The impact of the web on our environment with Eric Horesnyi
What are the environmental impacts of the web? Can we calculate it? How can we reach a more sustainable model?
Eric Horesnyi is a computer science engineer with an enviable global career behind him, including database management in Mexico, data networking in San Francisco, High Frequency Trading in New York and finance in Paris.
In the meantime, Eric has had three sons. “Parenthood gives you a different responsibility towards the future”, he told us when we asked him about his commitment to sustainability. His favourite quote? “We do not inherit the earth from our ancestors; we borrow it from our children”, from Wendell Berry. “This applies to the environment as well as to the society we are leaving them, to our actions at home and at work,” believes Eric.
Eric’s passion these days lies in IoT and Cleantech, and running his recently founded company StreamdataIO, headquartered in the tremendous environment of the Alps. We asked him about the impact that the web has on our planet and settled in for an interesting tech-heavy conversation.
Hello Eric and thank you for your time. Streamdata’s slogan is to “Efficiently turn APIs (application programming interface) into real-time experiences”. Imagining that I have no idea what that means, how would you explain it to me?
Let’s say you look at a transport application, and want to know when your bus is arriving at the stop, you expect to see updates presented to you without you having to click on refresh and wait, as with Uber and Lyft interface for example where you can see a car moving on a map. On the data provider side, it means you want to avoid having millions of people sending you requests for updates every second: you give them guarantee that they will receive updates when they arrive.
That’s what we do. We help make sure that live data — from financial markets, or open APIs in healthcare or transportation — is disseminated to its audience as efficiently as possible, in 10 minutes of code. Efficiency comes from avoiding to send unnecessary traffic: we just send differential information, and only when an update has happened.
This allows people to build dynamic applications and bots, reacting to upstream events rather than expecting for end users to refresh data presented to them. The end user sees information updating automatically, hence trust he will know whenever a change happens, as in real life. Actually, our brain is trained to consider a thing to be live whenever it presents new information at its own pace, that is one to two iterations every second.
Tell us about your work analysing the impact that the web has on our planet.
As a company contributing to the web, we wanted to analyse our impact on the environment. We believe a careful use of captors, the web and machine learning can reduce our impact on the planet to reach sustainability for future generations. The introduction of Jeremy Rifkin’s Zero Marginal Cost Society book on Commons and IoT (Internet of Things) revolution is interesting. To be consistent, we wanted to make sure first that the web itself was not becoming a sustainability problem in itself. We hence started to work with greenit.fr who had gathered a community to develop a framework for such assessment.
What are the environmental impacts of the web? Can we calculate it?
Let’s define the web as the group of equipment from end user devices (captors, mobiles, computers) to data-centres interconnected by the Internet. We can estimate it with a light version of ISO 1404x for Life Cycle Assessment (LCA), considering the impact of building, running and disposing of products. GreenIT.fr has actually evaluated its impact in terms of Greenhouse Gas (GHG) emissions, water and energy depletion. The end result is that the footprint of the web today is similar to that of 40 nuclear plants, and growing faster than any other industry.
Who is responsible for this?
The main component of the web footprint comes from end user devices. End users can, for example, decide to select devices that they know will last more than three years because they can change the battery or upgrade the memory. Developers always look for efficiency in coding. By selecting technologies that are efficient, they naturally contribute to minimizing the impact of the web. The more they know about the impact of the architecture they selected, the more efficient they will be and the better for the planet, hence our focus within the Green IT community on a common framework we can all communicate within.
How can we reach a more sustainable model?
Just being frugal in our use of technology, and careful about its impact on the environment. A sustainable model can be reached by optimising our use of resources through sharing and reuse. The web is a powerful tool for sharing, and was actually built at CERN in Switzerland just for this: sharing information. If your app knows that best transportation mode to your next destination is an autonomous car that is now programmed to come to your door in five minutes, why would you ever need your own car that will be resource-intensive to build? What is the use ratio of my car? What if we used our existing resources at 95% by sharing as much as possible, rather than wanting our own. What if each house could generate its base power and stock it? What Tesla is doing in this area is very promising.
Can a tech company measures its environmental impact? What happens then?
Yes. There is no decision possible without data. You start by assessing your impact, which can be positive or negative, and then seek to grow it or diminish it once you have understood the drivers of your impact.
In our case, we have a positive impact we can measure in terms of GHG, water and energy and would want to grow it by 2020 to erase the energy impact of a midsize city such as Manchester, England. A dream yes, but not impossible.
Anything you want to add?
While researching on the latest progress in Artificial Intelligence, I came across The Signal and the Noise by Nate Silver, and warmly recommend it to any data-driven person. Amongst other cases, he exposes a data-driven analysis of Global Warming, and the end result is that not only has correlation now been demonstrated over the last century, but most importantly causality since the 1930s.
Interview by Anne-Sophie Garrigou