On his weblog, Ray Ozzie says:
In terms of building software that people actually use, I strictly prioritize my platform investments and always have based upon "where the users are". No religion. Obviously that means Windows first. But we didn't know what to make of the Linux phenomenon when we were building Groove, so we covered our bases by funding a company (Macadamian) to enhance Wine so that Groove would run. We eventually gave up: nobody gave a hoot about Linux on the desktop. Regarding the Mac, two factoids - take them for what you will: a) the top personal request on the Groove website is currently "when will there be a Mac version?", and b) no major enterprise customer has yet asked to purchase a Mac version. Quite perplexed.
As CIO for a large organization (22,000 users) I've puzzled over what to do with OS choice (particularly for Mac users) for some time. I think I have at least part of the answer.
One of the problems with managing desktops in small clusters is that the workload is overwhelming. Suppose that you manage 22,000 desktops in clusters of under 100. That is, for every 100 users or so, there is a separate administrator who is responsible for buying the PCs, installing them, maintaining them, and eventually disposing of them. This one person needs to know a lot about a lot of things. They become a PC administration generalist of sorts. In this scenario, having a few users who want to use Macs is a nightmare because the administrator already is overworked and needs to know too much about managing Windows without having to learn Macs.
Now, imagine another scenario where PCs are managed by a team of people where all 22,000 PCs are bought, installed, maintained, and disposed of by one group (not necessarily centralized in a geographic sense, just unified and working as a team). One might think that this system would be more rigid and less willing to support alternate operating systems and software, but that need not be the case. In this scenario, there are people who are experts at purchasing and installing PCs, experts at mail, experts at LANs, experts at OS bugs, experts at specific software problems, etc. The organization has the same number of people, but now with the ability to specialize to a much greater degree. In addition, the organization as a whole can purchase and amortize enterprise-class PC maintenance software (like that from Altiris, a Utah company) to rationalize and automate the maintenance of PCs across the organization.
Such an organization could easily lay on a few people, at little additional relative cost, to support Macs and a few Mac users here and there wouldn't cause anyone undue stress. There would be no need for an entire division to have to make the jump to Mac or something else all at once. Indeed, as long as the software is interoperable (which mostly means that it supports common data formats) wider latitude in software selection could be allowed as well. Modern LAN technology and PC maintenance software make this scenario much more plausible today which is why more and more organizations are going this route. In addition to saving a lot of money, you also get increased levels of service and greater flexibility. Who could argue with that?