Here’s a blog post on O’Reilly suggesting that sticking to web conventions for web apps (e.g., gmail) is better than trying to emulate the desktop UI for web apps (e.g., Zimbra). Interesting. I don’t think I agree. I think users want a full desktop UI in their web apps (drag and drop, etc.), and have just been willing to put up with the limited web UI because of the benefits of anywhere, anytime access. I especially think this is true in daily use business applications like MLS systems or contact management systems (e.g., Salesforce, etc.). What do you think, should web apps be limited to the web UI or shoot for more of a desktop UI?
Tangentially, here’s a post about not letting marketing derail the software development process with too many incremental “wouldn’t it be great if(s)…” As we march toward our next major release this fall, I best remember that. I’m hopeful our clients and developers will help me stay focused on what’s important as I want more and more and more!
Web UI or Desktop UI? It’s all about capabilities and (agreeing with Matt L) expectations. If users expect the visual language of the web, primarily defined when browsers were quite limited in their capabilities, and you give them a desktop UI in their web browser then it may be confusing. But as browsers continue evolve, web UI can evolve to include more capabilities and different ways of doing things. Let’s not get stuck doing things one, less than ideal, way just because the old Mosaic browser made that the only practical way to do it back in the day. I think over time the distinction between web and desktop UI will continue to break down and that the question of whether one has to have one or the other will just naturally become less meaningful. But it’s not as if “Desktop UI” is a static thing or an ideal to shoot for – it can evolve too (e.g. multi-touch interaction video – http://www.youtube.com/watch?v=PLhMVNdplJc)
Jenifer Tidwell has an excellent book on these concepts (that I’ve coincidentally had waiting for months and just got to crack last week), titled “Designing Interfaces“.
She posits that interfaces that are labeled “intuitive” are actually familiar, and the most important facet to making an interface usable is establishing familiarity in context.
-Matt
10:38 pm
I’m mostly with you. There’s a reason people prefer Outlook Web Access to every other corporate mailserver web interface — because it works (almost) just like Outlook. Virtually all of the native UI conventions are intact, including good ol’ drag’n'drop.
Perhaps the point that they’ve overlooked is the users’ expectation in context. I think Zimba and OWA are better because they meet my expectations of what a _mail_client_ should do, not just how I expect a “webmail” client to behave. The original author seems to be stuck on the fact that he expects a webmail client to be a disjointed experience from a “real” application.
If the ideal circulating through Microsoft and Google is that web apps are the new thin client and The Way Are Going, then the new conventions will be learned and become the new expectation.
Microsoft’s Silverlight platform stands to spawn a whole new generation of this stuff beyond Flash + AJAX… if it takes off.
Regarding the don’t-let-marketing-define-the-app, I highly recommend Steve Krug’s book Don’t Make Me Think. I’ve built web apps professionally for, shoo, a decade now, and the book covers at least 90% of the pitfalls I’ve seen suck the life out of good sites. He has a whole chapter on mitigating management’s decisions.
-Matt