IT, DevOps, and Corporate Social Responsibility (Europe 2021)

Environmental Sustainability, Diversity Equity and Inclusion, Community Engagement, Stakeholder Capitalism - all of these aspects of responsible corporate behavior have become increasingly strategic, often urgent concerns of boards of directors. IT organizations have not yet come to terms with the critical role they play in planning, governance, and execution of their companies' social good efforts. In this session I'll show how critical that role is and provide some ideas and frameworks with which technologists can "engineer" solutions to these challenges.


Mark Schwartz

Author, The (Delicate) Art of Bureaucracy



Hi everyone. My name is Mark Schwartz, and I'd like to talk today about it and corporate social responsibility or CSR. It's been quite exciting actually to watch all of the activities around corporate social responsibility lately, there've been some determined efforts around environmental sustainability, diversity, equity, and inclusion, ethical sourcing, and corporate ethics. In general. It raises the interesting question of what role we do or should play in CSR as technologists, as it professionals. And as a dev ops community


Dev ops is built on a very extensible and flexible model. Really it's the idea of fast feedback of delivering something, monitoring how it actually works and what results it has, and then taking that feedback and feeding it back into the process to continuously improve what we've delivered, because this is such an extensible model. We've seen the rise of things like dev sec ops dev biz ops, fin ops, um, all of which really have to do with picking this basic fast feedback model and adding to it additional sorts of requirements or additional sorts of guardrails. What I want to suggest today is that we can actually take this model and create a, what you might call. I hate to say it that was sustainable ops or dev social ops by adding social responsibility requirements into our cycle. And by putting a place guardrails around social responsibility, I think it even more interesting, it's not this, this model of fast feedback is something that we can also apply to other corporate social good efforts, even outside of it.


So that's what I'll be talking about today because we have limited time. I'm going to limit myself to just a few areas within CSR. Although there's a lot more we could talk about, I'll be talking about environmental sustainability, diversity equity, and inclusion, ethical sourcing, and I'll touch on some issues around ethics of technology. Who am I to be talking about CSR? Well, I'm an enterprise strategist at Amazon web services. And that means my, my role is to work with leaders of large enterprises to try to help them overcome some of the non-technical impediments to digital transformation. So often that's things like cultural change, organizational structure, organizational politics, governance models, investment strategies, and my favorite bureaucracy. While, uh, while doing this in my spare time, you could say, I started to poke around AWS to see what we had going on in the way of social good.


And I became involved in some of our efforts first around homelessness and then around re-skilling unemployed and underemployed populations and trying to place them in jobs. And while I was doing this, uh, I became aware that companies now we're very interested, especially at leadership levels all the way up to the board level or interested in what they should be doing around social responsibility. And since my role is to advise companies on the things that are important to them, I started to include some of these ideas, some of the things that I'd been learning around social responsibility from my other activities, I should define my terms because this can be a little confusing. I'll start with corporate social responsibility or CSR. That means exactly what it sounds like. Stakeholder capitalism is the idea that businesses need to be, uh, addressing the needs of a broad variety of stakeholders, not just their owners.


So for example, their workforces, the communities in which they operate regulators and so on ESG is, um, making the next step forward from those two. The idea is, uh, it stands for environmental, social and governance. And the idea is to put in place the controls, governance, uh, process frameworks, incentives, training, all of those things around CSR to make sure that the company meets its obligations are around those things. It's been, it's been becoming a major concern, as I said, for companies and for their boards of directors. So 45% of board members now say that ESG has become part of their agendas. 82% rank being a fair employer and good corporate citizen is very or extremely important. Large companies. I'm talking about companies whose market cap is greater than 25 billion. 56% of them actually have ESG measures in their incentive plans.


And 90% of S and P 500 companies now publish a sustainability report. That's a big increase over the last few years. ESG is becoming important to private equity investors, venture capitalists. And one reason for this is that the data is showing these days, that companies that are good at environmental and social governance are also good at governance in general and return good results. S and P 500 has an ESG index, which has actually outperformed, suffered fewer losses and recovered faster than the S and P 500 as a whole during the COVID pandemic. You might wonder what technology has to do with some of these social problems, uh, as I did at first. So let me, let me tell you about something that I learned in the course of, of researching homelessness. Uh, if I, I didn't realize you might not realize that actually there's a good deal of permanent supportive housing available for people who are experiencing homelessness in, uh, in major municipalities.


And, uh, I'm talking about permanent supportive housing, not, not shelters, but actually, uh, semipermanent lodgings, and a lot of this permanent supportive housing actually is unoccupied. Why would that be? Well, it turns out that in order to establish their eligibility for this housing, the people experiencing homelessness have to produce a number of documents to show that they're eligible. So this can include things like a birth certificate, social security card, uh, proof of their income or lack of it. And, uh, of course, as you realize people who are sleeping rough on the streets, typically aren't carrying these documents with them. So it can take six to 18 months to assemble all of these documents. And in the meantime, the housing is sitting on occupied. Now this, this is a data problem. Why couldn't we, for example, directly interface the it systems of the social security administration with those of the public housing authority.


So the social security administration could just assert that this person does in fact have a social security card or barring that why not keep copies of scanned versions of these documents available so that they can be produced whenever the individual needs to establish eligibility for, for something. So, uh, that's actually what we did. We created a, um, a scanned document, uh, library on behalf of people experiencing homelessness. We did this working with the Baltimore city government it's available for other municipalities as well, but the potential for alleviating homelessness with data goes far beyond that. Typically in a continuum of care, a, a local, um, set of organizations that is working together on, on homelessness. Um, it requires the coordination of a lot of different types of services, food kitchens, shelters, medical services, dental services, substance abuse, counseling, employment placement, and the potential benefits of being able to share information easily between all of those organizations is tremendous.


But it's very hard to do today to give you a, another example of how technology can apply in areas that you might not expect it over a million children each year are reported missing. There's this organization, the international center for missing and exploited children that has a website where they publish a brief snapshots of 6,000 or so missing children. But in 2018, they started a new initiative using the AWS cloud, where they combined machine learning and advertising technology to start scouring the dark web and the clear web looking for images of the children they know are missing. And when they find a hip, they can then pass on the lead to law enforcement. And I'll give you one more example here. Um, AWS has a disaster response team and when there's a natural disaster, uh, earthquakes, hurricanes, flooding, wildfires, things like that, they go to the scene of the disaster and they help local governments and local nonprofits re-establish their operations and, uh, and conduct rescue operations.


So for example, they might come with our very ruggedized compute and storage devices called the snowball edge. And, uh, it might be filled with detailed images of some of the area that is covered by flooding. It might have maps of the area and they can then fly drones around to update the maps and using this information, the local authorities can target their rescue efforts better. So I wanted to mention these just as a few examples of why technology can be so important in social good efforts and, uh, the potential it has. So now I want to go into the four areas that I, that I mentioned before. And first we'll, we'll look at environmental sustainability, and I think there were just a few numbers that help me understand the story behind this. First of all, the annual carbon emissions of the world these days are around 42 giga, tons of carbon dioxide scientists tell us that the atmosphere can absorb a total of about 400 more gigatons before the temperature changes more than 1.5 degrees centigrade, uh, which is a threshold that they've said this was as of 2017 and some time has gone by since then.


So at this point, there are less than seven years at the current rate before this, this limit is reached.


So what can we as technologists do about this? Well, my first answer, I'll give you a few. The first one is kind of self-serving, but I will say it anyway. Let's get into the cloud. When a workload is moved from a corporate data center into the cloud, typically its carbon footprint is reduced by about 88%. That's a very big number. How is it possible? Well, the first thing to realize is that servers in corporate data centers are generally operating at about 18% of their total capacity. So in the cloud, we can get much better at capacity utilization by combining workloads on servers. We've also designed our infrastructure to be much more energy efficient. And we work with our electricity providers to, uh, provide electricity from renewable sources. We generate a lot of our own electricity from renewable sources. We have around the world, 127 solar and wind farms.


Of course, it's not just about carbon emissions. We have to look at other sorts of emissions. We have to look at water usage, um, recyclable packaging and so on. And we hit, we have efforts going on around all of those things, but there's this immediate impact of moving to the cloud. The next thing I'll mentor mentioned is that we can actually engineer for better to sustainability. So if you go back to my dev ops diagram at the beginning of the presentation, I said, we can incorporate social responsibility requirements into our process and guardrails. Well, it turns out that we can actually learn to engineer code in a way that leads to better sustainability. I'll give you a few examples here. There's so much more, uh, but just to give you an idea of what I'm talking about, first of all, there are serverless computing, which is a big potential impact where you can use event-driven asynchronous processes so that you're not wasting idle time on servers that are still running and consuming, uh, electricity. You can also engineer to choose the most energy efficient instance types in the cloud. You can optimize your network routing to reduce the burden on infrastructure of routing, and you can reduce your infrastructure, your storage needs by using more efficient file formats like parquet. For example, again, there are lots more and we're hoping to release some more guidance for engineers on how they might do this.


Another area where technologists can be involved is in producing transparency and data for your entire companies to use in becoming more sustainable. And that can be a combination of internal data and external data sources. We've tried to make data available at AWS through our Amazon sustainability data initiative. Uh, we also have something called the AWS data exchange and an open data program. But, uh, what I'm thinking of here is, for example, if you had data on temperature inside your facility is in data on temperature, outside your facilities, you could probably do a more efficient job with your heating and air conditioning. Or for example, if you knew the location of your delivery vehicles and the conditions around them, you can maybe optimize their routes so that they use less fuel. A really good example of this is Quantas airlines, which developed a simulation, a flight path simulation, essentially, where they could figure out the optimal flight path for, for their aircraft.


And by doing that, they found they were able to number one, save a lot of money, $3.2 billion in just 2019 and fuel costs, but they also found a two and a half percent reduction in their fuel consumption, which if you think about it as a really big impact since they, uh, use a lot of fuel, as you would expect, um, let me move on and talk a little bit about diversity, equity and inclusion. One of my colleagues and I published a blog post on this recently. Uh, so you might want to refer to that, but, uh, the question we posed to ourselves is what can it organizations technologists be doing to help solve diversity, equity and inclusion. And I think the first area that we have to look at, of course, it's sort of a no brainer. It's the composition of our dev ops teams dev ops thrives when there is diversity, diversity of background, diversity of thought, and diversity of personality is on a team.


It was built on the idea of a kind of diversity of technical diversity, putting developers and operators together. But the potential of having truly diverse teams is tremendous. I'll quote you a couple of stats that I found quickly. Um, Harvard business review article found that diverse teams out innovate and outperform others. There are 45% more likely to grow their market share. There are 70% more likely to capture new markets, a study by Deloitte and Bersin said that companies that rate high in diversity, um, are 1.8 times more likely to be changed ready. And our 1.7 times more likely to be innovation leaders in their markets. So as with the fuel cost savings before increasing diversity, it can have a powerful impact on the business in general, besides being just the right thing to do.


I think an interesting thing to think about is this question of what would happen if we took our dev ops mindset and applied it to engineering for diversity. So, so what I'm thinking is this, let's say you realize that your company is not diversity enough. Well, let's, let's think about that as a defect essentially, and that's a defect that must be remediated. So how do we handle things in the DevOps world, in the digital world? When we have a requirement that's a new requirement to us? Well, uh, usually what we do is we, we, uh, construct hypothesis for what might solve that problem. And then we test those hypotheses and really double down on the ones that seem to be effective. So let's say you have a hypothesis that you're not recruiting a diverse pool of talent. Um, and, and your hypothesis is that maybe your job descriptions are inadvertently, uh, uh, excluding certain people.


Okay. That's, that's a hypothesis and you've probably have other hypotheses about what's happening. So why not rig up tests and minimum viable solutions, let's say. And then as you find which hypotheses are moving the needle the most, then you'd double down on them and you continuously improve whether your fast feedback cycle and then, uh, analytics. This is the, uh, data oriented data based approach to improving equity and inclusion. And, uh, my colleague gave me a good example of something he'd seen in one of his previous companies. He said, uh, the company had a mentorship program. And when they looked at the data and poked around a little bit, they found that African Americans were not participating in this mentorship program as much as they would expect. And so that caused them to do a little bit of digging and looking around. And what they realized is that the, um, the leaders of the mentorship program, uh, in a very well-intentioned way had set things up so that African-American mentees generally received African-American mentors.


But the problem was there weren't very many African Americans in higher levels of leadership. So there were very few mentors available. So here's a case where looking at the data, um, triggered something, uh, made them aware of a situation, and then later on how to solve it, ethical sourcing. So you, as a customer of your suppliers have a lot of power customers generally have power. You can try to insist on high standards from your suppliers. You can change suppliers when necessary. Sometimes you can even write conditions into the contracts that you're assigning with suppliers. So we do at Amazon, uh, here's a list of some of the things that we're looking for from our suppliers. And, um, we conduct a ballot 4,000 audits a year both to make sure we're choosing the right suppliers and also to help them with feedback, to improve their own practices.


An interesting example of supply chain management from this perspective comes from Nestle, where they've started using an AWS managed block chain solution to try to ensure the integrity of their supply chain traceability, and, uh, to make sure that the products that they're sourcing actually come from the parties contracted with rather than from other parties who for example, might be contributing to deforestation. So this is sort of a chain of custody application of blockchain to ensure that the supply chain is performing as it should. So I mentioned before that you should be looking at your suppliers and making sure that they are meeting your ethical standards. And I include cloud providers there, AWS included. So I owe you a little bit of an explanation of what we're doing at Amazon on sustainability with such a huge worldwide operation. So here's a, here's a scorecard on our emissions, our carbon footprint, and it's divided in the usual way between scope one, scope two and scope three.


You mentioned so scope one emissions are the ones that result directly from our Amazon operations scope. Two has to do with our purchased electricity, the carbon emissions that are involved in generating the electricity that we use. And then scope three is all of the indirect sources of carbon emissions through our supply chain and related activities. And as you can see, a total is just over 50 million metric tons a year. So what are we doing about this? Well, our commitments now are our climate pledge, which pledges to have net zero carbon emissions by 2040. This is 10 years ahead of the Paris Accords by 2025. We're going to be 100% based on renewable energy. That's just 40 years from now in 2030, we're going to be, or 50% of our shipments are going to be net zero emissions. We're also investing in several funds to support sustainability initiatives.


Uh, this, this thing about being 50% shipment net zero gives you a little bit of a sense of the complexity of the, of the undertaking. What we're trying to do is for packages that are shipped by Amazon to be net zero carbon emissions all the way through the entire process. So, uh, in our fulfillment centers where the orders are packaged, um, all the way through to the delivery to people's homes. So if the fulfillment center is it's the packaging and it's that last mile delivering. So our fulfillment centers now 60% of them, uh, how solar panels on the roofs and those solar panels are able to generate about 80% of the electricity requirements for the fulfillment center. Uh, in terms of the delivery itself, we have purchased a hundred thousand electric delivery vehicles about 10,000 of which should be on the road this year. Um, so when we're talking about, um, becoming environmentally sustainable, we're talking about that big breadth of activities, scope one, scope, two scope, three water packaging, uh, and all of these other things.


And then quickly let me touch on ethical technology. And, um, here I I'm, I'm going to hit a few points that you might not have been thinking about in terms of responsibility. First of all, there is product quality. So applying, uh, your customers with products that actually work is an important thing and a matter of responsibility, I think it can be especially important, for example, uh, in in fact it can, it can lead to dangerous if you're not responsible about it and things like the software that runs medical appliances security in general, um, is a social responsibility. I think, um, Josh Corman and his organization, I am the cavalry ha have done a very thorough job of presenting the importance of product security, um, and talking about some of the things, uh, that are vulnerable medical devices. As I said, um, systems in vehicles, cruise control and collision avoidance, uh, things like cameras and microphones in your home security systems, um, this security, um, ultimately is something that you owe to your companies and to your customers.


Um, privacy, of course, uh, you certainly owe that to customers. And so I think we're, we're all moving to an approach to privacy that involves privacy by design thinking it through carefully in advance and, uh, making sure it's designed into everything that we do. And I'm finally, I, I, can't not mention artificial intelligence, but I'm not really going to say anything about it because I'm really not the expert on this, except to say that I think we're all realizing that this is something that we have to take very seriously now. Um, the ethics around the use of machine learning and artificial intelligence, and, uh, it's important to, to pursue this and learn more about it. I think so, uh, to sum it up to recap here, I think there's this amazing potential to drive social change, a social responsibility from the it organization and from the dev ops community. I think we can be doing better with environmental sustainability and diversity equity and inclusion, uh, pushing change throughout our supply chain and making sure that we're handling the technology itself in an ethical way, and that actually technology organizations can and should be the drivers within their broader enterprises, have a lot of these things. So, uh, please think about it and please think about what you can do to support the efforts.


I do have a few areas where I can use some help, um, and let me quickly go through them. First of all, if you got other ideas on how technology organizations should be supporting ESG and CSR, please let me know. Uh, especially if you have some good examples of companies that are doing it right, that we can use. When we're talking about this with the rest of the world and any input you can give me on how cloud providers like AWS could be better supporting your company and its social responsibility efforts. So please join me in, uh, in advancing, I hope social responsibility from the technology point of view, but from the technology part of the organization and helping our companies act responsibly in their markets and with their workforces. And thank you very much.