Technologists want to lower carbon emissions with even more code
By Melanie Ehrenkranz
Much of our digital lives are in the cloud, a seemingly abstract digital space. But it’s not abstract—while the data is intangible, the storage centers where this data lives are very real in the physical world. They have a carbon footprint, a truth opening the door to new sustainability solutions.
“So far a huge amount of effort has gone toward trying to figure out how we design better algorithms to predict who should get a loan, or who should get bail, and is it ethical to have machines making these decisions,” Daniel Larremore, an assistant professor in the Department of Computer Science at the University of Colorado Boulder, told me last year.
“But there’s this other ethical component about the externality of computation itself: by doing so many computations, are we putting a lot of carbon into the atmosphere that wouldn’t otherwise be there?”
Over the last few years, the leading providers have taken steps toward cleaning up their clouds. It’s an ongoing contention—figuring out how to efficiently house hoards of data without radically fueling the global warming crisis-and climate technologists are heeding the call.
“It should be a commodity to understand the footprint of what our decisions are, whether as humans or as companies,” Olivier Corradi, a machine learning engineer and data scientist that created Electricity Map and founded Tomorrow, said.
Electricity Map is an open-source mapping tool that color-codes regions based on their carbon emissions, specifically in electricity production.
The map was built a few years back with just a few countries on the map. Corradi said they now have more than 200 unique contributors and thousands of contributions. “One person says it feels like taking a brush and painting over the world, turning gray countries into brown or green.”
Over the last few years, the leading providers have taken steps toward cleaning up their clouds.
Google uses this project to optimize their data centers worldwide, according to Corradi, tapping into the data when making decisions on when to train energy-intensive artificial intelligence models. (A paper published in June of last year by researchers at the University of Massachusetts Amherst found that training one AI model has the same carbon footprint as the lifetime of five average American cars.)
Google has some wiggle room when it comes to scheduling their cloud computations, and Corradi’s team provides them with the data-driven insights needed to make climate-friendly decisions.
For instance, looking at a 24-hour forecast, Corradi’s team might find that the least carbon-intensive time is in two hours when the wind is blowing more and the sun is shining. That’s because wind and solar plants might take over for coal-powered ones at that time.
Google has the flexibility to pick the ideal time and data centers based on when dirty fossil fuel plants are less likely to run and when, in general, less people are consuming electricity.
“I’m very bullish that all cloud players should be offering this as a service to cloud users to make sure big companies can schedule their workload at the optimal time,” Corradi said, “not only Google, but Google customers can help reduce emissions.”
“I’m very bullish that all cloud players should be offering this as a service”
Ethics and sustainability are relatively new drivers among idealists in Silicon Valley, and hardly a mainstream principle yet among the likes of manufacturing-heavy factories and telecom infrastructures.
Corradi said that they’re seeing digital tech companies like Google, Stripe, and other small and medium-sized enterprises in the tech industries focusing on sustainability because in general, you have younger entrepreneurs in these spaces guided by ideals and that know climate change is an issue.
These companies are also competing more aggressively on talent, Corardi notes, “and for those with the luxury of picking the company they can work for, it [climate change] becomes a more important issue as time goes on.”
Austin Whitman, the CEO of Climate Neutral, noted that not all companies have a large (or any) budget to easily and cheaply measure their carbon emissions—that the alternative to these tools might be hiring a consultant to spend months looking at their operations and putting together a custom model for more sustainable business practices.
“That process is prohibitive for many companies,” Whitman said. “We thought if we could build something that is useful to people, maybe we can accelerate the rate at which companies learn where their carbon emissions are coming from and start to do something about them.”
Whitman said that if you get these tools into the right hands, like a sustainability marketing or operations person, it’s a powerful catalyst for change.
He specifically cited one of Climate Neutral’s users, a sustainability manager from a large, recognizable footwear brand. “Her day-to-day is a little bit unstructured, she’s got very little resources to hire consultants or invest in software, so a lot of what she does is just try to work internally on building support for new initiatives,” he said. So with access to a tool like this, she can access the data she needs to come up with a carbon emissions estimate, print out some data visualizations, and talk to leadership about what they can do better.
“Even simple observations like, most of your carbon is coming from your supply chain, that can be helpful and eye opening,” he said.
“If a company has no idea what their impact on climate change is, they can’t take meaningful steps toward doing anything about it. A user that is engaged and under-resourced is pretty much the sweet spot.”
Tristan Mallet, the associate director at BCG Gamma, echoed that adoption of these tools is powerful at the individual level, because at the end of the day, these engineers are the ones structuring the company’s code. “A corporation or organization is a sum of its parts,” he said.
BCG Gamma along with a number of other AI experts just deployed CodeCarbon, a software package that estimates the carbon emissions from the computing power needed to execute code. The tool helps developers understand how to make more energy-efficient coding and cloud infrastructure choices. Mallet said that the tool can be implemented in about five minutes.
It’s misleading, though, to imply that these corporations are a closed loop when it comes to creating more energy-efficient operations. No single entity holds all the power and influence. Corradi says they work with three different types of actors.
“We have the citizens that are asking questions like, what can I do to change the world and behave in a better way? You have policymakers and politicians in general, they can pressure individuals and corporations to do the right thing through regulations. And the third factor is corporations. They are pressured by regulation, but also need to sell to their end users. In this interconnected triangle, there’s not a single one that really trumps the other.”
The tools of course are just one step one of these entities can take toward building systems that take into consideration their impact on the environment. Simply using the tool won’t automatically change a corporate strategy or redirect infrastructure choices or share with the public information they can use to make sustainable purchasing choices.
The tool is just knowledge, it’s what companies do with that knowledge that matters. Knowledge precedes action, and these tools are banking on that.
Melanie Ehrenkranz is a writer with a focus on tech, culture, power, and the environment. She has been featured in Gizmodo, Vice’s Motherboard, Medium’s OneZero, National Geographic, and more. You can follow her work here.