For 25 years, I've preached the superiority of the PC as an application platform, but times change and reconsideration is in order. The web browser and the infrastructure of the World Wide Web is on the cusp of bettering its aging cousin, the desktop-based graphical user interface for common PC applications.
The advent of Google's gmail service has been a signal occasion in the evolution of web apps. At this point, the main reason I am not trying to use gmail as my principal email client has less to do with its being a web app, and more with particular features and policies Google has chosen.
FYI, I would consider myself a demanding email user. I receive well over 100 non-spam messages a day, subscribe to multiple mailing lists, and use folders on a desktop client to manage my workflow. I keep many years of email, though I only rarely need to look at a message older than six months. My working set of mail is about 500 MB.
The great search, relatively simple filtering and labeling features, and auto-complete of addresses more than meet my needs. Surprisingly, the UI, while hardly enjoying the visual elegance of Apple's mail.app is just about up to the task. Technically, there is no reason valuable, missing capabilities like drag and drop can't be put in a browser. Google just hasn't, yet. Even without drag and drop, the UI is serviceable.
If gmail offered a way to synchronize gracefully with IMAP (including associations between labels and folders), and if it let me store more than 1 GB of mail as well as retain more ownership and control of my mail, I'd be happy. I need to keep a unified archive, and in six months I've already used 500 MB.
Then, how do I get my mail OUT of gmail if I want to? Roach motels where you check in but never check out don't work for me. These are basically policy issues, not technology issues, so there could, for instance, be an gmail alternative with more liberal policies that would meet my needs.
The greater convenience of the browser has been evident for many years. Browsers work from every PC, while desktop applications do not as they have to be installed (purchased, licensed, etc.) where they are to run. I can check my mail from anywhere. I like that.
The exception to the far greater convenience of the browser is off-line usage. With no net connection, data stored in a web app is inaccessible. So, infrastructure to support local storage of data (via caching, via something fancier) as a standard affordance of web-based applications is perhaps the biggest remaining barrier to be overcome. There is no fundamental reason I am aware of it can't be overcome either on a case-by-case basis, or better, in a more general way which would work not just for a given application, but for many of them.
So far, I've been describing redoing the feature set of a conventional app for the web. When an application, like Chandler, tries to break new ground in functionality or interface, matters grow considerably more complex, a subject I may take up here in the future. But for any new application project I get involved in starting, my strong predisposition is to think in terms of a web interface as primary.
Posted by mitch@osafoundation.org at January 02, 2005 04:11 PMIt's exciting to see us finally getting to the point where we can contemplate that all applications will be effectively supported by the web model. We were focused on this for years at Kenamea. It's great to see the industry adopting technologies and developing the skills that will (finally) make this kind of experience ubiquitous on the web. You're right, the final frontier is offline use... we had some of this working at Kenamea some years ago. Then for a while there it looked like Adam Bosworth was going to make it happen at BEA. I'm sure someone will step in and solve this...
More: http://tomconrad.blogspot.com/2005/01/mitch-kapor-on-web-applications.html
Posted by: tomconrad ![[TypeKey Profile Page]](http://blogs.osafoundation.org/mitch/nav-commenters.gif)
at January 2, 2005 05:46 PM
Gmail now supports POP3, so you can download your email to a desktop app if you want to get your mail out of it. I now read my gmail in mail.app most of the time.
However, you lose a lot of gmail's advantages that way. All of your labels are lost when they get downloaded. One nice thing - spam stays in gmail's spam folder and never gets downloaded.
Posted by: mike3k at January 2, 2005 07:03 PM
There was a really interesting IT Conversations audio show with one of the architects of GMail -- you can listen at http://www.itconversations.com/shows/detail302.html. He talked alot about the user data island issue, and it sounds like they want to make GMail open to movement both in and out. However, I worry that there's no way to do certain kinds of things well -- for example, sending signed email either requires a browser extension that GMail can use to do local digital signatures or trusting Google with your private key.
Posted by: unwiredben at January 2, 2005 09:38 PM
It's interesting that you say you won't switch to GMail unless you are able to switch away. I commented on this idea in a recent blog post talking about GMail's forwarding service: http://weblogs.mozillazine.org/gerv/archives/007058.html .
Posted by: Gerv at January 3, 2005 03:03 AM
Your summary of gmail echoes most of what I've read across the web. It seems that gmail is "good enough" for many moderate to heavy users and all the criticism I've encountered has been about google's policies rather than thier application.
So I have to ask, given Mitch's relationship with OSAF, what does this mean for Chandler?
Probably nothing, since the Chandler team seems to have a fairly solid goal and enough project management experience not to switch in the middle of a project. But I'll pretend they don't.
I've been secretly hoping for years that Chandler would magically turn into a remote XUL or DHTML application. I did see the wiki page in the jungle where XUL is discussed, hence the "magically" part.
As an explanation of why this is, I'll turn to university students. For context, I'm a recent grad (Dec 04) from gatech.edu. I know from reading the roadmap that the plan is to initially push Chandler at universities.
If Chandler were a gmail-grade web app, it would be in a class by itself. The best OSS competitor is IMP, which is under the GNU Horde project. The only other widespread webmail app I know about is squirrelmail. Both offerings have a slightly chunky UI and aren't as good as the second-tier commercial offerings (hotmail, yahoo). Despite this, well over 80% of the students I've met use the school-provided IMP system for reading their mail and everybody uses the webmail system as a repository for transferring files by mailing their working copies to themselves.
[rant]Companies providing webapps to universities produce TERRIBLE software from a UE perspective and charge a fortune for it. I've used the apps at three different .edu's and IMP is both the best webmail app and the most usable webapp of the whole lot. The next closest was a CS assignment turnin that was written by a group of students.[/rant]
I'm confident that a high quality OSS solution in a web-based PIM would take a huge chunk of the higher education market, not to mention elsewhere.
Posted by: grayrest at January 5, 2005 05:56 PM