In-app-purchasing in games relies on conditioning the user across many hours of gameplay. 98% of users never take the bait. Roughly 0.15% of users make up 50% of the revenue.

In short: I don't think the "skinner box" model is relevant to web content. I can't set up a pain/pleasure response in readers and condition a tiny fraction of them to pay vast sums after 10 to 50 hours of engagement. That's not an option in journalism.

(Not that this has stopped the FT from trying it: $1 for a one-month trial, $480 a year for a subscription. I wonder how good the conversion rates are?)