Re: ACTION-47: Draft a chapter outline of "What is a Web App?"

This is a great effort, if only to start a debate that needs to be
had, because our view of a web app is very different to this.  I'll
counter each of these points first and then give our definition:

On 10 August 2012 05:33, LEONG, JENNIFER <jl3101@att.com> wrote:
> Here is a stab at ACTION-47 (http://www.w3.org/community/coremob/track/actions/47), drafted by Bryan Sullivan and I. Please let me know if this helps to define "What is a Web App" at least within the context of the CoreMob requirements scope.
> -----------------------------------------
> What is a mobile Web app?
>
> First, a mobile Web app is mobile.  That means that it is meant to be used on a mobile device, such as a smartphone or tablet.

The question was actually 'what is a web app', and 'mobile' has snuck
in there.  I'm not sure that's helpful, as web apps don't have to be
mobile (see Chrome Web Store), and mobile doesn't have to mean
smartphones and tablets (see netbooks with built in 3G).  We don't
think the device you choose to use it on should affect whether we
consider it to be an app or not.

> Further, it is likely to be used in and between various network environments, and usable
> when the user's device is network-connected (online), or offline. It may be expressly intended
> for offline use only, requiring no network connection after being loaded or installed.

It's reasonably safe to say that a static website that adds a manifest
so you can reliably view it when offline isn't an app.  So it feels
like we're saying here that there is no definitive definition of a web
app, but rather various things (like this offline point) add up until
we reach some arbitrary 'bar' at which point we tip the scales into
'app' territory.  In that case, it would surely be virtually
impossible to reach a tangible definition.

>
> Next, while being based on Web technologies (e.g. HTML, CSS, and JavaScript), mobile
> Web apps can be implemented using WebView APIs in a native code wrapper, as installable
> Web apps using various standards-based and proprietary app packaging, or as Web browser-
> based apps.

I agree, but given that a) you could use a webview to load the worst
website in the world, and b) we're not insisting on the webview
wrapper as a pre-requisite for 'app-ness', this seems a moot point.

>
> Mobile Web apps  may be loaded on-demand from a Web server, may use
> installed or persistently cached Web content, or may use a combination of
> both installed/cached and on-demand Web content.

This is de-facto true, since it allows for every possible option and
doesn't mandate any or prohibit any, so it doesn't seem to narrow the
definition.

>
> Mobile Web apps are typically single/special purpose rather than general purpose apps.
> They also focus on simple presentation and ease of user interaction, leveraging diverse
> forms of user input including touch, on-screen or dedicated keyboards, accelerometer,
> and speech.

I'm uncertain about what a 'general purpose' app would be, but
otherwise agree that simple presentation and ease of user interaction
is important to apps.  However, it could be seen as important to
websites as well. I think we can drill down on this point a but more
(see below).

>
> Mobile Web apps are designed to work well in diverse environments and device form
> factors or orientations, and often use device information (e.g. geolocation) to provide
> a more contextually relevant user experience.

I've seen a number of web apps designed purely for iOS, which don't
really work on other mobile platforms and not at all on desktop.  But
they're still apps in the sense that they feel the same as a native
iOS app.

Here is my take on what an app is.  First, I would say we need to
ignore distractions involving technology choices - the very point of
the web is that it is ubiquitous and ever evolving, so tying the
definition to any concrete form of current technology is likely to be
a bad idea.  The closest quantitative definition that I'd be
reasonably happy with is that apps lack a page metaphor when moving
between states.  A traditional website will, at every state change,
load a new page.  An app does not (appear to) have the page concept at
all.

However, there are counter arguments to that as well, and I actually
prefer a qualitative definition - an app is something that is designed
for the user, which feels 'made for me'.  The problem with the web
when the iPhone came along was that the sites didn't fit well on the
screen and using them was fiddly, because they weren't made for the
iPhone.  Apple released the iOS SDK so that third parties could build
apps that were made for the iPhone, and as a result the user
experience of those apps feels perfectly tuned for that device.  Any
website that does the same, even if only for iOS, has become an app,
in my view.  The great potential of using web technologies is that you
can potentially make a single app which feels 'made for me' on several
different platforms, by having it adapt automatically to the
constraints and capabilities of the platform on which it is used.

I would include desktop websites here as well.  It's become the norm
to set a fixed width for a website, and make the design job easier as
a result. But doing so means the site is no longer 'made for me' on
any screen that doesn't match the fixed width that the developer
chose.  For example, 'apps' from the Chrome Web Store tend not to make
use of fixed width layouts nearly as much as the wider web does (for a
specific example compare www.bbcgoodfood.com with
www.bbcgoodfood.com/chromeapp).

In conclusion, I realise I've countered all of Jennifer's points and I
really don't mean to devalue her draft, as the points made are all
valid in a certain context, and it may well be that my view is the
outlier because we see apps in a different context.  A few months ago
I posted an article on the FT Labs blog trying to answer this very
question, and for those interested, it makes roughly the same argument
that I've presented here.

http://labs.ft.com/2012/06/what-exactly-is-an-app/

Cheers,

Andrew

Andrew Betts [skype:triblondon | @triblondon]
Director, FT Labs [labs.ft.com | 0870 085 2038 | @ftlabs]

-- 

------------------------------
This email was sent by a company owned by Pearson plc, registered office at 
80 Strand, London WC2R 0RL.  Registered in England and Wales with company 
number 53723.

Received on Monday, 13 August 2012 12:33:35 UTC