Where do you draw the line on a modern browser?

In his article 
 
 Modern browsers among Czech users, Ondřej Bojar presents unusual statistics on browser use across 
 23 working days and eleven days off around the end of April and beginning of May this year. The data covers
 1.68 million users in all. It shows the proportions of modern and nonmodern browsers by
 operating system, type of machine and type of connection.

It’s fascinating data, with one small catch. How do we tell modern browsers from the rest?
 Where is the line that determines how advanced a browser is, and which standards or specifications should define it?
 Those basic questions put the article’s conclusions in a different light. There are two competing views of the web, as a place for applications and as a place for information.

For a web whose job is to provide information, we could count even Internet Explorer 5.01 as modern. 
 That browser already lets you separate the visual 
 presentation of information from the information itself. Its support for CSS
 and client-side scripting is plenty for an informational website.

For web applications, though, this is where I part company with the author.
 I wouldn’t call any member of the Internet Explorer family 
 a modern browser. Its level of support and compliance with the standards needed for 
 web applications, such as B2C services, contrasts sharply with browsers built
 on Gecko, for example. I’d even say that technologically, Mozilla 1.0 was at least half a generation ahead of Internet Explorer 6.0. The gap keeps widening now that development of the standalone Internet Explorer has stopped.

To answer the original question of how to distinguish
 modern browsers from the rest, we’d have to agree on a basic criterion
 to judge them by. Without one, we’re left with a subjective view of where
 to draw the line on technological maturity, as in this article.