I think there is a big question mark over the 'Show me more like this' button in the app. What does that actually mean? To what extent is the algo capable of qualitatively identifying 'more like this', and hence pushing story X by writer Y to those who are likely to want to read it, either because they have given that explicit signal by pressing the button, or simply because they registered a high engagement score for the similar story Z by writer Y (or writer W) recently?

I think the answer is 'not at all, in fact'. And that 'more like this' simply means 'more stuff with a similar combination of tags', e.g. 'fiction' and 'scifi'. But does that mean Michael Moorcock? Ursula Le Guin? Arthur C. Clarke? JG Ballard? And can it distinguish short stories from '10 Great Sci-Fi Movies you MUST See'?

I suspect that the lack of the sophistication in the algo means that it distributes most stuff poorly, with a fairly random scattergun approach, which leads to most of it starting off with a very low read ratio - unless

enough followers/subscribers pick up on it straight out of the gate - and it then reaches the self-fulfilling conclusion that 'no one likes this - spike it'.

But how to enhance the algo? That would require AI being far more sophisticated in its capacity to interpret style and nuance, and maybe we don't actually want those 'powers' to exist.

Right now, I think that anything under 'For you' that isn't explained by 'in your network', 'user you follow highlighted' and those specific factors is simply: 'boosted, and with a tag or two of relevance to you based on explicit preferences and reading history'.

Under the current Medium model, the boost is the key tool to decent distribution, regardless of genre, except for stories that get lucky with a high read ratio that then snowballs from the get-go.

That's my perception, and I think that's what we have to live with.