Want to take part in these discussions? Sign in if you have an account, or apply for one below
Vanilla 1.1.10 is a product of Lussumo. More Information: Documentation, Community Support.
What would people think about running Google Analytics on the nLab? It’s be easy to set up: just add a few lines of javascript to the <head> template that’s used on every page.
What information would you like? I’m not too keen on GA and I think we can generate a lot of the information that GA would provide straight from the server logs without any of the privacy and inaccuracy issues that go along with GA.
What are your complaints about GA?
Personally, I don’t even know raw numbers about how much the nLab is used. It would help me get a feel for what sort of a scale of system we’re working with, for when we discuss technical and hosting issues. It would be interesting to know the geographic distribution of readers, what percent of them ever make an edit, and what sort of paths people take through our links. Do people use the pull-down “tables of contexts”? (I’ve found that I never do, but I’m probably not typical… but it’s impossible to say without data.) Where do most people arrive here from? How many come from a google search, how many from MathOverflow? Are there blogs that we don’t know about where the nLab is being cited? Those are just a few things that occur to me off the top of my head.
Okay, in the last 24hrs (roughly) we have had:
show)Blogs might take a bit longer to look for as I’d need to extract all the referrers and scan through to see where they are. The “pull-downs” are almost impossible to track unless you know exactly what links are in the pull-downs and so know that if someone clicked on that link from this page then they used a pull-down. This is because the pull-down effect is pure CSS and so sends no signal to the site that it’s been used (so GA would be equally useless here, I’m afraid!). Geographic information isn’t hard but would take a slightly more sophisticated script than I can knock up in 5 minutes.
GA is mainly for marketing purposes. The little extra it gives on top of scanning through the logs is verging on invasion of privacy as it involves tracking a user’s path through the internet.
That’s great! Can you set up a cron script so we can collect that data over time?
I’m a little puzzled by the 884 edits. Are those requests for the “edit” page or the “save” page? I would have guessed a lot fewer than 884 edits happen in 24 hours; surely even Urs can’t account for more than a couple of hundred. (-: Are most of them hits on “edit” (e.g. from spammers or bots) that never make it to “save”?
Is it possible to correlate the number of times each IP visited with their referrerers?
Yes, “save” would have been a better choice than “edit” to look for.
The access log give us a fair amount of information on each request: http://httpd.apache.org/docs/2.2/logs.html#accesslog. Then Instiki’s log gives us a bit more about what Instiki does with each request. From those, one can build up a fair picture of what’s going on. What sort of thing are you after? It’s probably worth thinking a bit about how to present the data before writing the script.
Basically I’m curious about how different people use the nLab. For instance, knowing how many people actually click on links might be useful as page authors, to know in terms of how much explanation should be given on each page versus referred to the relevant other page. I’d also like to know how many of our visitors just drop by once and how many are “regular readers” (which would involve connecting data over weeks or months). One reason this occurred to me is because there was some talk of having an nLab session at ScienceOnline 2013, and although that fell through for this year, if there were something like that in the future then it would be nice to have some statistics to help explain to an unfamiliar audience what the nLab is and how it works.
1 to 7 of 7