The Washington Post did some kind of study (free link, no paywall) to discover the sources behind "AI like ChatGPT" -- and they have a dialog you can use to find out how often certain sites are used in Google's C4 dataset (whatever that is). So of course I checked to see how much data from scripting.com is included. The answer: 710K tokens, which is 0.0005% of all tokens. Screen shot. Some comps: daringfireball.com had 610K, seths.blog had 750K and kottke.org had 590K.