I know nothing of Evans Data Corporation, who ran the survey in question, and I have no idea what the results were of the survey, as they want to sell the results.

You'll note that the Register harps on how much PHP is dropping, but the survey was not global, and in fact, EDC found that PHP use in North America is growing significantly. (Perl was up as well, but less so in comparison)

I also have no idea how these increases were calculated -- if it's based off of the total population of users, we need to also factor in how many people have been laid off or otherwise aren't considered to be part of the population. If we had the other numbers from the survey being cited, it would be much easier to tell if there might be some form of bias of this nature. (for all we know, developers are moving to management of COTS software, so their programming language 'use' is going down, but it's a factor of the nature of their jobs changing)

I'd be interested to know what the rest of the survey was, and why Perl, PHP, Python and Fortran were in the section 'ARCHITECTURE AND TECHNOLOGY ADOPTION', while C, Java, XML and ColdFusion were in 'LANGUAGE USAGE'. (ColdFusion being similar to PHP, and XML not being a programming language)

Unless I see the whole report, I'm going to take it with a grain of salt -- the Register took a long report, which may have known biases in their assumptions and methodologies (and may even be fully disclosed in the report), and they've taken the part that they feel is sensational, and posted it to their website to drive up traffic and ad revenue.