tech decentral

I agree with Alex Barnett on the prospects for “occasionally connected computing” and rich internet apps in 2007:

As I understand it, the RIA meme has the Occasionally Connected Computing (OCC) theme running through it - think of this as hybrid of desktop / web apps. I think the buzz will certainly be there for RIA in 2007 but it will take time (years) for development tools that enable these scenarios to become even close to mainstream for the developer communities. Lots of experimentation, no doubt, but the number of implementations will be few and far between compared to the progress in the delivery of rich experiences delivered purely via the web. Ajax for 2007 - my bet here is that pure play brower-based app development will be the winner of 2007, 2008 and 2009 :-)

I believe that browser-based apps are the way forward. They remove so much friction from computing. And even more important, they put the emphasis on information and data and navigability rather than on the application itself. Google Reader is a good example: the interface packs tons of information about unread items into your browser window. Then it gives you keyboard access to move through those items quickly. At any time, you can move externally to seemingly infinite sources of information via the links in the articles.

But it’s obvious that there are situations when a desktop application is the only way to go. Right now, if you are a graphic designer, you’re not going to use picnik to edit your photos. You’ll use Adobe PhotoShop. If you are writing long, heavily-formatted documents, you’ll use Microsoft Office or something similar, not Google Docs & Spreadsheets or Zoho Writer. If you’re a software developer, you’re using a desktop IDE like Eclipse or NetBeans, not entering code into a browser-based code editor.

How do we integrate the desktop and web app experience? Is there even any need to? Right now, there seems to be. Medium and long term–I’m not sure. Will it all move onto the web? It’s hard to imagine Google reproducing all of Word’s feature bloat in Docs & Spreadsheet. The flip side of feature bloat is that one feature you really need. Microsoft didn’t stick in all that stuff just for fun.

I’m going to assume there’s some need to integrate the desktop and web experience… though I’m not convinced that’s the case. If it is, what’s the right way to go about it? Do we extend web apps so they can work in disconnected mode (that’s what Zimbra’s doing)? Or do we move web apps and web development paradigms onto the desktop (that’s what Apollo promises)?

How about something different? How about integration at the data level rather than by making applications live both in the browser and on the desktop? An idea along these lines was posed to me by Simon Phipps at the Sun Analyst Summit… what if OpenOffice and ODF could serve as an integrating link across web and desktop apps?

There are lots of things I like about this idea:

• What we need to share across the web and our desktops is not specific application functionality and features but rather data–emails, documents, calendar items, etc. I don’t expect web apps and desktop apps to act the same, but it’d be great if they shared data better.
• The web paradigm works well because it focuses on information and relationships, not on application details. Del.icio.us has very few features but is rich in networked information. This is another way of saying what I just said: it’s data we need to share not applications.
• At this point, most people can’t do without an office suite. The online office suites just don’t offer the performance or features to replace MS Office… but OpenOffice does. Since the one desktop application everyone is likely to have is an office suite, why not make it the center of disconnected web access?
• OpenOffice is a package whose time has come… people have been trained now to expect free, downloadable desktop and web applications. Maybe medium-sized and big corporations can afford to spend big bucks for MS Office for their employees but if an acceptable free alternative exists, why will people stand for that kind of gouging? I’m all for the free market but prices are sticky and people don’t have time to become totally informed. Microsoft has taken advantage of this for far too long, and it no longer fits with the way things work.

I haven’t dug into the technical details of how this might work; Simon mentioned UNO (Universal Network Objects, or the component model for OpenOffice), the ODF Toolkit, and XForms. The fact that this might involve some kind of pluggable component model evokes the olden days of COM and OLE, when Microsoft proposed that maybe a person’s interaction with the computer should be data-centric rather than application-centric. That’s a fine idea–probably the right idea–but because software reflects the organizational structure that produces it (just as a corporate website often reflects organizational structure and power), Microsoft wasn’t in any position to deliver on that idea. As a massive top-down silo’ed organization, all they could deliver was massive top-down silo’ed software.

What’s needed to deliver on the idea of data-centric web/desktop integration is a decentralized development model… an open-source development model, for example. Suddenly, Sun is looking quite prescient in their acquisition of StarOffice and then open sourcing of the code into OpenOffice.

Update 2/9/2007: My WWD colleague Mike Gunderloy reminded me that OLE 2.0 already happened. That was the distributed objects version of COM/OLE. Perhaps I should have called this OLE 2.0 2.0. Or not brought up OLE at all–except that the underlying motivation seems so close to me.

Mike also pointed me at Ray Ozzie’s live clipboard idea, an RSS-based clipboard spanning web and desktop applications as a basis for wiring together websites and web apps and desktop apps–i.e. as a mashup enabling tool.

Disclosure: Sun is a client. Adobe may be soon.