I'm not sure about this. Non-member read ratio doesn't affect story earnings, which makes sense, as the non-members put nothing into the pot, so their opinion is irrelevant.

But does it affect distribution in general? The algo could see 'overall low read ratio, including non-members' as a signal for 'poor story quality - do not distribute widely', even if with a higher weighting ascribed to the member RR.

Does anyone know if this happens? Is it explicitly stated anywhere? I kind of doubt it, given the veil of mystery that shrouds most Medium workings.

Imagine a story on Day 1 has 98 non-member views and 2 member views. Both members read >30s, so that's a 100% RR. Of the 98 non-members, only 7 actually read = 7.1% RR.

The algo needs to 'decide' what to do with that story on Day 2. What number does it use? 100%? 9%? A weighted combination that maybe works out at around 30% or 60%?

Because ultimately earnings also depend on views (since members can't read what they haven't seen) which depend on what the distribution algo does with each story.

I don't include these 'read for free' links, partly out of that suspicion/lack of information, and partly just because I feel it looks a bit tacky, like handing out free samples of cake cut into tiny squares on cocktail sticks in a supermarket aisle.