Some sites won't let you click Accept until you've scrolled all the way down. But you could track how long they spent on the screen, etc. (Some may copy the terms and paste them into a full-screen view elsewhere to review them, I do this, so I don't even scroll it, but I do spend more than a few moments on the screen.)
That aggregate data could then show that the average user spends 2.3 seconds on the screen, just enough to find and click Accept, and has no idea what they're accepting, therefore yadda yadda.
I would love to argue that because the user has no meaningful negotiation power, it's a Hobson's Choice, they can't even submit feedback like "The reason I'm noping out is because you share X and Y data", there's no ability to edit the contract and send it back and see if the other side accepts your modified terms, etc, that it's not an equitable arrangement.
How would your method work on that?
My point being: statistics has developed arcane methods to describe a population. One should not blindly force the most common measure on every case.