Minor reduction in sql here: WITH RECURSIVE dates(date) AS ( SELECT DISTINCT start FROM terms WHERE start >= "1975-01-01" ) , members_gen as ( SELECT members.*, members.birthday < '1928-01-01' as other, members.birthday between '1928-01-01' and '1945-12-31' as silent, members.birthday between '1946-01-01' and '1964-12-31' as boomer, members.birthday between '1965-01-01' and '1980-12-31' as x, members.birthday between '1981-01-01' and '1996-12-31' as millenial FROM members ) SELECT dates.date , sum(other) as num_other , sum(silent) as num_silent , sum(boomer) as num_boomer , sum(x) as num_x , sum(millenial) as num_millenial from terms inner join dates on terms.start <= dates.date and terms.end > dates.date left join members_gen on members_gen.bioguide = terms.member where terms.type = 'rep' group by dates.date

Yo thanks for the suggestion, just added this + republished the notebook!

Love, love, love. I just started playing with this, and installed Datasette in my Docker locally (because I would like to deploy the image to heroku later). The instructions in the Datasette docs worked great: https://docs.datasette.io/en/stable/installation.html#using-docker but in order to use localhost in my notebook, I had to of course insert --cors in the command line and also, because localhost didn't have an SSL cert, I couldn't have the https observable notebook fetch from the http localhost... so I had to use Chrome and allow the notebook sandbox to allow insecure content. I had to first find the domain for my user content by simply typing 'location.origin' in a cell to get the Observable worker's domain for my login, and then setting 'Insecure content: Allow' for that domain in Chrome. That allowed me to then finally do db = DatasetteClient("http://localhost:8001/fixtures") and the rest was great. Not sure if you want to perhaps explain some of these roadblocks to people who want to run their Datasette locally or in a local Docker. Thanks again for a great notebook!!!!