Unfortunately the original article was published in late 2006, so the statistics are obsolete - but still interesting.
Upon some contemplation these numbers aren't that surprising. ASP.NET ever since Visual Studio 2002 has made it incredibly easy to create Web services contract-last. The fact that the WSDLs actually made it easy to identify the implementing platforms suggest that many of them were created contract-last, though some tools will still deploy their own identifiable WSDL even if you create the web service contract-first - so one cannot make the assumption that all identifiable WSDLs relate to contract-last WSs. Of course JAX-WS now supports contract-last just as easily as ASP.NET does - which isn't necessarily a turn for the better as the generated WSs are often not all that WS client-friendly.
However it is interesting to note that if you needed access a public web service in 2006 on the internet there was a +66% chance that it was implemented in .NET and that you would have to deal with the quirks that were typical for WSDL/SOAP web services that were built with that platform.
Wonder what that percentage is these days? [ November 10, 2008: Message edited by: Peer Reynders ]
I really wonder if these numbers are representative. DotNet may well be way up front, but Apache Axis way behind Apache SOAP? That sounds strange. I bet these numbers would look quite different if non-public and commercial services had been considered as well.
Originally posted by Ulf Dittmer: Apache Axis way behind Apache SOAP? That sounds strange.
I had the same feeling about it but unfortunately there is noting specific said about when the sample was taken. Notice that both Axis and Apache SOAP are associated with the "news and reference" category. I would associate "Apache SOAP" with early adopters that haven't moved to Axis yet while "not so early adopters" would immediately go to Axis. Axis 1.0 was available end of 2002, so it is possible that the sample might be as old as end of 2003 or 2004. During that time many organizations with ASP-based web sites might have finally made the migration to ASP.NET and some would have opportunistically added SOAP web service support.
Originally posted by Ulf Dittmer: I bet these numbers would look quite different if non-public and commercial services had been considered as well.
Possibly. However not every organization had the need for a "distributed component architecture" - and ASP was quite popular (probably thanks to VB) in its day for plain business web sites. These organizations would have had the tendency to (possibly grudgingly) move on to ASP.NET unless someone could make a business case for a total transplant to J2EE's distributed component architecture (though interface21, Spring's predecessor did emerge as early as the end of 2002, marking the growing doubt that J2EE's complexity was in fact necessary, even back then).