There's an interesting article by Nate Koechley over on the Yahoo Developer Network entitled [weblink:45/27]. Nate proposes a "broader and more reasonable" definition of support, where a graduated model of browser compliance encourages web developers to support all browsers. Nate defines three grades of browser C, A and X:

• C-grade browsers are well-known, incapable, antiquated and rare. QA tests a sampling of C-grade browsers, and bugs are addressed with high priority.
• A-grade browsers are well-known, capable, modern and common. QA tests all A-grade browsers, and bugs are addressed with high priority.
• X-grade browsers are generally unknown, assumed to be capable, modern, and rare or fringe. QA does not test, and bugs are not opened against X-grade browsers.

Nate subtly differentiates "graceful degradation" and "progressive enhancement" suggesting the former prioritises appearance while the latter prioritises content. I like this differentiation. I can see how the adoption of this approach could help clarify a QA strategy, ending long arguments over pixel perfection in rarely used browsers. This approach could enable us to decide on a policy where, for example, IE 5.5 for the Mac is deemed a C-grade browser (where it's acceptable to deliver unenhanced but highly accessible content), while IE 6.0 for the PC is deemed an A-grade browser (where content may be enhanced by modern functionality and the results are tested thoroughly for visual accuracy).