From our networking perspective, the short answer is "no" -- wireless is not a viable *alternative* to wired networking for professional or knowledge-worker offices, much less for departmental computing facilities. Wireless support is inevitable; it's addictive technology that won't go away. People will insist that we support it, and in certain venues such as classrooms and auditoria, it will be the preferred/only medium... But I don't believe that it can serve as a *replacement* for wired networks except in very specific circumstances.
Last, but not least, there is the question of wireless infrastructure cost. The motivation to go "wireless only" is clearly based on the (tempting, but suspect) assumption that it costs less to deploy wireless than wired, combined with the (correct) assumption that some amount of wireless support will be needed anyway. However, wireless deployment costs are a bit deceptive: it actually requires a surprising amount of wired infrastructure to build upon, in addition to the wireless-specific costs associated with providing a scalable, manageable, and utility-quality service offering. So savings are more illusive and elusive than is commonly thought. There is this notion that you just buy a $300 Apple "base station", plug it in somewhere, and you're done.
Unfortunately, there is much more to provisioning a high-availability network utility than that... You need to start with site surveys and spectrum analysis, then engineer locations for antennas and base stations according to radiation patterns, interference sources, number of expected simultaneous users and desired performance levels. Once this is done, physical plant issues arise for placing antennas, getting cables to the access points, and connecting those access points to conventional Ethernet infrastructure, which needs to be specially designed to keep all of the access points in a building on the same subnet. (This is necessary to allow wireless "roaming" within the building, and also to facilitate connection to the special access control servers needed for wireless LANs.) Of course, network management and monitoring systems (for both wired and wireless infrastructure) are a necessary evil for any large-scale production service.
Note also that when considering wireless vs. wired in terms of cost displacement, a lot of the wired network infrastructure is shared with telephone infrastructure... so unless wired phones are also eliminated from a building plan, the communication rooms, riser systems, conduits, tray, and some amount of wireplant will still be needed. Then again, much of that infrastructure will be needed to support wireless access points, too.
Wireless can be less expensive, and certainly more convenient, in a number of specific situations, such as classrooms and auditoria. However, even in those venues, it will be important to provide a stable high-bandwidth (wired) network connection for speakers to use, in order to enable advanced demonstrations. Similarly, planning wired network connections for videoconferencing equipment is prudent, and should be treated as a separate problem from providing connectivity for the audience or class.
It is true that buildings exist where retrofitting wired network infrastructure fully is horrifically expensive (think asbestos), and deployment of some wireless capabilities to provide connectivity in certain offices, in spite of all their limitations, may make sense --but this is certainly not the case for offices in new buildings or major remodels.
So the "bad news" is that we really need a hybrid strategy; we need to do both, and that is bad only because the costs are non-trivial. But I do not see any viable alternative if we are to meet the conflicting requirements of mobile freedom and also support for high-performance advanced applications.
The counter-argument to all of the above is that "We are only going to use laptops, and those will need wireless connectivity anyway, and we promise not to have any servers, so why should we spend money on wired infrastructure if it isn't going to be used?"
One answer is that the limitations of wireless may begin to cramp one's style as applications and departmental needs evolve. Moreover, in the "University of a thousand years" no one can claim that every successor occupant will be content with the same assumptions and limitations, and if wired infrastructure isn't planned-for initially, the cost to retrofit will be prohibitive. Also, the wireless stuff will need baseline wired network infrastructure anyway.
In any case, to abandon standard connectivity infrastructure in UW buildings (or at least, in offices) on the grounds that "what we can do with a wireless laptop is good enough, and all we'll ever need" is surely a CIO/VP-level decision. It would be analogous to deciding that "point of use" solar power technology had matured to the point where we could radically scale back on conventional power distribution infrastructure. Would that it were true, but it isn't.
In my own home network, I have the option of using either wired or wireless technology. I chose a hybrid approach, with some machines connected by conventional wired Ethernet, and others via wireless. I believe the same hybrid strategy is appropriate for UW.
It is important that wireless infrastructure on campus be considered part of the over-all network utility infrastructure. If departments roll out their own wireless infrastructure, their users will suffer because they will need to reconfigure their laptops to use the wireless facilities elsewhere on campus. This is because departmental installations need to be configured differently than the campus-wide wireless infrastructure in order to avoid interference. There are also some access control issues that need to be managed, as well as normal network management and security operations that I suspect most departments would be happy to have someone else worry about. So in general we do discourage departmental "do it yourself" deployment of network infrastructure, wired or wireless.
Wireless networking is a wonderful thing, and should be encouraged (although with eyes open about its limitations, volatility, and costs); however, it is not an adequate replacement for all wired networking infrastructure. In particular, for a facility to be considered "state of the art" it must not only support wireless LAN connectivity in conference rooms, classrooms, auditoria, lounges, etc, but it must provide stable, uncongested high-bandwidth connectivity (read: wired connections) for knowledge workers and lecturers who may wish to use demanding advanced applications that cannot be supported via 802.11b technology. Of course servers and high-end conferencing systems are examples of other hosts that need wired connections.