Search for a report, a publication, an expert...
Institut Montaigne features a platform of Expressions dedicated to debate and current affairs. The platform provides a space for decryption and dialogue to encourage discussion and the emergence of new voices.
24/03/2021

The Environmental Impact and Potential of Digital Technology

The Environmental Impact and Potential of Digital Technology
 Gilles Babinet
Author
Former Advisor on Digital Issues

In recent months, there has been abundant media coverage regarding the greenhouse gas (GHG) emissions of digital technologies - particularly due to several fresh alarming reports. The Shift Project and the High Council for Climate Change reports both mention a significant increase in the negative environmental externalities of digital technologies. While these figures have admittedly been controversial (as explained below), they nonetheless show how little is really known about the environmental impact of digital technologies. 

Regardless, France is seeking to take up the issue without delay. On January 12, the Senate, the upper house of the French parliament, adopted a transpartisan bill aiming at "reducing the digital world’s environmental footprint". This bill is based on four main principles. 

  1. Making users aware of the environmental impact of digital technologies. This means educating individuals and companies about how to avoid digital waste and a disproportionate use of energy. In other words, this means introducing and mainstreaming the notion of digital sobriety. 
     
  2. Limiting the renewal of digital terminals. Their manufacture is the main contributor to the French digital industry’s carbon footprint. The bill particularly intends to sanction software obsolescence, to limit programmed obsolescence and to support reconditioning and repair activities by reducing VAT to 5.5%.
     
  3. Promoting ecological digital uses, by making the eco-design of websites compulsory and by creating an eco-design reference system.
     
  4. Introducing environmental regulations to prevent increased consumption and emissions from networks and data centers.

These proposals come after a report from the Senate, which attempted to estimate the environmental impact of the entire digital sector value chain - from terminals to data centers and networks. 

Obviously this is a laudable effort considering the current and future state of the world - one in which environmental concerns must take center stage. This endeavor sheds light on a sector which has not yet made environmental issues a priority (probably because it sees itself as the vanguard of modernity, and therefore unencumbered by a critical look at its negative externalities). 

The challenges of measuring the environmental impact of digital technology

Nevertheless, measuring the environmental impact of digital technology is not an easy feat. The "rebound effect" (or "take-back effect"), a renown theory stating that technological improvements increase the efficiency of using services, and thus their energy consumption, is compelling, but very hard to measure, for several reasons. 

Firstly, because the cost of data administration has been divided by a factor of around 70,000 over the course of the last twenty years (i.e., a processing and storage system that cost a million euros in 1995 was subsequently only worth about 15 euros in 2015). This principle applies to computing, storage and transport performances (as per Moore's law), and is also found, to varying degrees, in energy efficiency. For instance, a 1990s computer like the Macintosh II consumed 230 watts, on top of 205 watts for its screen, which is a total of 435 watts. It is, of course, significantly less powerful (150,000 times less transistors) than a 2016 Samsung S8 smartphone, operating at 12 watts when in use. 

In the digital world, technological breakthroughs are constant. As a result, the most sophisticated processor on the market, the Nvidia A100 tensor with its 52 billion transistors, is now in competition with the Lightmatter company’s Optronic processor (which offers 1.5 to 10 times better performance, for 6 times less energy consumption). These disruptive innovations also apply to technology architectures. A recent data center using adiabatic cooling technologies can consume 40% less energy than its predecessor using traditional digital technology.

Measuring the environmental impact of digital technology is not an easy feat. 

Accordingly, calculations which were correct at a certain point in time are quickly outdated, due to accelerated technological obsolescence. Most analysis of these technologies does not sufficiently take this into account. They therefore construct models based on more or less constant efficiency, despite this efficiency increasing in reality, thus demanding less energy. 

Furthermore, the energy source is an essential factor that is sometimes miscalculated. For example, the CO2 emission intensity of Chinese electricity is 9 to 11 times higher, and American electricity is 7 times higher, than it is in France. If a piece of equipment is manufactured in China, its carbon intensity will necessarily be tied to the emission intensity of Chinese electricity. This item’s footprint would be significantly lower if it were manufactured in France (which is rarely the case). Another example, a Netflix user’s footprint varies depending on whether the movie they are watching comes from a stream hosted by France or Norway (65 gr/KWh in France, compared with 50 gr/KWh for Norway), or from an American stream. Note that this type of company uses content delivery networks (CDN) that store the relevant content as close as possible to the end user. Thus, in France, Netflix has several data centers that meet the demand of French users, reducing the distance between server and user to a few hundred km. However, the world being what it is today, the fact remains that at least 70% of technological equipment is manufactured in China, and it is estimated that just over 50% of data centers are located in the USA. This will likely evolve, particularly as environmental issues are gradually taken into account. 

Finally, it should be pointed out that equipment depreciation strategies have a major impact on the digital industry’s energy consumption. A leasing company will seek to boost its offer slightly and, in return, provide accelerated renewal of the equipment it supplies. Similarly, a telecom operator may decide to stop subsidizing terminals, and optimize its equipment replacement by taking carbon externalities into account - which only few are doing as of yet. Since the carbon footprint of manufacturing (including usage) represents between 75% and 95% of the total footprint (the latter figure pertaining to certain passive equipment, e.g., optical equipment), it is obvious how much we would benefit from considering the aforementioned approaches.

These considerations highlight the intrinsic complexity of these issues for the digital sector. They cover particularly sophisticated technological, scientific and economic aspects, which are often prone to methodological errors. While the complexity of the subject is significant, it is not necessarily inaccessible. Giving credence to recent studies, some of which indicate worrying figures regarding digital externalities, it is essential not to facilitate an activity that would develop at the planet’s expense. With this in mind, understanding the transformation of production models is crucial.

The positive externalities of digital technology is an important issue that has been largely bypassed in recent studies.

Generally speaking, the massive individual development of digital technologies (more and more of us are connected and increasingly multi-connected) has a structural consequence on the increase in the direct environmental footprint of digital technology. This plays out on the purchase of terminals, on the development of digital infrastructures, and to a lesser extent, on usage. Nevertheless, the positive externalities of digital technology is an important issue that has been largely bypassed in recent studies. Moreover, this point is subject to methodological weakness, since - no doubt intending to make their work more spectacular - many authors have not hesitated to highlight the most problematic uses of digital technology in their studies. The example of streaming, mentioned above, is striking. The authors of the Shift Project's first report claimed that it accounted for 1% of total CO2 emissions - a figure that they later admitted was false, with more recent studies suggesting a footprint 22 to 57 times smaller. 

Positive externalities are manyfold and frequently underestimated. For example, the European Commission reports that the European Union has seen a 14% increase in truck loading efficiency in the last 15 years, due to the development of integrated information systems in the logistics chain. Many nations are seeing a significant drop in fuel consumption per kilometer driven on their roads. The significant quantity of connected GPS systems, which help avoid traffic jams, is probably a major contributing factor. Following the same logic, the French government has been providing financial incentives and assistance to install connected heaters (which go on standby when no one is home), because their efficiency compared to their cost is unbeatable, and so on. The case of home-office, in the context of the health crisis we are enduring, is also reflective of this point, as it has considerably reduced the use of highly energy-consuming means of transportation. 

In fact, in a finite world, in which resource exploration generates significant negative externalities, making information widely accessible and usable is one of the most effective ways to reduce our environmental footprint. It allows us to synchronize needs with supply, and flows with infrastructures, at all levels of the production chain. Information technologies are developing at the right time, when the aim is no longer to produce more, but to produce better

These aspects are overlooked and understudied. For instance, legume farming easily involves more than a hundred variables, some of which cannot be controlled (e.g., temperature, humidity). Yet, optimizing these variables can have very significant environmental repercussions on the quantity and quality of production, but also on greenhouse gas emissions (GHG) and other environmental externalities. In any case, machine learning could quickly become a powerful auxiliary in managing these multivariate environments and in optimizing the environmental requirements of these activities. 

The digital revolution is responsible for a paradigmatic shift of at least the same magnitude as those that propelled us into the modern era. This revolution is multifold. It is firstly anthropological, altering our relationship to space, the nature of our social interactions, our psyche, etc. It is also economic, as those who traditionally dominated the 20th century have been replaced by digital players. Finally, it is geostrategic, as the preeminence of the State pre-eminence is no longer as clear-cut, threats are evolving, as are their instigators. We should therefore view the challenges we face in a new way. The aim is not to promote an ideological position, but to grasp the nature of this revolution in order for it to benefit the greatest number of people, and to allow it to become an auxiliary to the environmental revolution which, without a doubt, represents the greatest challenge of the 21st century. 

One of the likely aims is the transformation of the digital industry, freeing it from its enslavement to mind-numbing consumerism and applying it to the fundamental challenge of the century. This means designing more virtuous systems, including equipment and software, and fostering the skills needed to bring together human activity, the environment, and digital technology. The dual challenge of the digital revolution’s acceleration and the imminent threat posed by climate change is a powerful incentive. 


Copyright: Photo by Claudio Schwarz | @purzlbaum on Unsplash

Receive Institut Montaigne’s monthly newsletter in English
Subscribe