I like his point that Tablets are very different from Smartphones:

People tend to lump smartphones and tablets together as “mobile”. This can be misleading. Ask people who run internet companies and they’ll tell you that user behavior on tablets is far more similar to user behavior on desktops/laptops than it is to user behavior on smartphones. That said, the software on smartphones and tablets is similar, as are the discovery mechanisms (mostly app stores) and monetization techniques.

I like his point that App Stores terrify everyone:

App stores have had a few important effects: 1) They take 30% of revenue, which scares away most big companies (e.g. Microsoft) and also startups/venture capitalists. Not many businesses can survive an immediate 30% haircut. 2) They’ve led consumers to expect very low prices for software. It’s hard to imagine charging $30 let alone hundreds of dollars for software through app stores (although some mega-hit games do get near these levels with in-app purchases). This is why many big software vendors are scared. 3) The discovery mechanisms (e.g. top download charts) tend to have a rich-get-richer effect, making it very hard for software to grow from niches, as they often did in the past. Just as in the movie industry, the trend is toward creating blockbusters that appeal to everyone. The emergence of new app discovery mechanisms (e.g. FB & Twitter) might alleviate this problem.

I like his point that the biggest winners in mobile have "attach" business models:

The best entrepreneurs understand these dynamics and have been exploring “attach” business models, which basically means charging for something outside of the app store, like offline products/services (e.g. Square, Uber), online services (e.g. Spotify, Dropbox), and sometimes even hardware. Most of the companies that have succeeded (= generate real revenues/profits) on mobile were either desktop incumbents (e.g. eBay, Amazon, Facebook) or have attach business models.

The online service attach model that's most interesting that he didn't mention is Evernote.

Discoverability in app stores is universally terrible.  As an outside observer it seems to be just a combination of best seller lists and naive matches against keywords that are trivially spammed for SEO.

Netflix, Tivo and Amazon show that recommendation can be usefully done, but it's not there for apps yet.

I vaguely recall some startups eyeing to be the "recommendation engine" for App Stores .. haven't heard of anyone that's made it significantly yet