36C3 preroll music Herald: Our next speaker is Chris Adams, and we were talking about reducing carbon in the digital realm. How to understand the environmental impact of the digital products you built and take measurable steps to green your stack. Floor is yours. Chris Adams: Cool! applause Hello everyone. First of all, can you folks hear me in the back? Yeah. If you can just raise your hands. Excellent, cooI! All right. Hello everyone. My name is Chris Adams. As you can see, Mr. Credit, Chris Adams on pretty much every online ralley. Please don't try to follow this link because it's not actually gonna go anywhere yet, but it will be up at the end of this talk. I'm just going to introduce myself. If you don't already know me, my name is Chris Adams. I have a background working in environmental kind of wacky startups from Loco2, which was all about trying to make trains easier to book than planes. A.M.E.E, which was all about putting kind of carbon calculation as an API, so you could work out the carbon footprint of anything. And now I spend a bunch of my time working with the Green Web Foundation, where our mission is to basically make the web green. And I also help organize an online community called ClimateAction.tech, which as you can see is for folk like yourselves who want to work out how to do something about climate in their day job and what they do. I've got about 45 minutes with you and this is a kind of rough outline for the day. I'm gonna briefly give you a kind of primer on how you measure carbon and why you might measure that. I'll share with you a mental model which I found useful for helping think about what steps I can actually take as professional working in this field; and then I'll give you some pointers of where to go next. If you feel compelled to do something about, well, what is essentially an existential challenge and the biggest challenge we are already facing. So first of all, measuring carbon! So, can you just show you... raise your hands if you've seen this before? Yes. The water cycle. So you get the idea that we've got like a water, kind of, evaporates, goes along into the sky, then comes down and rains. And then... generally it's best if it doesn't go in one place because that's that's a good thing. But generally we have like cycles in nature and one of them we have is for water and we also have cycles elsewhere. And that's kind of why I want to share with you here, because there are also cycles around carbon. So what you're seeing here is actually some freeze frames of a really, really cool but somewhat confusing video. And as I was saying, kind of carbon works on a kind of slower timescale that might have here. So, this is an... this is basically a diagram of all the carbon in the world with some idea of proportions. All right? So, the green stuff up here, this is us like we're made of carbon. It turns out. So a tree. Trees are made of carbon. And when living things die and decompose, some of the carbon ends in the atmosphere as we decompose. The purple stuff here - this is like the ocean. Fish are made of carbon, too. And so are plants and everything like that. And when they die, they sink down, which is why you've got this massive chunky like stock of carbon down here. Eventually, some of that might become sediment, then form rock, then end up in this kind of black stuff around here, which we kind of consider that the earth's crust. So then at the top, we've got carbon in the atmosphere. So you really can't see it very well. But there is carbon up here. This is kind of atmospheric carbon, really. And there is like if you watch this video that I've linked to, you can see all moving around as a cycle. But this is kind of wealth carbon, really. So now in about 1850, we started using a lot more energy. And to meet that demand of all this energy, we started burning fossil fuels in earnest as well as burning wood for fuel. And you can see the fossil fuels here as kind of stuff that's kind of come out of here and is now represented here. And we've also taken some carbon in the earth and burn some of that, and like... What is it? Where does this all go? There's a clue in these arrows on the screen. Right? Now if we fast forward to 2017, it looks a bit more like this. You've seen like there's less of this red stuff down here because it's got into here and because it's circulated all the way around. We've ended up with carbon in other parts of Earth. And, in particular, you'll see that we have fewer red dots down here, but we've got a lot more around here, which is going to pushed some of the carbon higher up here. So this is kind of... if there's one thing I want you to kind of really take away is that when we talk about climate, it's really about carbon. And we need as an industry to be able to get good at understanding how to reduce carbon more than anything else. If we're going to be thinking about climate. Now, I'll try and break this down to a kind of simple way, which I found really, really useful for this. There's a lady on Twitter. Her name is Professor Julia Kay Steinberger. She's awesome. She has this really nice way. She recently shared this thread on Twitter where she basically explained how she explains kind of climate change to her children who are 4 and 7 years old. And she basically did it this way. She got like a globe, wrapped it in some plastic and then said, well, this is basically what happens. You understand the idea of being wrapped up in things getting warm and things getting too warm. That's it. And like, that turned out to be a really effective way for her to do it. And it turned out if kids can get this stuff, then I think we can. And I think it's really, really useful for us to kind of be aware that carbon is actually a thing that we do need to be able to count, and track and account for in our jobs. So we've spoken about this on a planetary scale, which isn't all that actionable. But if we were to kind of bring this down to organizational scale, where the most of us might kind of work and more specific, it might be easier for us to act upon. It might look a little bit like this. There are established ways to measure carbon within an organization you work in and in the same way that accountants might talk about legal ownership of a company. There are guidelines for talking about who has responsibility for emissions when you're reporting and deciding who should be changed, who should be making reductions. And I'm sure in this diagram here to actually take some very, very dull, dry material into something that we can relate to, which is hot beverages and coffee. And, generally, you can think of it like this: if either these three kind of scopes, there's this idea of scope 1 emissions, which is basically if I burn fossil fuels to heat up a, say, container so I can have coffee, then that's the emissions from me burning less fossil fuels myself and my scope 1 emissions. Scope 2 might be me using a kettle. So if someone is a burning coal to generate electricity for me to kind of boil the kettle, then that's my scope 2. Now scope 3 might be me walking to a store or into a coffee shop and all the emissions and all of that supply chain - that's my scope 3. So there's this idea of kind of like dependency chain of carbon, which... because I'm speaking to a bunch of people who have some understanding where technology I think you should be, kind of, you should be ready to be comfortable with the concept of dependency chains and things like that. And to make this a bit more concrete, I refer to, say, a good example. So Stripe... Has anyone heard of Stripe here? Okay, a few of you. Okay. So, basically, payments company. All right? Now they are actually pretty good about sharing information about what they do and what their emissions are. And you can see this here. They've basically got... in 2017, they started making noises about becoming carbon neutral and they started reporting information in these kind of scopes here. And you'll see this phrase - TCO2E, but basically just think of that as carbon dioxide, the stuff that's warming up the world because there's different gases, but they all have more or less the same effect. So in many cases, people will use the term CO2E2 to describe all of these kind of basket of gases. All right? But you see here that from a scope 1 emissions are quite small because they're usually just heating a building. Scope 2 is a little bit higher, because they pay to, say, keep a building running. And then you'll see down here that there's massive scope 3, because they tend to pay for a lift infrastructure and have a lot of people flying around as if they're prevent a lists??? and so on. All right? Let's look at some other other examples. Amazon! So Amazon repeat started reporting on CO2 emissions for the first time this year. And long story short, their carbon footprint is about the same as the country of Finland. Right. Which is kind of large. And as you can see, because obviously they run lots of services, which is why you got 4,5 million tonnes of CO2. They also have a lot of warehouses and things like that, and have a lot of kind of cars or vehicles that are getting things around. But because they actually have a large supply chain themselves, they have a massive honking great scope set of scope 3 emissions. So let's look at another company, right. Google. So Google is a Google has a surprisingly low emissions considering scale, about the same as the African country of Liberia. All right? Now, there's something interesting here. Google have reported their emissions with scope 1, 2 that you can see here. And 3, but they have this interesting thing here, where this is the energy they're using. But this is the energy they say that they're using because they are purchasing what are basically renewable energy credits, which is one way of using green energy. If you do not have access to green energy in certain parts of the grid and Google are pretty good in this field, but it's worth bearing that in mind. Google are transparent about this stuff, where at Amazon, you don't see these numbers. So this number could be somewhat higher for we know. All right. So this gives you an idea of what some of these numbers might look like. And then let's look at another company, which is also, well I assume you might have heard of. Is this works? Yes. I go to Apple. Yeah. So Apple has... these are their emissions with just their facilities. So just their data centers and just, say, the offices that they have. Right. So this is quite low, by comparison. Right. So this is carbon emissions of maybe Gambia - another small African country. All right. And you'll see once again, by sourcing renewable energy, they've had a kind of quite big reduction in the emissions that they otherwise wouldn't had. But you'll also see that in scope 3, because there's lots of flying and because there's lots of commuting, they have a lovely new office miles away from anywhere. So you have to drive to get there, you know, it's like, I don't know, 40 percent of the emissions were just their facilities. We have to drive there and back. All right. But if you look at the emissions from Apple as all the products that people buy, you'll see that the emissions are somewhat larger, about the same size as Mongolia, which is kind of large. And you will also see that there is a large... this is a breakdown from Apples annual report, and they basically... they're pretty transparent about this stuff and they're good on this. And they say this is where the emissions come from. And you can see all the way around here, this is the manufacturing. The machines they have. And then if we zoom in to like this bit over here. Right. You'll see that there's a little bit of information there. There's like, obviously use phase. But the main car, the main impact from what they do is actually making the electronics. So this is kind of how we tend to think about a carbon and how one might report on carbon and make it and make decisions to reduce carbon. All right. Now, I'm going to share with you a mental model, which I found helpful in this field. I call it Platform, Packets and Process. And it's generally aimed taking something which is quite abstract to this scope 1, 2 and 3 thing, to something that you might act upon inside the teams that you work in. And it kind of maps to the kind of groups that you might be working in if you're, say, a frontend developer or a backend developer, or a designer, or product manager. So it kind of tries to map to that. And as you can see, platform might be infrastructure you run; Packets is infrastructure other folk run, like the rest of the Internet; and Process would be decisions that are made inside your organization that cause that to be emissions. So if you highlight a large company and you have a massive, say, corporate campus which is miles away from everyone else has to drive to. Then there's gonna be a load of emissions from that decision that you actually make. So let's run through this. On the scope 1, 2, 3 thing I shared with you, this is a kind of how it maps. And this might look like some of the things, some activities you might have been involved in that will create emissions. So, generally, on the platform side, if you're running infrastructure yourself, there'll be scope 2; if it's scope 3 you'll see here and, well, I'll go into the other ones in more detail later. All right. So this is the model that I'm sharing with you today. All right. So let's have a look at Platform: infrastructure that you run. There are kind of three levels, which I'm going to share with you today, which might be of use to you. All right. There are so basically Provisioning, Provider and I'm a bit worried about showing this last one because it's kind of new: the other programing languages. And that's partly based on some stuff from yesterday. But also, I'm a bit worried about kind of inciting the pitchfolks when I share this. But we'll see where we go with this. So Provisioning is one level you might actually have if you build digital products and you run servers. All right. So this chart you see here is a chart from the CEO Power of Wireless Cloud Report. This basically shows how Australia uses the Internet and it gives you an idea that, well, we don't use the Internet in a uniform fashion all the time, because basically if you think about us being set at midnight as we tend to fall asleep, we tend to use Internet less. All right. And then as we wake up, we might say kind of have coffee or something like that, then as more of us come to work, we start using it more. And then later on we work during the day, then we'll go home. But you will start watching Netflix and things. And then once again, we fall asleep and then we get to there. So there is a kind of pattern. And if you have any analytics on how your own tools are using your ad services are used, you'll see these kind of waves and patterns in how it works. And this is worth bearing in mind, because traditionally when we've had to provision services to support something like a Web site or a video streaming service, the way that we used to do it was basically buy a big server that can handle the peak usage. And we just accept that. Because Provisioning is quite a difficult kind of, because it's a pain to do. We just leave this big box idling most of the time in the hope that we could actually hit the surfing at the peak, right. Now the downside of that is that, well, if we can make this conceptual leap that the cloud computing is someone else's computer, then it shouldn't be that much of a leap again, to realize that computers run electricity and we generally burn fossil fuels right now to generate electricity. So in the example we had before, when we had this big box that wasn't being used, we would basically be burning money to pay for capacity, we didn't have. But we're also burning fossil fuels and causing emissions one way. Now we've got better kind of running infrastructure. All right. So we've had this kind of trend over the last, say, 10 to 15 years to abstract machines away to make them easier to manage. So this might be VMs or containers, or dinos, or unicorns, or whatever you might prefer to use. But in general, the pattern is make something more abstract and then make it easier to spin it up and down in response to demand. All right. This is better because we now have something looks a bit like this. So we are not... we're wasting less well, we're burning less money, but we're still burning money. And as a result, we're still burning fossil fuels. Now, there's some new changes in the last few years, which are relatively recent, and we're seeing kind of changes in how we work now, which look a lot, look and sound is sometimes referred to as like service lists or functions as a service if you work on the backend with the servers, right. Now here, this is interesting, because we're rewarded for efficient use of computers, because we basically pay on a per request basis. So if someone tries to load a page, we pay for specifically that request. And then when we're not using it, theoretically we're paying for nothing and theoretically things can spin right down. So there's a much clearer mapping between the usage and what you would pay for here. Now, there is a trade off here that is the number of providers that if you want to use tools like this and have this much tighter mapping is that we end up with a small number of people who do provide this. So if you were to go to, if you want to move to something like this, you generally end up using something like either Microsoft Stacks or Google or AWS. And this kind of presents us with a kind of awkward problem right now. If we care about climate and we care about like basically being able to solve problems, as say professionals, we have to choose two of these three things. So we kind of care about the fossil fuel thing because we're in a climate crisis officially now. All right. Also since I think December, when the EU parlament declared this. If you've used to running any kind of online service, you'll be aware that in many cases there is a kind of shift to basically use a hosted thing, because it turns out that running infrastructure is extremely complicated. And in many cases, if you can buy versus build, then you'll often save yourself a lot of hassle. And the final one thing is, well, avoiding oligopolies is quite a good thing, and because diverse ecosystems are healthy ecosystems and we kind of don't really have so many options right now; if you do care about this, it does feel a little bit like this. And as we saw from just yesterday, if you are at this stage, you saw Gillian Oliver talking about this and how much infrastructure you had to run themselves, in many cases because we didn't win it with this consolidation, where if you want to use one service, you have to buy into all these other things associated with that company. All right. So that's one of the trade off you might have to think about when it comes to, say, Platform and Provisioning. Provider is also another decision that might affect it. Now, I didn't mention in the beginning of this talk that you can basically reduce emissions from what you do. If you say a cloud provide like AWS just by running it in a different region. So you can see this map here. This is by this person who is best paid by Amazon to build kind of sketch naughty things. And this is a list of all the data centers and you'll see the ones with the green leaves, which are marked as kind of sustainable regions. And you see on the right hand side on the east coast of America, which is around North Virginia, which has traditionally been coal country - you don't have much in the green regions there, but on the left hand side, where you see a lot of kind of leaves. And that's because on the west coast of America, there's a lot more hydro and things like that. So, you know, with kind of a lower CO2 for each kind of unit of computer that you're paying for. So, yes, you can reduce emissions just by switching from one region to another region. And this is because, basically, to expand that point I shared with you before, where you are in the world... like the place will affect the carbon intensity of the emissions that come from running infrastructure anywhere. So let's have a look at Europe. Right. So France kind of green full of nukes, right. So that's good for some people, maybe less so in Germany. Right. Germany, which is up here, this is the land of solar and coal. Right. So we're not all that green. England is actually got lit up quite a bit better than it was before. But up here, you can see Poland: aww, not so good. Poland is really, really into coal. And as you can see up here in the Nordic states where there's loads and loads of mountains and wind and water, that things are really, really green. And this gives you an idea that you can kind of see where the emission, the likely emissions might be depending on where you're on the world and that you might make decisions based on this. But to have to kind of compare against this is gonna be quite complicated process. So one thing that we do. All right. The Green Web Foundation, which is where I currently work, is we build like a directory to make it easier for you to make that, to do the right thing. And we present this information as an API and as data sets for you to kind of build into your own tool in which I'll expand a little bit later. But one of the key things is that your provider will actually have an impact on who you... basically, where something is will have an impact on emissions from here. There is also... At this point here I wanna segue a bit into kind of energy markets, because when you start working with computers and if you have to, if you're responsible for any service, the more you think about cloud or how to work with cloud, the more you realize how many parallels there are to energy markets. Now energy markets are really interesting in lots of strange ways, right. So in Europe, but actually in lots of places around the world, you can have cases where the cost of electricity is actually negative rather than positive. So like on a day where there's which is really, really sunny or really, really windy, it's actually cheaper for the grid to basically pay people to basically take energy off the grid to keep it stable than it is to say power down a nuclear power station or power down some kind of large coal fired power station. And then under assault, you end up with these scenarios where you have a negative and you have negative cost. And one of the reasons here is to basically get people to kind of take this demand and put it to use in other places. And I'm sharing this with you because it's the kind of the idea like shifting load might be might be something you're aware of as developers or trying to kind of delay jobs, for example. But you see this manifesting now in how we use energy, but basically things like with IV's and things. So this is an example: Bulb energy in the UK. They will basically... if you have a car, they, basically have time rest now, where depending on the time of day or if you're prepared to kind of provide, be a little bit less strict about when you need something running, then you'll get a cheaper electricity. And why am I sharing this with you? Because I think you start to see things like this in the realm of like computing now. So this is a paper that was shared this year, the ICT for Sustainability (ICT4S), basic conference in Lapperate, which I've misspelt. Sorry, Finnish people. And the general idea is that these people started building Kubernetes Scheduler to basically run a docker container, run machines, run workloads where energy was cheap and green by basically tracking where it was sunny really. And they were able to do this on where they end up working with Microsoft to do this, because Microsoft is one large company that has a number of data centers all around the world. But it kind of sucks that there's only one company that you can actually get this stuff from, or if you wanted to do something kind of cool with, say, a more decentralized use of the web that... you have to kind of go through one large company. And if we were to look at say, well, maybe a more kind of open, green and decentralized web or Internet might look like there is actually some lessons we could learn from the energy sector over the last, say, 10 to 20 years. So Germany, one thing that we saw was the Energie- wende over the last 15 years. And the result of that was that we had like cheap green distributed energy. So that we, Germany is interesting in the sense that it has quite a heterogeneous grid. So there's lots and lots of smaller providers of energy rather than just a handful of huge providers. And there's lots of reasons why having a diverse ecosystem is helpful in this, right. And I kind of wonder like this is one I share with you as an idea. What if we'd something like a digitalwende? Right. What if we could do something like this to, kind of, abstract computing away to the point that you can run these in the same way? There are companies that are now doing stuff like this right now. And there's one company called Helios Exchange that does exactly this. But the paper I showed you before, shows you this stuff around there. So this may provide a way away from having to rely on just basically an oligopoly and concentrating more power. If we were to kind of be prepared to think a bit more about how we run computing around. So final thing is, as this possible way of sharing was programing language. So where appropriate, you can have an impact here as well, because different languages have different goals and that can result in reducing emissions from just much more efficient use of resources. So this is Hannes Mehnert yesterday. He was presenting some work he's been doing on MirageOS unikernels. All right. And whether you probably can't see it. He was basically making the point that this is how he used running this stuff before. When he switched to using unikernels, he saw memory usage and CPU dropped massively by just having a kind of better use of the existing resources. So this was like a 25 fold decrease in compute use and a 10 times decrease in like RAM usage. And you see the same things with other computing, other languages. And the nice thing is this is recorded now so you can see it too tomorrow. All right. But there's also papers that talk all about this stuff. So depending on what your goals might be, there may be certain languages which are really better kind of optimized for the task that you might actually have. Now, does this mean that I'm saying that we should all go out and code everything in Ocaml, C and Rust? No, that's a really, really... we choose languages for a wide range of reasons from ecosystem to hiring, to like developer happiness. And when you look at a project or a product level, you'll see that these kind of micro optimizations, more than fun, might not be the most effective way to achieve some emissions reductions. But it's still out there and it's worth being aware of. And also, it's worth thinking about if you are able to kind of think about the entire stack of tools you might be using, then you're kind of doing something like this in many cases. If you say use like Redis or nginx or something to serve things. So that's the idea for, like, platform. Now let's talk about packets. So this is... I've spoken of infrastructure you control, this is infrastructure you do not control. All right? Now, you cannot really control the other parts of the Internet, and that's generally considered a good thing. But what you can do is control how much data you send over the wire instead, allright. And if we were to kind of look at, say, the amount of energy we sent over the wire and we figured out that both sending data uses infrastructure, it uses energy, which uses fossil fuels. Then we've got some bad news, like we've seen pages growing in size to the point that in average they... I think the mean web page size is now larger than the original download of Doom. All right. But we're also seeing it, because we have mobile phones, so we use this more. All right. And then because cellular network tend to use more energy to shift the same amount of data as, say, wired or Wi-Fi networks, we have a loss here. So from a kind of energy and climate point of view, this is like the worst scenario we can imagine right now. Thankfully, there are this, if we think about Web page budgets as basically carbon budgets, we realize we have lots of tools that we can repurpose for carbon reductions. So one example is Google's Lighthouse. It basically runs checks against your page, then it grades you on how well your page is optimized. So what we've been doing at the Green Web Foundation is taken Lighthouse, we forked it and we made Greenhouse. Which was basically the same idea, but it kind of looks at how many resources you run and then says, well, yeah, climate emergency folks, maybe you don't want to get all your stuff from fossil fuels. And there's some other things far worse than we can make it, kind of, work out the carbon footprint from this, because these numbers exist, but that's further down the line. But you can also see that I'm referring to the ethical web principles here. If you care about this as a professional, this is something that the creators of the web are now saying and giving you license to be doing. If you do say, look, I don't want to do this, there is a moral argument for doing this. And if I want to build the web, as Tim intended, then you can refer back to these ethical principles now for this. But I will need to share with you that that doesn't mean that we should just like reader optimize every single web page and they'll be fine. Right. It's worth getting a sense of perspective around this. All right. Video like, video just dwarfed web traffic. When we think about designs, we might make like to give you some context. This chart is just showing you an idea of where like usage of like data flows. Right. Now, 60 percent of this basically is just telling us that all the video used and streamed is about 300 kind of megatons of CO2. And that was what it was in 2018, which is roughly the carbon footprint of Spain. All right. So all the video = Spain, that's kind of numbers and numbers you might want to look at. Now, video on demand like, say, Netflix and stuff - just like the country of Chile. Porn - that's like Austria. All right. So these are some kind of like reference points is for you to kind of refer to now. All right. And I speak about that process now. And like, this is why it's worth thinking about some other thing. This is why I also care about kind of making the web green, because I think it's going to be easier to make the entire Internet green than it is to stop people watching porn, basically, which is a statement, I suppose. So I spoke a bit about the process and how there are other things you can do outside of computers. All right. So they are obvious. There's two ways I'm going to show this. So there's kind of inward looking at process, like the greening of how we build digital products. Right. This isn't visible to the end users, but it's still a useful thing to do. Now, an example of this is the company called WHOLEGRAINDigital. I really admire that. They're really really cool company doing some good stuff on the web and they are one of the original kind of WordPress agencies and they basically say, yes, we do everything with WordPress and green energy and they started working at their own missions and blogging about this. And what they said was that they know, that they looked at these figures and they switched to running on green infra because that was the kind of right thing for them to do. But when I started doing that, they started looking at where else are their emissions and what they do. And they basically looked at and some realized that, oh, wow, a large part of their emissions just comes from travel. And you guys, you can see office and home energy. And of the travel, around 94 percent of their emissions came from commuting. So this is why I'm saying that it's more than just playing around with computers and optimizing stuff. All right. So this was kind of useful and this kind of inspired some of the work agreement foundation for us to start sharing this. And what we've been doing recently is basically take this model and build some, like I guess in VCC, which is I'm afraid is minimum viable carbon calculator, not the post-growth, other kind of acronym there. And we basically built like a simple spreadsheet that is very, very fast to fill out. So we picked two people to kind of get an idea of, okay. These are the things I'm paying to run. This is the like if I build a web project, this is how much data I'm shifting over time. And then because it turns out that it uses energy to keep people warm and dry inside buildings and people tend to use energy commuting. Then we tracked it, we tracked that as well. And we do this to give people some figures and get some idea of where they might want to act upon this, because the current ways that you report on, say, emissions or think about this is like an annual report every year. That's a really, really slow debugging cycle. That means we have like eight or nine kind of hits of the F5 key before, like it's a climate apocalypse. That feels like we should all do a little better than that. But this is entirely open, you can link to the template yourself. And we've we learnt some interesting things and we did this. We realized that when we were working on this because the team were pretty good on commuting the let them cycle to work. They were emissions are quite low there. But we also realized that there was actually an argument for changing how we design, say, a Web site. We found that one chunky background video, you know, those things that we all hate. Right. That had the same carbon footprint as basically the entire team commuting for the entire project. All right. So it's quite easy to make an argument to get rid of that and it kind of meaningful, measurable reduction there. So these are some of the stuff that we do now in this. The other thing is an outward process. So these might be decisions you make that affect the kind of emissions through use or for your end users. And these are gonna be much more visible to end users. The example I like to refer to, because I'm a fan of the company. is Fairphone. You folks have heard a Fairphone here, right? Yeah. Like they're basically the canonical Fairtrade smart smartphone, really. All right. That's the best way to describe them. And they also... there's lots of good things about what they do, but they also share lots of information about their own carbon emissions and what steps they're trying to take to reduce them in a relatively honest way, which is also really good. And because they publish it, we can read it. Now, we've seen trends in electronics over the last, say, five or six years where we had modular things and we saw a few failed attempts at having modular smartphones. We've been a shift towards like systems on a chip and stuff like that. So if you are going to do that, that means in some way that having a modular design is a challenge. But it means that if most of the energy or impact on building electronics is coming from basically turning sand into a chip with lots, lots of energy then we'll need to think about where we might have a way to decouple this from the rest of it. All right. To make us the electronics we do have last longer. All right. And this is what I found in there, over LCA life CO2-analysis report, where they look at the emissions over the entire process. They... basically the first thing is their actual use is pretty small for this. But what you can see here - the emissions were from the production. And the thing they decide to do was make the phone as easy as possible to repair for end users or replace parts of it. To make it go from, say, a 5-years project product to a sort... from a 3-years product to a 5-years product. And this had the impact of reducing the emissions by a measurable figure. And it's a pattern that we might follow ourselves. All right. Now, actually. Fairphone and it make a real feature of this now, they've talked about this to their, kind of, audience. So they basically say if you can just say upgrade just the camera rather than the rest of the phone, that is still kind of working, then you can reduce the carbon footprint over the lifetime by these kind of figures. And I think this is actually worth sharing because it hints that there are options for us to actually be doing something until we can run everything on green power, which would affect these numbers. And I think I've shared with you like a mental model and see how it can be activated. I've spoken with you about kind of carbon. Now I'm just gonna give you some steps of where to go next. So, okay. It's 2020. Right. And I kind of feel this feels like this table stakes. And I'm really glad that at least one person is taking photos of this. I'd really like it, if more of you to take some photos of this and share with your peers, because I think the single most effective thing we can use in community is stuff like this. All right. We would expect we need to kind of make this just... we need to change the static about how we build things. So in the same way that we would expect a builder to know about asbestos and we expect automotive engineers to know about lead poisoning in particulars. I think as professionals we need to know about the impact of carbon and what we do and that we don't need it to build digital services. And I think if we can build electric cars in the automotive center, then we can build green stacks in technology. And I think we need this to become the norm. So this is what I need. I need your help me to share this with your boss, with your coworker. Set something like this. We've had people talk about just how bad the situation is. And like this is really like one of the minimum things we can do. Which doesn't actually have a massive cost for us, as you do. So if we do, we need something like a cloud moonshot to get off fossil fuels. And this is one thing I really would ask you to really consider. So asking for this is a bit easier if you have friends. So I'm in a group called ClimateAction.tech, where it's basically just a slack group with a few other things that we now run from here, on meet ups. And the idea is that we do this to kind of share what the strategies are to actually kind of push for this kind of stuff, because not everyone can join Extinction Rebellion. And I'm not sure that everyone should join Extinction Rebellion. They used that we have different aspects of ourselves that we bring in our work compared to where else we might work. Now if you are interested in acting upon any of this. I work at Agreement Foundation and we provide, we have like an open source platform where you can check your stack and then the tools you have to do this. And if you are interested about any of this stuff, we are trying to find a way to make it easier or to make some of this much more transparent, because I shared with you before about how large companies can basically fudge some of the numbers to make them look greener than they otherwise might be. So tomorrow we're running a kind of on lecture room M2. We're running something like a workshop to figure out what some of this might look like in the same way that we set up robots.txt. So carbon.txt a way to verify this. There's a Web site that that's been hastily put together to give you some idea for this. And finally, I'm just gonna wrap up now. Thank you for letting me talk to you about this stuff here. If you're interested in getting in touch and talk to me, please do the Green Web Foundation. I'm Mr. Chris Adams on Twitter and GitHub. We're asked to do some training around this because we realize that although most of us do care, we don't have much ways to act upon this kind of stuff. Then finally, if you find this interesting, there's a newsletter which I've started with a friend of mine, Martin, where we're basically sharing what we learn as we go through it, which bits are hard, which bits aren't so hard. And yeah. Thank you everyone! applause Herald: Thank you very much. We do have plenty of time for Q&A. Nonetheless, I would like the Q&A to be of high quality. So when you ask a question, it should be a question. Thanking the speaker is lovely and nice, but don't waste our collective bandwidth on that and do that afterwards on your own. And I don't really care what your name is or what your affiliation is. Just ask a question, and make it short and so on. And the first two questions go to the Internet. Signal Angel: You've mentioned the preview for other workshop, a carbon.txt just now. Can you already share some strategies how, for example, the Green Web Foundation and other sites directories can protect themselves against being abused by companies, for example, through greenwashing. Chris: So I think... you just need to read some reports about this kind of stuff. And get familiar with this. Now, I really apologize - a large bunch of this was incredibly dry. There was lots and lots of dry material around this to, actually, figure this stuff out. I think it's mainly a case of working out what numbers you... what questions you might actually have to ask. And I can share a link specifically for the questions that you need to ask. But generally, I think the four things I would look for: if the organization hasn't made a public statement about when they're going to hit zero emissions; if they're not sharing their progress on an annual basis; if they're not sharing how much of their business they get from basically the from fossil fuels right now; then how much of their business is still involved in extracting fossil fuels from the ground. I think these are the key things and if they're not using this kind of scoped emissions process, which is a clear thing of finding numbers, those are the biggest ones. There is a page on that I've got which lists these questions to ask and I'll share link to everyone since I get on the webs to link to this. Signal Angel: And the second question from IRC: The numbers that you've mentioned that the companies themselves publish. Can you verify them? And if yes, how? Chris: So the thing that you can... Oh, that's the thing is - independent verification is the thing you need to ask for, it's the final one. So I have just finally seen where the voice is coming from. It's like the voice of God speaking. The independent verification is very important. All the examples I pointed to, had independent verification, usually from a set of companies that do auditing of this kind of stuff. Now there is again, there's a firewall of tedium around this stuff, like that thing where I showed you with a Google with Google's numbers being high and low. This is because you need to kind of go into the minutiae of understanding market based reporting, because location based reporting and there's reporting around this stuff, there's like lots of academic literature, but there's just not very accessible to lots of people right now. They need to be more of us who do do this stuff. This is partly one of the reasons we want to have something like carbon.txt was to basically give people a chance to see this, but also linked to the specific documentation they're referring to. So you can ask. Okay. So it's nice you've done this. But whereas this was a third party verification of this or why is the same guy called Trevor being basically audited all of these companies and like literally it's the same dude who's reported Amazon's and Apple's. And I'm not sure about Google. Right. But Trevor is probably a good guy. But the fact that there is only one person doing this is kind of... come on, we know that there are things that you probably don't want to do in this field. Also in Europe, at least, if you want to sell renewable energy, you do need to register this with the government registry. And there is something in there... I didn't really talk about this in the session. The aim is talk about more in carbon.txt, how to find this and how to look this up. Basically, everything we know about SSL and DNS, you can apply that to solve this kind of problem without needing a freaking block chain. All right. And you can actually find something useful here, right? Like this data is out there. It's just the we need to better know the right questions to ask and make sure that, well, we are running stuff on a green stack rather than a brown stack. Herald: Ok. Microphone number 6, state your question. Mic6: Hello and thanks. Very good. Thanks for your great talk. Herald: I've said something about thanking the speaker before. Mic6: I know. That's why I said it. I understand you could try to convince people to reduce their carbon emissions. Now the managers I have been working under are generally good in reducing cost. Do you think it would help to translate carbon emissions into cost and have this problem solved by the invisible hand? Chris: So this is actually the approach that Amazon use. And for some of the spreadsheets I've showed you before where we do not have numbers forecast for emission, for reported emissions, the best thing you can go on is going to be sector level averages for the carbon intensity of spending a 1000 pounds or 1000 euros in a particular area. So you might look at it like that. I kind of feel that there are certain people who really respond to this kind of like cost base messaging. But I think it's more attractive for us to kind of change the aesthetic around what we do so rather than us continually striving. I mean, if we had the narrative for people for our generation was to basically save the planet, that feels a much more attractive thing than using cost. But I do... I'm aware that if you are in an organization where the primary driver is cost, then you need to be able to use that language. And that's why we've been speaking about some of that. But there are lots and lots... there's actually stuff in this field to show this. There are organizations that will basically help you quantify the risks from doing nothing versus the cost of action. Because in many cases, when we talk about this, we think that we often use... you hear the phrase: "Well, what's it going to cost to shift to kind of green energy?", for example. But it's like - what we pretend is there is no cost to inaction, when it really, really is. And we've seen just the five years massive, massive changes in the destruction of companies and industries and, well, hundreds and hundreds of lives as well. So I think that you can use costs for this, but I think it's a bit reductive to only use costs. Herald: Shall be microphone number 2. Mic2: You in your research, did you encounter a tradeoff between privacy and security and carbon neutrality? And if yes, in which cases? Chris: Yes. In fact, I used to work at a company, Amy. One of the key ideas behind the company was that if you can understand the company's CO2 emissions, you can understand the supply chain. And if you understand the supply chain, then in some ways you can use it as a kind of cost to kind of beaten organization to reduce their prices, because you can see that they're much more wasteful compared to other ones. But at the same time, this is... so that works at an organizational level. There's also a personal aspect to this. And I think that this is a field that in many cases is focusing on individual action and kind of shaming people has been proven not to be very, very effective. But that doesn't mean that you shouldn't do anything, that there isn't a cause for individual action, because I think that provides the cover for politicians to make the decisions which will result in kind of real meaningful changes. So there is a tradeoff, because in order for you to understand emissions, because basically emissions are essentially a proxy for activity and you will usually see this and there are plenty of stories around this and hopefully there might even be a topic about this specific subject on Republica, in May. Because, yeah, there is lots of interesting literature around this tradeoff that we do actually have to make. Herald: Microphone number 1. Mic1: Hi. Can you make an educated guess on how much emissions could be spared if like the big providers would follow your advice? Chris: So I guess it's tied to the, you know... I've showed you the numbers of Google like that different kind of charts. Right. You could make the list, kind of like coyork??... just cheat argument. You say, well, maybe all of it. Right. But I don't think that's really accurate. Generally, if we are looking at just the CO2 emissions from just like running the Internet, right. I reckon you could probably wipe out two thirds to 80 percent of it, what actually depends on where you might be looking at this, actually. So more than half comfortably. All right. Because if you look at, say, data centers, the main driver there, the main source of emissions is from them being on 24/7, continually use like three quarters of the emissions, assuming that data centers are full of servers which are used for three to five years, which is common. I'm not sure this is the case. Please talk to me if that's not the case, because I hear rumors that that might not be the case at large companies, but no one will write this down. So until you know that I can't give you a really educated, a really kind of better guess than that, but it'll be lovely to find out if that data exists and there are lot of hackers who might get to know about this kind of stuff. Herald: Microphone number 4. Mic4: When you talk about moving computation around the globe, basically cause to the weather. I wonder if there isn't a lot of overhead associated with that, like additional communication. Maybe if you know, you're far away from a database, you need rooting hops or whatsoever. Do you know anything about that? Chris: Yes, I spoke to the guy to Alan James, who was actually working on that. And I said, hey - what you said, I said, yeah, we take that into account because we can work out the emissions from moving a container of this size to over there. And in many cases, we might move to the state that we watch you when a query there as well. So his approach was to basically apply various kinds of metadata tags to the kind of jobs you might want to run to provide this kind of flexibility. And this is not new idea like a Mastodon ... is a kind of data science company that started doing something like this 10 years ago. They're kind of the ... Green Cloud people doing this kind of stuff. And even before then, there's this phrase called "chasing the moon". Where was this idea that you can do this if you run stuff on the dark side of Earth, which sounds super metal. Right. There is... this is not a new concept, really, but it is cool. Herald: Microphone number five, please. Mic5: Hi. Do you see any chance of getting governmental support with this, for example, like with tax cuts for electronic cars? I mean, maybe it would be possible in here, too. Chris: Yes. I was actually at the EU Commission Green Public Procurement workshops. They've been doing for the last like few months. And I was... there were a tiny number of people from small companies. There were lots of large, large companies who were there saying, yes, they think you should do is move to our cloud. That's the clear solution to the climate crisis. But they actually, it does look like there is guidance and there is going to be support in this field. All right. So I do know there's going to be... well, we've already seen this like we've seen the European Parliament declare climate emergency saying: we need to halve emissions by 2030. That's around 8%, year on year for the next 10 years. Now, most of us don't know what that looks like because the last time you saw 8% drop in a single year was the collapse of the Soviet Union. Which is partly why I'm kind of sharing stuff like this, because I think the idea of having a more managed reduction of emissions feels more kind of conducive to I guess a kind of continuities of how we live than the collapse of the Soviet Union. So I think there is stuff out there. But in many cases, we don't have the knowledge right now as people in the sector to know what is effective. And this is something that we need to, as professionals, learn, to learn where the levers are if we want to consider ourselves as professionals facing the scale of the challenge that is ahead of us. Herald: Microphone number two, please. Mic2: Hello. So my question is about you talked a lot about how much a carbon emission happens because of running a server in the server side. But if the traffic goes really, really high in a real larger scale, you might have a lots of emission just because of their transmitting the package through the datacenter from and ... and backbones plus like there carbon emission from rendering the page in the like formZ and everything around that. Is there... I was looking for it for a while, but I couldn't find a number like saying okay, one TB of traffic in the data center from United States going to cause like this much of carbon. Chris: Yeah. Is that your question? What is the carbon footprint of a gigabyte of data or something? Mic2: I want to have some numbers saying like which one is like translate traffic to a number. Chris: Okay. So there are two organizations who are doing some work in this field. In fact, there's a whole kind of community around greener web performance where they are tracking this kind of stuff now. There's a group called the Shift Project, who I've referenced before, who talk about the carbon footprint... the video being the carbon footprint of Spain. They've actually got some browser extensions which you can install into Firefox. I'll give you numbers as you browse to see this. I've also put together just like some interactive notebooks, you can get some ballpark figures of this kind of stuff yourself. So very quickly, you can decide, well, do I do this or do I do something different. So, yes, look at Green Web Foundation, there's a link specifically to a notebook with the numbers for this. Herald: Next question goes to the Internet. Signal Angel: Hey, IRC states that in their experience, Kubernetes has a quite a high CPU idle usage, about 40 percent as mentioned. This mechanism that you showed to adapt the server usage to the demand. Does this mitigate against this? Chris: I don't really know enough about Kubernetes to give a useful opinion on Kubernetes. So my thing was like this is an interesting idea because it's treating cloud and compute like a utility. And as a result, we see patterns that we've seen that have success in other sectors. But that could be applied to us. I don't know beyond that, but I can tell you that there's a lot of well, there's basically funding going into this kind of stuff now. But I'm sorry, I don't know much more than that. But if you do work with Kubernetes, please you talk to me because I would be really nice to have a better answer than I don't know in future. Herald: Microphone number three. Mic3: How many talks do I have to attend here? So that it will have been worth it traveling here from Munich instead of watching this online. Chris: So it depends what... Okay. First of all, I think coming to a conference, just because there's talks, that's not the reason to come to a conference. You come to a conference to have high quality, high context conversations with other people and get something useful from that, because like you said, you don't need to do that. That said, the idea of like having kind of physical community is actually very, very useful. I think that there isn't a number, I can't say like seven because that's going to be meaningless. But there is a whole issue around basically the carbon footprint of traveling to events and in many cases, so for some context: I was organizing a conference in London called Helping Organize your Conference called Map Camp, and we were trying to internalize the carbon costs of people traveling, and we found that some basically a minority of people coming across the Atlantic Ocean wiped out, I think, by half the carbon footprint, the kind of carbon budget for a 600 person conference. So there is some numbers around that. We've actually hired a group to give to actually publish some of this information out there. And there's a couple of widgets to figure this stuff out. But if you're here for the talks, that's one thing. But really, you should be here to make speak to the other people and get some kind of meaningful connection you can have from there. Herald: Microphone number one. Mic1: Is this possible to create like an automated way, a platform or service that tech companies can connect to and estimate, like have a rough estimate about the carbon footprint based on the stack they use, based on the bandwidth, based on the, you know, different process information. Chris: It depends. This relies on the organization having access to the matrix that will go in as an input. So garbage in, garbage out. Right. So the spreadsheet I pointed to gives you a very, very low quality version of doing that. There's also a tool called AWS Green Cost Explorer. We've basically forked a diverse cost. AWS Cost Explorer worked out which ones, which regions are running on fossil fuels and then we present that information back to you so you can get some idea for this so you can work out these numbers. But I don't see them right now, largely because a lot of organizations see this information is commercially sensitive. So they don't like to share this. So we have to go on basically kind of some rough numbers here. And this is one of the problems that we do have. And that came up with the grim public procurement thing was that we don't have the transparency right now to make the particular informed decisions about this. But theoretically, yes. Herald: Microphone number five. Mic5: Hi. I have a question about this double things that you are doing. On one side, you have the getting things done like building sustainable infrastructure. And in the last days, yesterday, there was a couple of examples of that. And on the other side is generate momentum like convincing people to join the movement and be more aware of that. So I was wondering, how does, for example, the Green Web Foundation apply to that? In the sense then how radical can you be like, can you kick out people from your directory because you think they are not doing green enough? Or how does this work? Chris: So I should be clear about the Green Web Foundation just being a handful of guys. It's not a big thing at all. Right. So it's been running for about 10 years. And this is a thing that we have been doing for the last six months. I mean, I joined in March and we started looking through this and we are basically now, we've been contacting our providers and say: look, we need you to provide some more useful and some more rigorous data evidence to back up your green claims. For this reason, because you can't... basically, it's as the stakes are got higher, it's become more and more important to actually do this. And if you're going to base decisions about how you kind of choose infrastructure from now on, it makes a lot of sense to do that. So we are heading in that direction to say, look, if you can't share this information, we're going to stop listing you. But we have given, because we're not so sure ourselves in this, we we're giving people a deadline to get this information. So you probably see some the stats change over the coming weeks as the way that we do our reporting changes. But because we released open data sets around this on a regular basis, you can actually see this. Herald: Microphone number two. Mic2: Hey, have you heard about the Science Based Targets initiative? And if yes, what do you think of them? Chris: So Science Based Targets is interesting because they are one of the drivers to basic... So if you're not familiar what Science Based Targets are: Science Based Targets are basically a way to say, well, if you're an organization and you want to hit net zero, the science dictates that you need to take these steps here. I actually think they're better than nothing in a lot of cases and I think they're probably one of the more effective things to use. And they also insist that you do need to understand emissions in your supply chain. So I imagine organizations that sign up to Science Based Targets will come on pump against the problems that I've just explained about, trying to get numbers from the large companies who tend to be coy about sharing this stuff. I think it makes total sense at corporate level. If you're not a state body. But I feel that the legally binding targets that are now in place in the U.K. and we are likely to see in Europe in the next six months to a year, would be greater levers, because they provide a degree of certainty for people to then justify decisions. Because it's the law now, rather than being a thing that you might get unemployed for. Herald: Question from the Internet. Signal Angel: You mentioned that video on the Internet is a large majority of the energy usage. Can you say something about how this breaks down to encoding, storage, transmission and decoding? Chris: Yes. Those numbers, I understand, are all about just the transfer. So I don't think there is much about the encoding parts on that. That's just sending. Signal Angel: And quick follow up: can you approximately say how much people can save by, for example, staying on single definition or SD instead of HD? Chris: I... was it four times? I don't know what that the change in resolution would be and it's not something I feel comfortable showing numbers on, because I basically be making up on the spot. I feel that the solution is telling people to not do something like this - I think it's really, really hard ask and seriously, speeding? Getting off fossil fuels is a much better way to solve this problem than telling people they're not allowed to watch Netflix ever again or only in low resolution, for example. Like we like technology because we're like 15000 people here because we like technology. Telling everyone you don't get to use technology anymore is gonna be much, much harder to sell than just use green power and stop running fossil fuels. Herald: Okay. Last question goes to microphone number one. Mic1: Hello. I work for a company that... Herald: I've said something about affiliations and introductions. Stage your question? We're pretty much out of time. Mic1: If a company owns a lot of servers and the only solution to reduce the carbon footprint was to switch to green energy, this would increase the costs for the energy. And I'm afraid that I don't have good arguments to go to ask them to switch to green energy because of the costs. Chris: Green energy is cheaper than fossil fuel energy now, like ... so, we've seen this massive reduction in costs like a like storage has come down by 85 percent in the last say 10 years, we've seen a massive drop in renewables, like this argument is kind of being solved at that level there. It's a choice of provider or something separate. But I think in many cases it's going to be a case of choosing who you want to do that. And if you're trying to make this argument here, you can make the argument that generally people tend to want to work in companies that are not destroying the planet. And if you want to retain people or attract new people saying, hi, we're part of the solution, not the problem is a good way to present this. And that's why lots of organizations talk about kind of green credentials because it's a recruiting tool in the same way that they talk about open source or working from home or anything like that, especially as we get older and have more kids and then realize that: wow, they'll be alive when this stuff happens! Okay. Herald: Thank you so much. We're out of time. applause 36C3 postroll music Subtitles created by c3subtitles.de in the year 2021. Join, and help us!