Comment by michaelt

Businesses that scrape websites for a living hire people in third-world countries to solve captchas 24/7 to keep the scraping bots running.

So when I successfully solve a captcha, that doesn't make me 100% trusted not-a-scraping-bot. Instead it's an input into a statistical model, along with all the other identifying information they can hoover up, and that statistical model may still say no.