Third, you didn't use Linux if you think stability and speed won't increase. A user spending five minutes a day less on opening/saving/closing documents, applications, tables, etc. (be it Word or StarOffice/OpenOffice) means quite some bucks a year more for the company where he or she works. Add several additional minutes a day spent for reboots after crashes or other sorts of problems, and I think it rounds up to a pretty significant amount of money. Of course, there must be an initial investment in training the users so they learn to efficiently use the new platform, but this is a short-term investment, while switching to Linux is a strategic decision.

Fourth, development of Web-enabled Notes applications is not necessarily slower and more effort-consuming than developing for the client alone, as long as you take the same approach in your development as Notes does in the design you describe they used for their client: isolating all UI (User Interface)-related issues in a separate layer and developing all other parts to just use a UI abstraction. Development of a reusable isolation layer is not simple and not the kind of project non-software companies do internally, but this can be bought from specialized companies for quite a small amount of money, compared to other costs involved in a switch to Linux.

Judging by a few not so big companies that switched to Linux and which I know switched to Linux (some of which I know personally), ranging in size from less than a dozen to a few hundreds of workstations, running mainly internally developed small business applications placed atop one or another SQL server--PostgreSQL (after switch), MySQL, Microsoft SQL (before switch), Oracle--along with an office package (so no Notes and Domino), it is always a smart thing to switch to Linux, and it pays off pretty soon and pretty much. The same hardware with Linux allows you to run more and smarter applications, available freely for Linux and not at all for Windows, in many cases. Internal development becomes a lot faster, easier, and cheaper after an initial phase of learning because you'll always have more and better libraries available on Linux than on Windows. Not to speak that most of them are free and have better support from the operating system, especially for the kind of things you need in business applications, which never run only for one user at a time.

So although I'll agree with you that the switch to Linux requires some initial investment, I won't agree that the costs for such a switch are significantly higher than those of switching to a new version of Windows. They're just slightly higher, and if you compare the costs and benefits of switching to a new Windows version with those of switching to Linux over a period of, say, three years, I'm 100% sure that Linux is the indisputable winner.

So I think what Lotus should do is not a Linux client, but a native Linux Designer! I'm happy with the browser, in fact happier than with the client, I just need a smart development environment for Linux. The day I have a native Linux Designer I'll switch to Linux and forget that Windows exists.