One thing which occurred to me while in the process of migrating the GWF tcs.json over to DIST is the issue of representing "absence of evidence" vs "evidence of absence":
In the TCS schema, we explicitly include every emissions category, including those which are not measured:
"indirect_emissions": {
"offsite_employee_hardware": {
"emissions": 0,
"notes": "Not measured, new category since we used TCS"
},
This is not ideal, as it becomes impossible to tell programatically which zero values are un-measured, and which are measured, and result in zero emissions. Compare with:
"server_hardware": {
"emissions": 0,
"notes": "Not applicable."
}
In the case of "server hardware", our emissions are measured at zero (as we have none), in the case of "indirect emissions" our emissions are not claimed to be zero, but rather an unknown and unmeasured quantity.
Happily in DIST, given there's no requirement to list every taxonomy category, we can avoid this problem - we simply do not list categories which we have not measured:
"impacts": [
{
"dimension": "carbon",
"categories": ["upstream:server_and_storage_hardware"],
"value": 0,
"unit": "kgCO2e",
"confidence": "high",
"notes": "GWF does not own any of it’s own servers or storage hardware. This means there are no carbon emissions to report in this category and we’re highly confident about this.",
}
/* employee hardware, which is unestimated, simply not included */
]
However, I suspect this should be formalised and documented somewhere: "If you have zero impact in a category, you should list that category in your DIST file, along with the method used for calculation and a justification if appropriate, whereas if you have not measured or have no data for a category, you should not list it in the file".
@cronky @hanopcan thoughts on this?
One thing which occurred to me while in the process of migrating the GWF tcs.json over to DIST is the issue of representing "absence of evidence" vs "evidence of absence":
In the TCS schema, we explicitly include every emissions category, including those which are not measured:
This is not ideal, as it becomes impossible to tell programatically which zero values are un-measured, and which are measured, and result in zero emissions. Compare with:
In the case of "server hardware", our emissions are measured at zero (as we have none), in the case of "indirect emissions" our emissions are not claimed to be zero, but rather an unknown and unmeasured quantity.
Happily in DIST, given there's no requirement to list every taxonomy category, we can avoid this problem - we simply do not list categories which we have not measured:
However, I suspect this should be formalised and documented somewhere: "If you have zero impact in a category, you should list that category in your DIST file, along with the method used for calculation and a justification if appropriate, whereas if you have not measured or have no data for a category, you should not list it in the file".
@cronky @hanopcan thoughts on this?