Vlad-Stefan Harbuz
Edinburgh, Scotland
Edinburgh, Scotland
The Low Carbon Software Collective is an informal group of programmers who make targeted and measured contributions to open source software in order to reduce global energy usage.
New! We just launched the Low Carbon Software Forum, a place where anyone can discuss low-carbon computing. Join us!
Energy usage from software makes a significant impact on global energy usage. If business continues as usual, by 2040, computing will use up more than half of the emissions budget acceptable to keep global warming below 1.5°C, as it will be responsible for at least 10% of global CO2 emissions.
In 2023, Microsoft fixed a Windows defender bug that affected Firefox, and as a result saved approximately as much energy as shutting down a coal fired power plant.
Our mission is to identify inefficiencies in widely-used open source software that increase global CO2 emissions, and to send patches to correct these inefficiencies.
Right now, we're just starting out. Making impactful changes requires accurate measurements, so we're trying to identify robust ways to profile energy the energy usage of software.
Another goal is to figure out what software should be targeted. Current estimates show that data centres use 700 TWh per year and networking uses 1400 TWh per year, while other consumer devices only use 240 TWh per year. However, data centre energy usage is projected to increase at an alarming rate. We're trying to use this information to figure out which software to target.
There are various challenges that might make our goal infeasible. It may be the case that the best way to reduce computing's CO2 emissions is figuring out how to optimise the energy usage of cloud services, or creating alternatives to existing energy-hungry techniques such as deep learning. Maybe embodied energy is just orders of magnitude more important than energy consumption from usage. Maybe as energy efficiency increases, corporations will simply use more computational power, which would negate carbon savings because of the Jevons paradox. And of course, political change, and the regulation of industry behaviour that directly harms people, will have the broadest impact, and political change depends on cultural change, which is ultimately most important.
But we believe that smaller-scale change, change that's realistic for individuals to do, can also be beneficial, and we're doing our best to explore whether software-based solutions that can meaningfully reduce emissions exist.
We could use your help.
We are slowly getting started with research, spare time permitting. We'd love to hear your feedback.