Interesting read. I get where he’s coming from vis a vis training data for LLMs. But if those are the problem, negotiate a solution with those companies or block their crawlers. Don’t kill the apps making the site usable for everyone else.
No doubt, his comments are accurate as far as they go albeit completely out of context. I’d be much more interested in knowing how many of the top 100 subs (rather than top 5000) have reopened. I’d like to know what “top” even means here. I’m sure that 97% of mods don’t use 3rd party apps (according to Huffman) because they mod subs of a few dozen to a few hundred members or their subs are almost completely inactive.
In other words, this is interesting damage control, but it needs a lot more context. And NPR’s quality control and fact-checking are sadly lacking.
Is there a new dataisbeautiful that celebrates actual good visualizations, or are we just going to go back to jerking each other off to poorly labeled excel charts with trendlines going every which way?