describe methodology

[potash]

we should describe the methodology somewhere. probably on istheweatherweird.com. by methodology i mean:

• deifnitions of typical/bit weird/weird/very weird. including an acknowledgement that these terms are arbitrary and that's why we also give percentile.
• data sources (these are currently listed on website).
• maybe mention the minimum coverage thing i just did
• acknowledgement that the world is already warming and so our baseline for weirdness the post-warming world. this is a departure from many other definitions of temperature anomaly that use a pre-1970 or pre-1980 or pre-industrial revolution reference period. this choice is because of 1) lack of much pre-1970 data (in some locations anyway) and 2) because the choice is arbitrary.
• and much more broad and not methodology per se, but maybe most important nonetheless, mention the ways in which the statistics being reported do and do not measure climate change. this only needs to be a sentence or two to distinguish between time scales (climate change often measured in annual mean temperatures) and spatial scales (climate change is often measured in global mean temperatures). @jsbj knows a lot more about this than I do!

dunno if we want to do this before you distribute/advertise the bot? I don't know what kind of audience you have in mind but i feel like if someone sent me a link to this my first impulse would be to shout about how limited or even misleading it is for measuring climate change so it might be good to pre-empt that... but that's probably saying more about me than anything else.

[beamalsky]

Thanks for writing all this out! I agree that documentation is important. I'm inclined to say that where the bot and the website share logic, the methodology should go on the website. And when they don't, it should go in this repo's README. So we could add something about the gap testing and Sentry we just set up, and then the rest of this sounds like a great thing to me to link to on http://istheweatherweird.com/

I can add a bit to the README, but I think some writing from you two on statistics and climate change on the website would be really cool things to roll out soon after it's launched, not before. What do you think?