This is effectively the condenser, but for code. All the code is identified by its hash, and is immutable, with names aliased to hashes. Also note the power this gives us. By throwing away some things that we're used to (that code is mutable, that code has names), we're afforded a bunch of useful properties (code is easily distributed, test caching, etc). We would call these properties "free", as in, we got them "for free" by chucking away some stuff.
A core property of things that are "for free" is that we get them when we generalise - by being less specific, we get more ⚡
I've been trying to reduce the amount of data I have... around, and trying to get all the shit jumbling around my head onto paper. Examples:
which in all honestly, I think could be automated
we have commercial big data tools, designed to crunch huge amounts of metadata, able to run ML cohorting and clustering algos, able to deal with heterogeneous unstructured data
and we have just... shit everywhere
things saved on Facebook, in your camera roll, in Dropbox folders 5 layers deep, saved Reddit links, random fucking .txt files, Google Notes, Apple Notes, shit you've sent yourself in Gmail, shit you've sent yourself on Slack, things you've starred on Slack, things you've liked on Twitter, OneNote, Evernote (edited)