First, most everything that's important to "web services" is either a) under a non-GPL license, b) dual-licensed GPL and something else or c) GPL but pinned to GPLv2 and can't be relicensed without the permission of everyone who's ever contributed to it.
\r\n\r\nSecond, there are some big conceptual problems with trying to "close the web services loophole". What the heck is a "web service"? That could end up being even murkier than GPLv2's definition of linking. What's the difference between using GCC to transform C source code into an executable binary, and using, say, the PHP interpreter to transform PHP source code into a renderable web page? How could a license be written which doesn't force source distribution of the former but does force source distribution of the latter? What about a language like Ruby which is used both for system and web programming? Its interpreter is GPL; would proprietary software written in Ruby suddenly stop being proprietary if someone ran it through an interpreter embedded in a web server? How would that make sense? And if that's not how GPLv3 would work, if it would only apply when the CMS was GPL, then how does this matter? Google didn't start from a GPL search application, and neither did any other big web-services player that I know of; they all used open-source platforms, but they rolled their own services/frameworks/whatever.
\r\n\r\nThird, throughout this thread you are arguing in entirely the wrong fashion if you want to accomplish anything. Millions of geeks could scream their heads off that this would be bad for business, and the FSF would not change its course because that's not a cause which matters to them. To get through to the FSF you must speak their language, which means you need to have a reason why GPLv3 would negatively impact user freedom.