jmcook1186
commented
3 months ago Hi - just to address some of your questions in the original ticket:
Regarding judging criteria "Synthesizing - How well does the model integrate and combine information from existing research? Are any coefficients, methodologies, or techniques backed up with good citations? How can we trust the outputs of this model are correct?" - how might we make our output as reliable as possible? What can we do to best meet this criteria?
A: This is kind of up to you to determine - some ideas might be being as comprehensive as possible with citations for coefficients that you decide to use, showing where your work has used insights from published papers or other trusted orgs, being explicit about where and why you have deviated from established norms from other products, etc. It is content dependent, but it's about providing evidence that your solution is trustworthy and likely to be correct.
Also, we're currently only focused on the land used in energy production. If you have any ideas on how we can include the land used in hardware production (i.e. all of the land used in the life-cycle of hardware) in our calculations that would be really interesting. It would also be interesting to factor in the land used by data centers and other internet infrastructure. We haven't found any relevant existing research on this, but if you do please send it our way!
We don 't have any specific references to share with you - our remit is really developing the framework itself and we want the community to come with new plugins based on their research. We'd love to see what you find!
Prize category
Beyond Carbon
Overview
We intend to create a model that will estimate the land-use of any piece of software.
Land conversion is one of the nine ecological ceilings. Land use change is impacting biodiversity, food security, and climate change.
We intend to consider the energy usage of the software and land-use intensity of the grid to calculate our output.
Questions to be answered
Regarding judging criteria "Synthesizing - How well does the model integrate and combine information from existing research? Are any coefficients, methodologies, or techniques backed up with good citations? How can we trust the outputs of this model are correct?" - how might we make our output as reliable as possible? What can we do to best meet this criteria?
Also, we're currently only focused on the land used in energy production. If you have any ideas on how we can include the land used in hardware production (i.e. all of the land used in the life-cycle of hardware) in our calculations that would be really interesting. It would also be interesting to factor in the land used by data centers and other internet infrastructure. We haven't found any relevant existing research on this, but if you do please send it our way!
Thank you! Happy hacking everyone!
Have you got a project team yet?
Yes and we aren't recruiting
Project team
@JonathanBell09 @corscada @rewildchris
Terms of Participation
Project Submission
Summary
We have developed the first tool to estimate the Land Use Intensity of Software (LUIS).
We’re bringing the cloud, back down to earth.
You may be asking, how does software use land?
Land is needed to produce the energy our software consumes and the hardware our software uses.
Our energy technologies, for example, solar photovoltaics, coal power, and hydropower, can use huge amounts of land throughout their lifecycle.
Problem
There is currently no tool to estimate the land use of your software.
Why should we care about land use?
Land Conversion is one of the nine ecological ceilings. Land use change is impacting biodiversity, food security, and climate change.
A significant amount of land is needed to produce the energy software needs. Using our new methodology, we estimate that the entire internet uses an area around the size of Belgium each year (at least).
Millions of species are threatened with extinction, and habitat destruction is the leading cause. We need more land for nature.
With IT energy consumption increasing rapidly, and more land being used each year, it’s time we reversed this trend.
Software will eat the world if it continues on it's current path.
Application
Use our plugin to estimate the land-use intensity of your software.
Input: Energy (kWh) [Required], energy source mix [Optional], zone [Optional]
Output: Land Use Intensity (m2/year)
Land Use Intensity of Energy (LUIE) is the area of land that needs to be used for a year to produce 1kWh of energy (m2/kWh/y). To calculate LUIE we consider the Land Use Intensity of different energy sources (e.g. coal power, solar photovoltaics, etc.) and the mix of these energy sources in the grid.
Our primary source for the land-use intensity of energy sources is a UN paper which details the direct and indirect land use throughout the lifecycle of each energy technology.
We get the default global mix energy mix from the Energy Institute’s annual report. Users can optionally use electricity maps to use current (and region specific) energy mix data, or pass in their own energy mix.
The output is the area of land that needs to be used for a year to produce the energy needed by your software (m2/y).
Prize Category
Beyond Carbon
Judging Criteria
Overall Impact:
Organizations can now achieve a broader understanding of their environmental impact. They can report on and reduce the land use intensity of their software.
It’s easy for anyone to use and could be widely adopted. Software practitioners can start using it now. We are already measuring the energy usage of software to calculate CO2 emissions, now we can use that same data to estimate land use intensity.
Educational Value:
This plugin educates software practitioners on their impact on planetary boundary ‘Land Conversion’.
It’s easy for anyone to use and understand, and we hope it will spark interest in learning more about how our software impacts the natural world.
Synthesizing:
See our sources in our documentation. Our data is primarily based on a 2022 paper published by the United Nations. To get the default energy mix we used the latest Energy Institute annual report. We also integrated the electricity maps API so that users can specify region and get latest data.
We acknowledge that estimating land use intensity of energy is an emerging and developing new research area. We need more data to improve the accuracy of our results. The outputs of the plugin are estimates.
Video
Video Submission
Artefacts
Github Repo
Usage
Example Manifest
Process
Inspiration
This project was born out of our passion for protecting nature. For years we have been interested in examining how software impacts biodiversity. When we discovered a recent paper on the land-use intensity of different energy sources, we quickly realized that this is the link between software and nature.
Challenges
Our biggest challenge has be the limited amount of data. While we found reliable recent papers on the land-use intensity of energy sources, we could not find reliable datasets on the life-cycle land usage of associated hardware, and there was no transparency on the land-usage of data centers and IT infrastructure.
Accomplishments
Learnings
We discovered how challenging it is to measure biodiversity. When we set out we wanted to directly measure software’s impact on biodiversity. We soon realized that there are few standard approaches to measuring biodiversity, little data, and huge variations depending on region. To be truly accurate we need to align with the pulse and life of the eco-systems that depend on the land areas themselves. This means that we need to have a sense of temporal as well as spatial. One measure is not enough this needs to be continually monitored. We also learned that land is more complex than reducing it down to absolute grid (or hexagon measures). Context is important in terms of north/south divide as well as placement of a power source(s) on a parcel of land. Even with hexagons land is not regular so some form of offset for irregular land use can/should be factored in.
What's next?
Develop and mature our approach:
Beyond Land Use: