Wednesday, August 12, 2009

What is a smartphone?

I’m reviewing a paper and the question of defining “smartphone” has come up.

As it so happens, the question is particularly germane today when Dataquest released Q2 2009 global market share data. For all phones, the top 5 are Nokia, Samsung, LG, Motorola and Sony Ericsson, while for “smartphones” the ranking is:

  1. Nokia: 45%
  2. RIM: 18.7%
  3. Apple: 13.3%
  4. HTC: 6%
  5. Fujitsu: 3%
  6. Other: 13.9%
It seems as though recently, the term “smartphone” has degenerated into a meaningless marketing phrase rather than an accurate description of a product category. Even the analyst companies (such as IDC, Canalys or Dataquest) don’t bother to define it — at least in a way that others know what the definition is.

Here is what a simple Google search shows:
Definitions of smartphone on the Web
This suggests at least three possible ways to define “smartphone”:
  1. Give up on the term. Features that were previously “smart” (such as web surfing) are now standard in most phones and thus the distinction is meaningless.
  2. Mobile platform. A phone that is programmable by users or third party using native APIs. (Contrast to the Phone Scoop definition of a “feature phone”).
  3. Smarter than average. A phone that has more “smart” features than the norm or typical phone
I think the case for #1 is probably the most tenable long-term solution, but since I use the term “smartphone” a lot in this blog, obviously I’m not ready to give up on it just yet.

The #2 definition was one being used to call the iPhone a non-smartphone back in 2007 — perhaps an answer without much face validity, but a moot point now that Apple has released its SDK and opened its iPhone App Store. Even without the iPhone 1.0, the definition also potentially has a problem with the BlackBerry — where the native APIs are Java-based APIs.

The #3 definition would continue indefinitely, but would raise the question of where to draw the line — and how to change that line over time. If a web browser was necessary last year, what is necessary this year or next? An AJAX web browser? A web browser with Flash?

A possible fourth definition is a phone running a recognized smartphone OS, e.g. Symbian, BlackBerry, iPhone, Android, Windows Mobile or webOS. Of course, that raises the question of “what is a smartphone OS”, i.e. the infamous Potter Stewart problem that applies to other areas like enterprise software.

Every possible definition raises other questions. Here are a few:
1. If the native applications are web-based applications, does that mean that it’s no a smartphone?
2. Suppose a phone has a valid smartphone OS, but it’s a locked down to disable the third-party (or user) supply of applications. (E.g. DoCoMo’s MOAP(S) or MOAP(L) phones, many Linux-based feature phones). Is it a smartphone?

3. (Extra credit) Which of the following is an oxymoron?
  1. Razr smartphone
  2. LG smartphone
  3. Palm smartphone


GMM said...

Yeah, just stop using the term. The main thing that I see separating pc-like devices from phone-like devices is the ability to easily execute 3rd party code. The word "smart" doesn't really have anything todo with that.

Joel West said...

Except of course we need a way to report "smartphone market share." It's not like people buying a $0 bundled throwaway phone are the same market segment as those paying $200+ out of pocket for a iPhone or Nokia. (Of course the $50 BlackBerries blur these lines).

GMM said...

Well, you can talk about programmable phones, or phones that are above the average cost by a certain amount, depending on what you care about. Those companies that track these things must have some definition that they use.

Joel West said...

That's exactly my point: we need a single definition.

The PC industry doesn't vary the definition of desktop vs. laptop on a firm-by-firm basis.

The consumer electronics industry knows how to distinguish flat panel TVs from CRT ones, and within flat panels, separate the LCD from OLED from plasma and DLP.

Let's have some sort of consistent definition, and make sure the reporting statistics declare what definition they use.