i m just backseat driving here, but the fact that paid apps sell less on android is damning evidence for devs to not migrate across (or target it as the primary platform).
A subscription based model is more demanding to maintain, vs a pay-once-app. You at least don't need a server to maintain subscriptions. Ditto with In app purchases (which also require you to maintain a server to keep track of who has paid for what).
For selling "just apps", subscription can definitely be not worth it. But a smartphone-app-centered business model is not the only thing that results in smartphone apps being made. My issue is with the general wording in the grandparent post, which seems to me to suggest Android isn't viable for all or most of developers overall.
If you're looking to reach users of your non-app-centric service that you're looking to provide a free app for, targetting Android is a no-brainer if your service is aimed at demographic groups in which Android has sizeable or even plurality user adoption. That's most of them, but not all.
Looking through apps on my devices (I'm one of the "cheap" Android users with zero paid apps), Flickr, Evernote, Mint.com, Opera, car2go, Skype, Starbucks, Kobo all stand out as quality apps that pay off for the developers in other ways.