Looks great!

One of the 'options' that can be passed to Sortable is "hoverClass" which is described as "Class that gets applied to the otherwise white space." This sounds like it will allow us to hook an appearance style to the empty slot pretty easily.

Sorry I didn't use clear language. I don't think we want the image itself to continue to appear but something like the dotted rectangle that you were using to show where the image could be dropped in the recent mock-ups. Perhaps a light shade of grey for the outline. We just want to lighten the memory load on users so they don't have to remember where they picked it up from (and where it will go back to if they don't drop it).

Just wanted to clarify this in light of our recent conversations.

JQuery's default behaviour is to create "clone" of the dragged image (to use as an avatar) and set the visibility of the original element as "hidden".

The desired behavior is for the original element to remain visible (and stationary) as the drag action is occurring.

Do we want the original element to "fade" or stay the same in visibility intensity?

Check out igoogle's behavior when the portlets are rearranged. The empty rectangle is left there until the actual portlet is dropped. This is parallel to what I'm referring to. When the whole closes up before the image is dropped there is just more for the user to keep track of thier head. Also, if the images change locations as soon as the users "picks up" an image it is a little jarring and confusing. For instance, I've selected the image I want to move. I've clicked on it and at the same time am looking forward at where I want to move it to. As soon as I pick it up, that spot I was looking at is now in a new location since everything moved to close the hole where the image was. A third reason is that the original location may be of interest as they decide where to drop it.

I'm not completely sure I understand the last comment. Do we want to display a shadow of the image as it moves along like we do with drag and drop? It makes sense but I'm not sure how much value it adds given the move happens so quickly (the drag and drop happen incrementally an image jump at a time). If this is extra work to make happen my vote is to hold off until we get user testing data so we can better evaluate the value of the shadow.