Author here, yes this is a big problem (the biggest?), as the incentives are all wrong. As I noted in the post: "Often ... apps will have features to integrate with other apps and the wider operating system - but not so much that they become invisible. Instagram still wants you to see its logo, consume its specific content and stay within its ecosystem. Once again, the implementation and architecture are driven by economic imperatives."
Ah, yes. But you seem to be saying in the article that these incentives cause the platforms to lack such features, whereas what I mean is that e.g. Android's intents and activities provide just such a mechanism (which could be used to great effect as indicated in the sibling comment about OpenIntents), but commercial application vendors don't want to use it. They would rather control the user experience than integrate seamlessly into the platform.
In particular, you mention sandboxes imprisoning the code. But Android allows an app in one sandbox to display an activity (essentially a dialog box) from another app running in another sandbox with different privileges in a way that appears seamless to the user. I could have one app with access to bluetooth (but no camera) call upon another app with access to the camera (but no bluetooth) in order to take a photo.
I believe apps can also expose services and data sources (ContentProviders) – e.g. your Images, Tables and Conversations – to other apps and define their own permissions[1] for them.
Obviously economic incentives have an impact on what problems get worked on, but economic incentives can’t reduce the complexity of problems. I think what we have here is a problem of irreducible complexity. The reason that abstractions like the “Image” or “Table” you described one your article are few and far between is because it’s really hard to implement these objects in a way that scales to fit a broad enough set of use cases. The design of such objects involves making a set of tradeoffs where tradeoff space has a very high dimensionality and lots of shallow local optima.
Not that I disagree with your overall assertion, but without economic incentive there is really no incentive in most cases. For someone who does have economic incentive, it's really business risk reduction 101 to not pin your application to a network of invisible and hidden dependencies that you have no control over.
Right, using f-droid and installing open source apps where the only incentive is the user's benefits.. that's a breeze of fresh air. Of course it can never give you access to instragram or any particular commercial platform like that, but you can handle your own data with dignity.
This has resonated with me as I've recently been thinking about service interopt. There is a strong push in my org to go full rpc, which will probably win out for not bad reasons, but I sort of wish we had but in the necessary up front investment to model it restfully.
