I feel like mining the text of Erowid/Bluelight/Longecity could be interesting... from a large enough corpus you could probably figure out some differences in p

I feel like mining the text of Erowid/Bluelight/Longecity could be interesting... from a large enough corpus you could probably figure out some differences in pharmacology between a few related compounds. And maybe get some adme/tox info as well. As for actual experiments, liability would be a nightmare... maybe someone in Sealand/North Korea/etc could administer a framework where they mail you samples+instructions and you send back a daily questionnaire or something