Wikimedia Bugzilla is closed!

@domas: If you should consider doing that, please provide this information in separate logs. Else, the logs would be blown up too much.

@Brianna: You know that just because an article (containing an image at the very bottom) has been accessed 10,000 times for example (incl. reloads, bots), that image has not at all been *viewed* or even useful that often, although accessed? If an image is at the very bottom of a very long article, only very few people might pay attention to it, actually. In my opinion, images of interest are clicked on to get them enlarged. Thus the pagecounts for "Image:Foobar.png" should be a pretty good indicator already. If Flickr wants to track their images, there would be the possibility to combine CheckUsage with the existing pagecount logs. Image *usage* could be computed somehow instead to have a bit more accurate figures...

(In reply to comment #1)
> @Brianna: You know that just because an article (containing an image at the
> very bottom) has been accessed 10,000 times for example (incl. reloads, bots),
> that image has not at all been *viewed* or even useful that often, although
> accessed? If an image is at the very bottom of a very long article, only very
> few people might pay attention to it, actually. 

Of course I realise that. Just as a page view doesn't mean a person read an article from start to finish.

However if the user doesn't even load the image then it's not going to be recorded in the logs, is it? 

In my opinion, images of
> interest are clicked on to get them enlarged. Thus the pagecounts for
> "Image:Foobar.png" should be a pretty good indicator already. If Flickr wants
> to track their images, there would be the possibility to combine CheckUsage
> with the existing pagecount logs. Image *usage* could be computed somehow
> instead to have a bit more accurate figures...
 
I strongly disagree. Firstly there is the difficulty of figuring out precisely when the image was placed in particular pages.

And it is just a nobrainer that people don't go to the image page for every image they look at. View counts are about eyeballs. Directly measuring views will not be 100% accurate -- just as our current article pageviews aren't. But it will be far more accurate than the indirect method (with compounded errors) you suggest.

And of course, I agree these logs should be separate to the page view logs!

Let me first say/write that I actually do *support* this request of course (what is possible should be done), but ...

> However if the user doesn't even load the image then it's not going to be
> recorded in the logs, is it? 

... note that if you visit [[en:Normal_distribution]] for example, everything is loaded from beginning on; text as well as all images (thumbs). This means that all images on that page would be logged every time the page gets accessed.

> And it is just a nobrainer that people don't go to the image page for every
> image they look at.

That's true. When someone can see all details from beginning on, he/she will probably not click on it, yes. But image clicks already could be existing indicators. I just think that both methods have major drawbacks, and both could be fairly helpful (but different).

I raised the importance of this bug. Solving this bug would mean a huge improvement for most image related projects like photo competitions and GLAM partnerships.

Marking as shell as this needs ops attention, although I'm not sure what steps should be taken. Hex, can you assign this to someone? It would be nice having this for WLM.

[mass-moving wikistats reports from Wikimedia→Statistics to Analytics→Wikistats to have stats issues under one Bugzilla product (see bug 42088) - sorry for the bugspam!]

Today we worked on https://commons.wikimedia.org/wiki/Commons:GLAMwiki_Toolset_Project/NARA_analytics_pilot . The data we produced should be suitable to publish.

Some early proof-of-concept visualisations using Limn:
<http://glam-metrics.wmflabs.org/>