now you might say, hey, its not my job to make more money for the site, just to make it better. true. but the economics ultimately comes back down into your schedule sooner or later.

this is not to say an interface cannot induce higher margins, but i would say look at something like linden labs/WoW for an interface that moves way beyond the commodity market, literally breeding addicts. the next step for internet UIs (and high margins) is low-latency and very graphical. note i said internet, not web. the web stack has hit the wall and is already a commodity market with flatish numbers (the only growth is in internationals). in three years, the traffic from graphical environments will be greater than all http traffic. its happening.

want to make the internet interface for the next ten years? in my opinion it is time to dust off the opengl guide.

2. Point well-taken. People do not expect D&D in the web browser because this is not an expected behavior on the web. The only way people know something is draggable on the desktop is by way of being told by MS or their friend. Well, the logic follows…

One of the best things to come out of the past few years in the web (still a really exciting time, no matter which road we all take) was a re-recogniton of the user (whether decision makers knew it or not)

Id hate to think we’d simply forgo any of these gains, giving them interactive decisions theyre either not asking for, or as you said, not expecting.

When theyre asking for them, ofcourse thats a different matter.

Like youve said, done well, they wont have to even ask, they’ll simply adopt the well developed, well designed (and well thought out) experiences presented to them.

But at the same time, Drag and Drop is often the *best* “alternative” way for the user to achieve her goals of collection or arrangement, given limitations intrinsic to primitive gestures available via mouse and keyboard.

What I worry about is that the user’s schema for rich desktop interaction, where complete, won’t make it to the web in complete form. We’ll provide drag and drop that mirrors the desktop, but not a selection model that does the same; as a result, the user, upon learning she can drag objects around, will have a moment when she is willing to suspend disbelief and imagine that multiselection with control and shift key transformations is possible. She won’t have to be taught or bombarded with discoverability at that moment; it will just be part of the schema she transfers over from the desktop. And if it’s not there, that little moment of disappointment will be followed by months during which we will have to convince her yet again to try something new once we have figured out a way to make the interaction better.

Gradual is good in some sense, yes, but moving the whole schema, the whole expectation set, over in one undamaged transplantation — where possible — has some compelling advantages for higher-end users.