The Beginning of the End of "Apps"

When Siri can replace much of an app's interface, and do so for the multiple apps a user may need to complete a complex task, it raises the question of whether we even need "apps" at all.

"App" is a funny word. An abbreviation of "application," it entered common parlance along with the present mobile revolution as a distinctive referrent for the sotftware programs that run on our devices. Apple was particularly responsible for popularizing it, featuring the term prominently in the marketing of its App Store and selling the iPhone on how many "apps" were available for it. It drew a through line from songs on the iTunes Music Store to apps on the App Store—distinct, individual, snack-sized pieces of functionality, available at modest price.1

With each release of iOS, Apple has expanded the facilities available to apps and the integrations with the operating system they are afforded, and these increasingly blur the lines and erase the boundaries separating apps from one another as well as from the OS itself. First came launch URLs, which clever developers figured out how to use to pass data between sandboxed apps. Then came share sheets, allowing apps to register themselves as consumers of various kinds of content and couriers on to social networks beyond the device. Then, extensions, allowing portions of apps to appear within system apps, or to be exposed to other third party apps via the share sheet mechanism—a tad crude, but still powerful.

Now come integrations into Maps, iMessage Apps and Siri SDK, and each of these further break up the notion of an app as a singular entity and disperse it's fragments throughout portions of the system. The most interesting and, I think, portentous of those is Siri, primarily because the virtual assistant can directly integrate with multiple apps and aggregate disparate information fragments around a single conversational context. Or, at least she can in the carefully curated demos Apple has shown so far.

Every application has its own unique interface. Often comprised of standard components, yes, but expressing a transactional or process flow that is fairly distinct to at least its application category. But when Siri can replace much of that interface, and do so for the multiple apps a user may need to complete a complex task, it raises the question of whether we even need apps at all.2

The reflexive response here might be dismissal: "Of course we need apps! Who is this guy?!" But if we think about the application categories that Apple demoed in particular—ridesharing with Uber and Lyft, restaurant reviews and reservations with Yelp and OpenTable, these aren't sources of deep content or environments for sustained productivity. They are in-and-out, satisfy an immediate need, transactional applications. If their information can be exposed to a common interface layer… why do we need custom apps?

Interface mediation changes our expectations, changes the requirements we have of our technology. Yes, at first it's just narrow commerce applications, but eventually it can be communications: having a digital assistant read your correspondences and present you with the salient elements, soliciting inputs as needed and responding on your behalf in all but the most intimate conversations is not at all far fetched—it's what persons with means or position have done for generations, and what better use of technology than affording us all the luxuries of the elite.

Take your Facebook feed—and I'm guesstimating here, because I haven't had a Facebook since 2010: how much of the stuff that's in your feed do you really have to personally read? If you had a personal assistant who you trusted, who would pore through your feed and select the truly important and relevant items for you, wouldn't you take advantage of that? (Arguably, Facebook itself does this, as we all know that your feed is not strictly chronological, nor is it a comprehensive stream of every single thing shared by the people you follow or are friends with. The problem is that just about none of us really trust Facebook's algorithms to match and respect our priorities.)

Yes, there are application categories that are resistant to this disintermediation, but they are likely far narrower than you may assume, and likely represent a tiny sliver of the average smartphone user's installed apps. The key to surrendering to mediated interaction is trust (again: see Facebook in the paragraph above), which makes Apple's extended riff on keeping information on-device and the mathematics of Differential Privacy potentially more telling.

I recognize this is a provocative argument, but I remind you that Steve Jobs didn't want (third-party) apps for the iPhone in the first place. Perhaps he'll get his wish in the end.


Once again, many thanks to Jeremy W. Sherman for proof reading this essay. Any flaws in reasoning or rhetoric are my reintroductions.

  1. This is rubbish, of course. Application software is limited only by the constraints of the platform it is built for and the imagination of its creators. There is nothing inherently small, or simple, or cute, or cheap about apps. This is equally true about music, by the way: the present digital economy for music is built around two- to five-minute tracks, not 20- to 60-minute.

  2. The instructive analogy here might be Google's impact on the web/search ads market, going from dedicated in-house sales teams to a proliferation of real-time auctioneering houses and ad networks that have both massively increased the inventory of ad units and driven per-impression and per-click costs down.