number of hops needed, because the majority of the hops are served somewhere close. So only a small portion are going all the way to Poland.
But you can go further than this though. There are new internet protocols like SCION, which allow for much more control over how you route data through the internet. Instead of relying on BGP, the border gateway protocol, that connects networks to each other, relying on that to make routine decisions, they work by annotating every node, along the path with information like latency, actual costs like money costs, or even carbon emissions. And then they allow you to express preferences about what path you want to take to actually get to your destination.
This makes it possible to choose a measurably greener routes through the network, kind of like low carbon internet trick shots across our planet.
I think this is pretty cool, and I think this is where we can and should be heading with the internet.
So that’s gold, green, open, lean, and distributed. I’d love to hear your feedback as it’s still a work in progress. Okay.
We’re on the homestretch now.
What can you do as a Djangonaut specifically? I’ll share a few examples and then I’ll wrap up.
I think there are a few things you can do that are worth bearing in mind.
So continue the climate Django tradition.
Understand, we’re entering a boom for climate tech.
Make Django the obvious choice for working on climate online.
As we discovered an hour ago, if you choose to work on climate with Django, you’d be in very good company, and continuing a tradition of working on meaningful problems with your favorite framework.
And if you thought the idea of something like that carbon account might be cool. Then one project I know of, and I’m a fan of would be the donation. They’re a Django shop, working with a bunch of companies you’ve heard of, and they’re hiring for loads of roles.
If you thought some of the energy stuff was cool. Then I know that octopus energy are also a well-known Django shop who are doing loads and loads of cool stuff with energy and are growing crazy fast. These are two I know of, but there are loads of others and I’ll explain why in a bit. But remember if you’re a djangonaut, you’re also part of society.
You don’t need to kind of quit your job and go working for a startup. I spoke about civil engagement with climate emergencies. And if you’re in the UK, it’s a really nice way to understand what’s already happening around you. And what plans are a foot that you can be part of.
And, while it’s totally cool to contribute with code, I hope examples of the laws being changed and the economics being changed show that there are other options available to you. And this is why I’m pointing you to things to help you engage on a civic level. But if you’re not in the UK, And you think it would be good to have something like this, there are two things I’d say.
First of all. I know that my society, the people behind these tools are actually looking to raise funding and extend them further.
Secondly, this is a Django app. You can deploy something like this yourself, and you could contribute to this Another thing you might do is take a look around and realize that we’re entering a boom period for climate tech.
These next 10 years are going to require a colossal amount of change, As people realize just how much work reconfiguring an entire global economy to not be based in fossils will be. I’ve used this phrase, climate tech a few times, and for better or worse, it looks like it’s become the phrase to describe this kind of tech company working on climate, not every problem is one that should be solved by venture capital investments, but there is an impressive amount of funding going into this sector from VC.
Last year, the State of ClimateTech, a report by PriceWaterhouseCoopers, a consulting firm was published.
It presented some numbers around how much money was pouring into this emerging climate tech sector.
It’s been growing incredibly quickly. And if anything is it’s accelerated since this report was published.
Climate tech as a category is growing five times as fast as the VC sector is part of. And if you look at how much money was investing into AI is three times as much as that. And we know how buzzy that was.
If you’re a developer now is a good time to get into climate. In addition to being existentially important society-wise many of these companies are going to be looking to hire as they build teams for the web.
But remember because Django is based on Python, if you know your way around language opens other climate related doors elsewhere, too.
I find transition elements a pretty inspiring example.
This company was started up by a few engineers and project managers, but used to work in the oil and gas sector. They found out that much of the pythonic skills and tooling that they had been using, And developing to figure out where to explore for oil could also help find safer conflict-free sources of materials like lithium.
The kinds of things we’ll need to get we’ll need to get our fossil fuels. So they left.
And they set up transition elements instead. But not everyone wants to set up new startups or even work in them. Since 2018. I’ve helped organize climate action.tech, an online community of technologists, just like yourself, who worked to push for changes internally or incorporate climate awareness or an awareness of this into their professional lives. It’
s free to join. And every week we add a new core cohort of joiners to the community. So, if you’re looking for other technologies to care about climate, it’s a good place. Each week we share a newsletter or events of who’s hiring and what what’s going on.
Finally one simple thing you can do to help as a Djangonaut is simply work to make Django the most attractive option to use, if you want to work on climate in a climate friendly way.
I know there is growing interest in doing sustainable web or sustainable software engineering. But right now, while you can find individual heroic examples in almost every software community, it’s not like there are any communities where this is publicly integrated into part of the culture.
Django is known for being flexible, and I think it’s possible to reuse loads of the existing work in the Python AI community who’ve already been working on how to do more sustainable tech already. I think you could reuse a bunch of those ideas, to make Django the obvious choice for doing sustainable software.
Doing this would make it much easier to argue for using Django on future climate tech projects.
And increasingly there are places to share these ideas, if you were to do this. Last month, the green software foundation was set up to take ideas like the ones I’ve shown, document them more fully and develop them into actual patterns and conventions and maybe even standards. It’s early days, but it’s worth keeping an eye on. Just like how POUR and accessibility became a non-negotiable thing for public sector, part of the goal of this organization, is to do the same with more sustainable ways of doing software engineering, like the ones we’ve discussed. I think there could be a space for Django in that.
So now, we’re going to really wrap up.
Thank you for spending an hour, looking at all this stuff with me and, the takeaways I’d share with you are this:
- continue the Django climate tradition, please. It’s really, really cool when it makes me so, so proud to be part of this community.
- To make Django green, think GOLD.
- Remember show up and support others working on climate. We’re starting to win, but we have an entire system to change.
All right. Thank you everyone. These are my deets.
This link shows you a browsable time synced video transcript, and there’s also a browsable Miro board where you can comment on, and link to, and share anything you saw today.
I’ll do my best to answer any comments that you place on there.