A publication that is paid for by readers and one that is paid for by are solving different problems.
The first must be worth the price to the person paying it.
The second must assemble an audience and hold it long enough to be sold.
Those objectives overlap substantially and at precisely the points that matter most.
An article that leaves a reader better informed and mildly bored serves the first model and not the second.
An article that leaves a reader and no better informed serves the second and not the first.
Neither model rewards the thing most editors say they value, which is being right about difficult questions.
Being right takes time, produces conclusions and generates no strong emotion in anybody.
The shift to has therefore been welcomed as a correction and is a partial one.
Subscribers are not a random sample of the public; they are people willing to pay for a particular view of the world.
A publication dependent on them acquires an to confirm that view rather than to test it.
is a sharper instrument than a declining click rate, and editors feel it faster.
The honest summary is that every funding model selects for something and none selects for accuracy directly.
Accuracy survives where it is protected by a convention that the funding does not reach.
Corrections policies, and the separation of comment from reporting are such conventions.
They are cheap, they are unpopular internally and they are the only part of the system anybody can actually design.
Readers judge a publication by whether it agrees with them, which is the one test it cannot pass honestly.
A publication that never irritates its own subscribers has stopped doing the thing they are paying for.