Talk:WikiFactor

It's certainly worth considereng. Proxima Centauri 08:05, 4 March 2009 (EST)

I like your "wikiFactor" :-) - for MediaWikis, number_of(edits)/number_of(good_pages) is interesting too. --Wolf | talk 10:00, 4 March 2009 (EST)

I'm not sure I understand this

hmm TedErnst | talk 14:52, 5 March 2009 (EST)

Lets have a look on our wiki:

Popular pages (of WikiIndex):

Position (p) most visited first	Page name (views)	Relation views [<=>] 1000*p
1.	Welcome ?(572,833 views)	572,833 >  1,000
2.	Bulbapedia ?(42,344 views)	42,344 >  2,000
3.	Community Portal ?(28,489 views)	28,489 >  3,000
. . .	. . .	. . .
11.	Wiki.ThePPN ?(14,724 views)	14,724 > 11,000
12.	Books ?(12,807 views)	12,807 > 12,000
13.	TourBusStop ?(12,371 views)	12,371 < 13,000

If you count the entries with views > 1000*p (marked with green) you get the wikiFactor (12 for WikiIndex). I hope this helps. Greetings from Berlin --Wolf | talk 03:07, 6 March 2009 (EST)

Thanks for the illustration Wolf. That is precisely how the wikiFactor is calculated. The wikiFactor is not an indication of the size of a wiki, but rather the interest that a wiki has attracted. It is entirely feasible to populate a wiki with many 'uninteresting' pages, that attract very few visits other than the Main Page. The wikiFactor is based on the highly popular (at least amongst the scientific community) h-index (Wikipedia entry). -- Carl McBride (talk) 05:39, 6 March 2009 (EST)

Ah, okay. And would this be added through code of some kind, or by hand? 24.12.195.33 22:06, 6 March 2009 (EST)

I imagine that initially the value of the wikiFactor would be inserted by hand, as is the case at the moment with the wiki_pages parameter in the Size template. -- Carl McBride (talk) 05:30, 9 March 2009 (EDT)

Very cool!

Carl, I do worry about how this will scale to non-mediawiki wiki. Any ideas on that? Best, MarkDilley

I think most implementations of the wiki model have a facility for listing the most popular/visited pages. Instructions on exactly how to do this for each wiki type could be outlined on the category page, in a similar fashion to the list on the Template:Size page for the size of a wiki. -- Carl McBride (talk) 13:22, 9 March 2009 (EDT)

New template

Given the generally positive response to the Proposal:wikiFactor I have started work on an 'experimental' template along the same lines as the Size template: Template:Size experimental. -- Carl McBride (talk) 11:52, 23 March 2009 (EDT)

Please, have a look on Proposal:integrate new data into template:wiki! Greetings --Wolf | talk 04:04, 24 March 2009 (EDT)

I included two new parameters wiki factor and wiki factor URL into template:size. Have fun! --Wolf | talk 17:25, 24 March 2009 (EDT)

WikiFactor and caching

I collect stats for Rosetta Code blog posts and the admin there noted to me not long ago that the views count in the wiki statistics was skewed because of caching. If someone views a cached page in the wiki, it is not counted in the wikiviews because it doesn't trigger the script that keeps track of those views. I realize there's nothing much that can be done here about that except maybe put up a disclaimer or somehow create/promote a caching plugin that does count cache views on the stats page. I thought it should be noted somewhere though. --129.21.93.164 10:08, 14 August 2009 (EDT)

One could just as easily argue that if a page was not cached then the statistics would be skewed. If someone has already seen a page (and thus forms part of their cache) then viewing it again is not a fresh view with a fresh pair of eyes. All said and done, if everybody has the same working conditions (i.e. their wiki does perform caching) then, with regard to the wikiFactor, everybody is working on a level playing field. -- Carl McBride (talk) 10:18, 14 August 2009 (EDT)

There are several forms of caching that can be occur. Internally, MediaWiki (and perhaps some other engines) cache various items of data used to reconstruct a page. Busy sites often use a "front-end" cache such as Squid (in reverse-proxy mode) or Varnish to accelerate the delivery of pages and files, shielding the web server from duplicate requests from different clients where permitted by response headers. Finally, there are the intermediary server and user caches, also advised by caching and age headers. The latter caches are the ones you refer to, but on busy sites it is probably the server-side cache for anonymous users that has the most effect on perceived popularity. Indeed, it will greatly affect the hit-count for most-accessed pages (since they are very likely to be served from cache), and so disproportionately affect wikiFactor. GreenReaper 14:37, 11 November 2009 (EST)

There is something that can be done - use a tool like Google Analytics (extension) to track website usage. This is triggered on each access, regardless of whether a page has been served from a cache. To me, this is the right way to do it - after all, a citation is still counted even if someone copied it from another paper's list of citations. Those wikis using caching are the very ones that you would expect to have a high wikiFactor - some get hundreds or even thousands of HTTP requests a minute. GreenReaper 14:25, 11 November 2009 (EST)

Interesting. I opened an account on Google Analytics, but I stalled not knowing what to do with a wiki site. Thanks for the info on the extension! -- Carl McBride (talk) 04:56, 12 November 2009 (EST)