On Tue, May 17, 2016 at 10:09 PM, Vladimir Batov < Vladimir.Batov@constrainttec.com> wrote:
On 05/18/2016 02:01 PM, Robert Ramey wrote:
Actually this has been entertaining on a number of levels.
Could we agree that this particular tool won't be of much use to our discussion?
Indeed. The only sensible conclusion possible. Moving on.
Oh no you don't. 1. I showed a graph https://www.google.com/trends/explore#q=%2Fm%2F034w42%2C%20cmake%2C%20bjam&cmpt=q&tz=Etc%2FGMT%2B6 that has no indications of bad data. It shows reasonable comparative numbers, uses proper category disambiguators, and has correct labels. 2. Robert showed another graph https://www.google.com/trends/explore#q=C%2B%2B%20boost that illustrated a bug in the system when using the term "C++ Boost". The fact that it is a bug is demonstrated by a.) investigation of the labels over the graph indicating "C+Boost" was the term used, and b.) comparing "C++ Boost" to C++ and seeing that "C++ Boost" has more hits, which is not a reasonable thing. 3. Folks don't understand how this system works. This can be remedied by going here https://support.google.com/trends/?hl=en#topic=6248052. These three points *do **not* imply that we should throw out the original graph. Just because the numbers don't support a particular position doesn't mean we should throw out all kinds of FUD over the data. The decline of Boost is real and we should face that reality. -- David Sankel