John Carroll's next tweet after that (6 minutes later).....

"I believe the average failure rate for anything coming out of preclinical research is 90%, some say higher. And that's for legit co.s"

https://twitter.com/JohnCendpts/status/777309208651767808