I don't agree with dismissing this mouse just because it uses a laser sensor. What matters in a mouse depends on what the mouse is designed for. In fact, I would go so far as to say that for most purposes, the difference in tracking quality between a top of the line optical sensor and a non-dirt-cheap laser is unlikely to be important.
There are definitely some nice features to this mouse, like having a dedicated middle mouse button (which is more important for things like CAD and 3D graphics than it is for gaming) and gesture support. That said, for the price they're asking, I would like to see a lot more in terms of workflow automation capacity, as well as for more of the capacities to be moved to the device itself rather than the associated software, in order to improve compatibility with unsupported software and with locked-down workplace environments.
Maybe it's a little unfair of me, but I can't help but compare what this offers in terms of automation to what you would get out of a GWS Omnimouse, if you were crazy enough to buy one. The Omnimouse, for those of you who are not familiar with it, is a mouse designed nearly a decade ago using a motley assortment of fairly old components (even then) and assembled in a random guy's basement with some screws and conductive tape. The sensor was barely better than nothing at all, the software was glitchy and generally uncooperative, and durability was nonexistent. What it did offer was a lot of buttons, and a scripting capacity that puts to shame, as far as I know, any mouse developed before or since. Want customized gestures running on the mouse's own hardware for complete compatibility? You could do that. Layers and chording? Fine. Want some crazy combinations of conditions (hold a button and lift the mouse? wiggle it in a circle and then stop without pressing any button? move it left sometime between four and five seconds after having lifted it off?) to set off a script that then stays active and and changes its effects based on other things that you do? That's okay. Want to use input from your keyboard in your mouse scripts? That'll work too.
For mice that are designed for work rather than gaming, I can't help but assess them using the "rule of Omnimouse." The "rule of Omnimouse" states that if a work-oriented mouse designed now, or in recent years, costs more than $30, its automation capacities, and in particular the subset of its automation capacities that can operate off of the mouse's own hardware, must match or exceed those of the Omnimouse in power and flexibility. Based on the material on the 3DConnexion site, the CadMouse does not seem to satisfy this rule.