Erik Moeller, 15/01/2014 01:21:
For the time being, in both cases we're counting crawler traffic, correct? If so, have we attempted to quantify the degree of variation in crawler traffic over time? My main concern is that major changes in crawler behavior could distort the overall numbers.
What sort of crawlers/bots are more likely to have such an impact? Are things like /wiki/Special:Export* or /wiki/*?action=render and so on filtered out? No idea how many of those (and all sorts of parameters to index.php explicit or implicit) there could be.
Matthew Flaschen, 15/01/2014 03:17:
Yes. Regular page views can also use index.php (https://en.wikipedia.org/w/index.php?title=Earth). For example, navigation popups constructs the main article link like this.
Navigation popups are really a bad example to mention in this case. :)
index.php with only a title URL parameter should be treated identically to /wiki/ requests without URL parameters.
How many actual/"legit" page views can we expect from there, as opposed to bots of all sorts, stuff loaded in background etc.? I think the subject "Page view data with Wikipedia app" went in the right direction, it's probably better to hunt for biggest sources of views legit but missing or fake but counted. Nemo