We are developing a web application, and one of or major concerns are about browser support. Develop a cross-browser application is expensive due to the time required for testing in legacy browsers (IE6-9).
Our interface is usable in all modern browsers: Firefox 3.6+, Chrome, Safari 3.1+ and Opera 10.6+. We are deeply considering throwing IE away in favor of Chrome Frame and advising the user to download a decent browser.
What are the odds of doing this, beside the obvious?
PS: It is a paid application for a restricted user base.
You’ll sometimes find that there is a large set of users with no choice as to which browser they use, so requiring or disallowing a specific browser could be a show-stopper. Large corporations and government agencies tend to restrict their employees’ access to their work computers, to the point where they can’t install any new software without approval from the IT department. Same thing with public computers in places like schools and libraries. (And most people aren’t likely to know about alternatives like PortableApps, which would likely violate the spirit of the IT departments’ policies anyway.) As other people have said, it depends heavily on your target audience.
In some cases for paid applications, it’s even risky to recommend a specific browser — e.g. if you say “works best with Firefox” but a company’s policy mandates IE, the reaction could be “we won’t buy that for our employees, it won’t work well in our environment.”
I think there are two “decent” approaches to this problem. The direct approach (perhaps what you call “obvious”) is to put in extra effort to gracefully degrade your application’s functionality in browsers that don’t support what you’re trying to do. (And whenever possible, avoid using user-agent to decide what browser is hitting your application.) A related approach is to rely on a client-side framework, such as jQuery, GWT, YUI, etc., to abstract away a lot of the browser-specific issues.