How the New York Times Uses Machine Learning to Make Its Paywall Smarter

I’m pretty mixed about this type of optimization. From a pure business perspective it makes sense I suppose but somehow it also feels a bit dystopian as a user. But from an ML perspective it’s a very interesting read.

it does seems pretty easy to optimize oneself into a corner if it reduces outside the box thinking too

Also Anecdotally that optimization problem doesn’t get at the two reasons I don’t subscribe, and neither problem is suitable for ML. Applying algorithmic pricing would be a massive turn off and ML won’t fix what’s equivalent to repeated rat dropping sighting in an otherwise 5 star restaurant.

The article states that only first party data collected by NYT is used for the model. They also use no demographic data.

It seems that the only data that leaves is your previously read articles.

Amusingly, this implies that there are articles that actually increase your free article limit when accessed.

Obviously because previously you can read paywall news in your incognito browser. With all the machine learning talk, it amazes me how the NYTimes dev team haven't yet implemented dark mode in their apps.