[This is the first in a series of posts from my time at the British Museum. If you would like to assist in this project (or just eavesdrop), please contact me to join the regular mailout list and receive news (and prizes) first. The project’s homepage is at http://en.wikipedia.org/wiki/Wikipedia:GLAM/BM]
Yesterday was my first official day as the volunteer “Wikipedian in Residence” at the British Museum (BM) – as far as I’m aware, the first serious attempt from the GLAM sector globally to bring Wikipedia in-house. The underlying mission: To build a relationship between the Wikimedian and British Museum communities that is mutually beneficial, sustainable and replicable.
The British Museum’s digital strategy specifically speaks about sharing the collection and the institution’s expertise with the wider-web, beyond their own website. A sub-point of this is that the British Museum should engage with partnerships with the “knowledge sites” elsewhere online. These two points place a relationship between the British Museum and Wikipedia as not just a good thing™ but as a strategic priority.1
I’ve been preparing the ground in the months before I arrived by putting together measures of the existing relationship – qualitative and quantitative – in order to provide a baseline against which I can compare the relationship at the end of my pilot project. Without this, it would be impossible to objectively assess whether my project here was successful or whether it could/should be implemented elsewhere.
Increased WP article quality = increased pageviews = increased clickthroughs to your GLAM website. Therefore, if you want to increase the number of people accessing the deep resources of your GLAM’s website:
* encourage qualitiative improvement in the Wikipedia articles that link to it
* make it easy for Wikipedians to reference your GLAM website.
Part 1 – Qualitative baseline:
This graph takes every article which appears in “Category:British Museum” and sub-categories such as “collection of the British Museum” (but ignoring sub-categories of articles about staff or trustees) and scores them on intersecting quality and importance axes. Thanks to Nihiltres for coding the “BM related” article assessment infrastructure up for me. The quality rating is consistent across all of Wikipedia and is based on the extensive documentation at the article assessment pages. On the other hand the importance rating is on the basis of how fundamental an article is to an understanding of the British Museum. So for example, the article about George Bernard Shaw is high quality but low importance, whilst the article about the Water Newton Treasure is high importance but low quality.
There are 148 articles that have been tagged as related the BM (no doubt this will increase by the end of the month) and these are spread seemingly randomly across the matrix. This is because, as volunteers, Wikipedians work on articles that interest them personally – not because the subject is more “important”. On average, the articles with the highest quality receive significantly more traffic simply by virtue of their higher-than-average quality. If something is good it receives inbound links (from the rest of Wikipedia and the wider web) and inbound links beget more inbound links. For the technically inclined you can test this on 2000 randomly chosen articles v. 2000 Featured Quality articles here.
Therefore, it would seem that the most obvious candidates for immediate improvement are those two articles listed as “stub” class but of high importance – the Water Newton Treasure and the Vindolanda tablets. With any GLAM organisation, not only with the BM, ways of making it easier for Wikipedians to use your website include:
- Persistent neat URLs for records. The National Library of Australia’s system is the champion of this. Not popups, not search-strings masquerading as URLs.
- Clear information/research pages. Not splitting information about an subject across different sections of the website making it difficult to find, collate and cite. If it can’t be found, it can’t be read, and definitely can’t be linked to.
- Citation templates. If you have a preferred method of being cited, make it easy for Wikipedians to use that method. You can even create a dedicated citation template for Wikipedians to use when citing your website ensuring consistency and completeness of metadata.
Part 2 – Quantitative baseline:
The graph below takes all of the articles in the same categories being assessed in the qualitative data and aggregates their combined page-views into one monthly number to produce a measure of popularity. Thanks to Magnus Manske for creating the fabulous “Treeviews” tool for me for this purpose. These numbers are marked in BLUE. The graph compares these to the BM’s website analytics that track inbound visitors originating from anywhere in Wikipedia. These numbers are marked in RED.
[n.b. to make the comparison clearer, I’ve dropped the pageviews (blue) down by two orders of magnitude. So – a monthly pageview on this graph listed as 4,000 is actually 400,000. To see the raw data go to http://en.wikipedia.org/wiki/Wikipedia:GLAM/BM#Quantitative]
Setting aside some extraordinary spikes (explained below) there is an absolutely clear correlation between page-views and click-throughs. In fact, averaged across the whole set it works out at almost exactly 100:1. Obviously some links are more clicked on than others but nevertheless it is fair to say that page-views = click-throughs at a quite a predictable rate. More page-views = more click-throughs!
Points of note:
- Jan/Feb 2008 – The spike in click-throughs is due to the fact that the BM made its online catalogue available during those months. All of a sudden there were a whole lot of linked footnotes that could be added where before there were none.
- April/May 2008 – The GIANT SPIKE in page-views, dwarfing the rest of the graph, is entirely the result of the article Crystal Skull. The spike corresponds to the release of the latest Indiana Jones film – “The Temple of the Crystal Skull”. People searching for the film stumbled across the article about the museum object instead. Interestingly, a fair number of readers actually clicked through to the BM website after they found the article they originally sought (as seen by the smaller spike in click-throughs for that month). This demonstrates that Wikipedia can successfully convert the casual pop-culture googler to a cultural researcher.
- July 2008– The big drop in page-views is because the stats failed to compile that month, not because people stopped visiting Wikipedia. Ignore that one.
- January 2009 (in fact January 15th each year) – The article about the British Museum appears on the main page in the “on this day” section. Even this extremely small reference on the main page results in a visible spike in page views for the whole category and also click-throughs to the BM’s own website.
- April 2010 – The article Disasters of War, a series of sketches by Francisco de Goya, appeared as “today’s featured article” during this month which accounts for the page-view spike. However, despite many of the original sketches residing in the BM there is not a single link out to the museum’s catalogue and therefore there is no equivalent spike in click-throughs. This is because all of the BM’s resources are object centric (catalogue references for individual sketches) whereas Wikipedia’s article is subject centric (one article about the whole series of sketches). There actually is not anything on the BM website that talks about the whole group of sketches and so Wikipedians cannot easily reference the BM even if they tried.
- Overall – Prior to the qualitative assessment of all British Museum related articles, many articles were not listed in a BM category. Consequent to this assessment, many uncategorised BM related articles were discovered, added to the categories and therefore counted in the quantitative survey. The results gave a significant increase in page-views reported. For example, before I undertook the survey the combined pageviews for March 2010 were 350,340. After the comprehensive survey and discovery of more related articles this figure jumped to 513,049 – an increase of 32%.
It needs to be said is that improving Wikipedia articles can and should be an end in their own right but this should not be at the expense of sharing link-love with organisations that actually host the original research that Wikipedia requires. Indeed, ultimately Wikipedia should be encouraging our readers to leave Wikipedia via external links in the footnotes. Wikipedia is a place to start but not end your research after all. Wikipedians should count every person who leaves Wikipedia through a linked footnote to continue their research as a satisfied customer.
1 The responsibility, the number of skeptical eyes watching (from both communities), and the potential to do amazing things together are all very large. And yes, I agree with those who have (pointedly) said to me that there are more qualified people than myself to have this opportunity. All I can say is that I’m doing the best I can to represent Wikimedia and free-culture well, I’ve earned my chops and that I’m a volunteer doing it because I believe it should to be done.
You should be able to do a reverse lookup too. From the BM’s end they should be able to look at the referrers from Wikipedia by month, as well as look at the time on site for those visitors – which will help you affirm your claim that every person that who leaves WP to BM is satisfied (at least from the point of view of BM – not WP). It is also important to check whether the page views on BM are generated from logfile data or from JS/browser data (Google Analytics etc). If it is logfile data then you may find there is even a greater relationship.
BTW Is Treeviews supposed to work on ALL WP articles? I’ve tried True Blood and it seems to come back with zero results.
Not sure what the problem was that you had with Treeviews, but yes, it works for any category on Wikipedia. Here, for example, is True Blood for May: http://toolserver.org/~magnus/treeviews.php?depth=9&date=2010-05&cats=True+Blood&combination=subset&autolang=0&doit=1
11 pages with a combined total of 918,825 pageviews!
Thanks for posting this.
I’m looking forward to reading more on your adventures at the Museum.
Pingback: Kristina Alexandersons blogg » Kunskap, korv och Wikipedia -CoooooLt!
Pingback: Efter Creative Commons-workshop med svenska museer « Wikimedia Sverige
Pingback: End of my residency | Witty’s Blog
Pingback: Projekt:Internet i Sverige « Wikimedia Sverige
Pingback: Des wikipédiens partout… « Archives masala