We don't A/B test core values

In the broadest sense it's almost always better to attract customers by simply building a better product rather than some form of trickery, like making subscriptions unreasonably hard to cancel. Sure you might leave a little bit of money on the table in the short term, but as the author mentions it's hard to quantify the long term cost of being a bit of an asshole.

The prevalence of these sorts of tactics demonstrates the extent to which we've MBA-ified the tech economy away from what people used to like about it.

Of course not.

If you A/B tested a core value, that implies you'd change your value system based on the outcome of the test, which suggests it wasn't such a core value to begin with.

The title sort of buries the lede. There's a solid point here about the abuse of easy-to-gather metrics being used to prop up tenuous arguments because the metrics that would refute that argument are hard-to-gather.