Why this Site?

  • Our Mission:
  • We exist to shine the light of scrutiny into the dark crevices of Wikipedia and its related projects; to examine the corruption there, along with its structural flaws; and to inoculate the unsuspecting public against the torrent of misinformation, defamation, and general nonsense that issues forth from one of the world’s most frequently visited websites, the “encyclopedia that anyone can edit.”
  • How you can participate:
  •  Visit the Wikipediocracy Forum, a candid exchange of views between Wikipedia editors, administrators, critics, proponents, and the general public.
  • 'Like' our Wikipediocracy page on Facebook.
  •  Follow Wikipediocracy on Twitter!

Wikipedia itself isn’t reliable, but that so many people believe it is slashdot.feedsportal.com/c/35…

Google Search

Press Releases

  • Please click here for recent Wikipediocracy press releases.

Google’s Knowledge Graph Boxes: killing Wikipedia?

by Gregory Kohs

Editor’s note: if graphics do not display properly in your browser, please adjust screen size by using Ctrl -.

At the beginning of 2012, Sue Gardner, the woman in charge of the organization that hosts Wikipedia, caused quite a tech media stir when she rolled out a chart that actually had been sitting around for months. The line graph depicted a trend of declining active Wikipedia editors that suggested the English-language encyclopedia would never again see as many editors working away on it as were seen in March 2007. It has been a downhill pattern ever since then. Many of the Wikipedia faithful, adept at mindlessly deflecting criticism, said not to worry as long as there were more and more people visiting the site, it didn’t matter if fewer were choosing to edit the content they found. And besides, other language Wikipedias were supposedly going to take up the global editor slack being let out by the English version of Wikipedia.

Now, two years after Gardner revealed the “holy shit” slide (as she called it) documenting editor loss, Wikipedia is finally losing its readers, as well. Wikipedia analyst Andreas Kolbe broke the news on Wikipediocracy.com on Saturday, January 4th, revealing a series of alarming graphs that unmistakably show that Wikipedia ended 2013 with far fewer page views than it began the year. And not just on the English Wikipedia; the same disappointing pattern was found on the French, Italian, Spanish, Portuguese, German, and other language versions of Wikipedia.

What could explain this decline in viewership of the Wikipedia.org sites? Are readers getting fed up with the unrelenting fundraiser drives that make us feel guilty for not donating to a $50 million, organizationally bloated Wikimedia Foundation budget? Or have readers been shamed into finding more reliable sources that aren’t written by anonymous teens in their parents’ basement? Neither is the likeliest explanation. As with just about every traffic-related puzzle on the web, it traces back to Google, in all likelihood.

The Google hypothesis put forward by one Wikipediocracy participant speculates that Google’s “Knowledge Graph” feature is the culprit. If you’ve used Google throughout 2013, you probably couldn’t help but notice that on many basic topical searches, a sidebar of pertinent info and images appears on the right-hand side of the results page.

 

 

Google hopes that the short, visually appealing semantic data that it displays in the Knowledge Graph will be sufficient to answer your immediate needs, thus keeping you on Google and not letting you slip off to another website to learn more about the subject. Google began to roll out Knowledge Graph to English-language search engine users on a staggered basis through mid-2012. Sure enough, by February 2013, the English Wikipedia began to show its page-view decline. With the success of the English-language Knowledge Graph results, Google expanded the feature to other European language search customers in December 2012. And just as predictably, the visitor traffic to those language Wikipedias began to drop in the early to middle months of 2013.

 

 

The sad irony is that Google is very much for-profit, while Wikipedia is non-profit. Google has donated several million tax-exempt dollars to support Wikipedia in the past, and the Wikimedia Foundation thought that was a lovely series of gifts at the time. But now, Google has figured out a way to take that same Wikipedia content and “import” it directly into Google’s own Knowledge Graph space, where it can be surrounded by advertisements that put money back in Google’s pocket. And if these recent Wikipedia traffic statistics are to be trusted, the shift of Wikipedia “knowledge” over to Google may be exactly what is simultaneously robbing Wikipedia of its readers.

Image credits: Google screen capture (fair use), and Wikimedia Foundation and Gregory Kohs (CC-by-SA)

 

38 comments to Google’s Knowledge Graph Boxes: killing Wikipedia?

  • HRIP7

    It is worth mentioning that if people search for information and find it on Wikipedia, they might edit the article, and become a content-adding Wikipedia editor. If they just find it on Google Knowledge Graph, that avenue of editor growth for Wikipedia is choked off.

    The English Wikipedia has now according to http://stats.wikimedia.org/EN/TablesWikipediaEN.htm for three months in a row had fewer than 3,000 people making at least 100 edits a month (2,858 in September, 2,981 in October, 2,917 in November, December figures not in yet). That particular metric had been consistently above 3,000 since March 2006, peaking at 4,797 in March 2007.

    The decline in almost all the page view graphs seems to take views down to levels not seen since 2011.

  • Radiant Orchid

    That panel of summary information on Google search results isn’t the “Knowledge Graph” (nor is it “semantic data”, but that’s a whole ‘nother discussion). The Knowledge Graph is what underlies the summary panel. It is a “graph” in the sense that data items are connected to each other. Like Da Vinci is connected to the Mona Lisa because he painted it (and understanding the fact that one is an artist and the other is a painting is where that whole “semantic” thing comes into play).

    Calling the display of summary information the “Knowledge Graph” is like calling any Google results page the “search engine”.

  • simsa0

    Thanks Greogory for the nice summary and esp. Andreas for providing the charts.

    Just one additional note:

    It is very reasonable to suggest that Google’s Knowledge Graph project tries to keep “users” on its site by serving their immediate informational needs. To do so, the KGP needs a database from which to retrieve this information. Even if Google should have huge amounts of information in-house, it would benefit immensely by having a machine-readable access to Wikipedia content at large. Enters the Wikidata-project of which Google as been a substantial donor ( http://is.gd/rgyZLo ). Enters that at the end of September 2012, Wikidata’s project director, Denny Vrandecic, left Wikidata project to join Google ( http://is.gd/bEjpqo ).

    All this seems pretty obvious to those following the developments. But it is very helpful to have some factual “evidence” in the form of numbers — if only to counter the habit of many Wikipedians to gloss over or deflect critique.

    • Wow, simsa0 — the Wikidata info about Google and my friend Denny Vrandecic is quite telling, also. I am disappointed in myself for not having pieced that together and included it in this blog post.

      • simsa0

        Don’t beat yourself up, Gregory. This post is fine. Take my comment as a plea to the people at Wikipediogracy to do some in-depth follow-ups on Wikidata and Google (something I’m simply incompetent to do myself and that wasn’t quite the topic of your article either):

        * Can Wikidata be seen as a reaction by the Wikimedia Foundation to the decline in editors of Wikipedia?

        * How much would Google’s Knowledge Graph Project in fact benefit from a successful Wikidata?

        * What happens to the licenses in such a machine-readable access to content on Wikipedia? Esp. when one day Wikidata might becomes machine-writable as well?

        That is: Is the content delivered from KGP and shown on the website of Google-search copyrighted by Google? Is it open? If it is copyrighted, how does that relate to the open licenses of content in Wikipedia?

        Questions over questions …

        And indeed, I find it quite ironic that Denny is a friend of yours. Good to see that differences with regard to Wikipedia don’t need to end friendships. ;)

        • Denny once said to me, “my first loyalty is towards research and not towards Wikimedia / Wikipedia”. That put him in very good stead with me, because I feel the same way. If Wikipedia and the people who ran it would just be HONEST for a change, about 90% of my problems with that project would disappear.

        • ken

          > Is the content delivered from KGP and shown on the website of Google-search copyrighted by Google?

          No, the content is not copyrighted by Google. Google displays the source of the information (usually wikipedia.com) right next to the snippet, so they obviously do not claim copyright.

  • metasonix

    “Enters the Wikidata-project of which Google as been a substantial donor ( http://is.gd/rgyZLo ). Enters that at the end of September 2012, Wikidata’s project director, Denny Vrandecic, left Wikidata project to join Google ( http://is.gd/bEjpqo ).”

    Got it. Thank you.

    Another item to consider: is it possible that Google is silently scraping all of Wikipedia and Wikidata content, plus other databases, with the intention of building its own semantic general knowledgebase? If Google is doing that, they will have very little use for Wikimedia projects going forward. And that will result in the precipitous decline of traffic to Wikipedia. Which already gets “60-70%” of its traffic straight from Google. According to Jimbo himself, quoted in the Financial Times in February 2010.

    Google is NOT Wikipedia’s friend. They are not the average user’s friend, they are nobody’s friend. They are the world’s largest advertising agency, and will clearly do anything to dominate the online advertising business. And they are succeeding. Okay?

  • Tony1

    Guys, I haven’t been here for some time, and I notice the quickly moving sidebar. Very very distracting for readers. I had to cover it up with another window to avoid my peripheral vision being signalled every few seconds. Also too frequent to actually READ what’s in each new entry on that production line. I suggest moving every 15 mins if you really have to have motion.

  • Tony1

    I meant every 15 seconds

  • simsa0

    Right now Google seems to present only abbreviated content taken from Wikipedia. (Click e.g. on the “Feedback / More info” link below such an entry. It shows the source & the license.) So Google acts as a kind of intermediary, condensing larger articles from Wikipedia to the “basic” needs of the user. In what sense could that harm Wikipedia? Its content is still used, isn’t it? Is Wikipedia really dependent on the increase of traffic and page views? It is well-funded enough not to insist on traffic as a metric of “success” (and “value” to cash in).

    With regard to “scraping” I share metasonix’s suggestion. The point really is no longer how to appropriate content but how to use already existing “open” content to create “walled gardens” that no-one can ever leave. These “walled gardens” are far more problematic than those whose entries are controlled and access to “their” content can therefore be monetized.

  • Bob Smith

    As a Wikipedian, I contribute primarily for the good of the world, not particularly for the good of Wikipedia itself or the WMF (although I don’t have any particular gripe with those institutions either).

    So if Google happens to scrape Wikipedia articles for use in its search results, (with the proper attribution of course), I regard that as a good thing. More of my content out to the wide world.

    The only qualification I would give to that, is that there should be no fork of the primary content host, as has happened with WikiTravel/WikiVoyage; so Google can display any text from articles, but should offer people the opportunity to edit the text by going to Wikipedia itself.

  • HRIP7

    For reference, the data plotted on the Wikimedia Report Card graphs are here:

    http://stats.wikimedia.org/EN/TablesPageViewsMonthly.htm

    According to that table, the English Wikipedia’s page views dropped by 21% from December 2012 to December 2013. The German Wikipedia’s page views dropped by 30%. Those for the Spanish Wikipedia by 29%. Those for the Japanese Wikipedia by 25%. Those are massive losses.

    Note that the Wikimedia Report Card graphs show non-mobile page views only. If the Report Card had included mobile (i.e. smartphone) page views, the drops would have been slightly offset by the rise in mobile views, giving us a drop in page views of 12% for the English Wikipedia, 17% for the German Wikipedia, 19% for the Spanish Wikipedia, 9% for the Japanese Wikipedia (which has a very high proportion of mobile use).

    The corresponding Wikimedia data table is here:

    http://stats.wikimedia.org/EN/TablesPageViewsMonthlyCombined.htm

    There is also a table that shows the drop for Wikipedia as a whole, across all languages:

    http://stats.wikimedia.org/EN/TablesPageViewsMonthlyAllProjects.htm

    In that table, Wikipedia (non-mobile) shows a drop of 20% from December 2012 (16.718M) to December 2013 (13.423M). That’s for all languages combined. This 3.3M drop is somewhat offset by a 1.4M rise over the same period in page views for Wikipedia (mobile), but there is still a net drop of close to 2M for the combined total.

    Note that mobile users will not edit, as editing on a mobile phone is nigh-impossible. In terms of editor recruitment and retention, non-mobile page views are crucial to Wikipedia’s health.

  • John Howard

    The Google Knowledge Graph (originally called Freebase by Metaweb — a company Google acquired in 2010) is (for the most part — Google includes proprietary data that it has purchased) Creative Commons data. It is available at http://freebase.com and you can download the whole graph as an RDF dump or explore it on the site.

  • Anyone who is curious the new infoboxes that emit micro-data, may be interested in the departure essay written by Surturz, now renamed Riggr Mortis. It has been deleted from WP, but I have reposted it under CC-by-SA license at http://neotarf.wordpress.com/2013/10/14/surturz-riggr-mortis-wikipedia-as-database/.

    Regards,
    Neotarf

  • hi Gregory, thanks for the data graphic. this is just what I expected in December 2012, when Google introduced the knowledge graph in Germany http://www.seolytics.com/blog/20121219/googles-knowledge-graph-affects-wikipedia/

    But when we compare the developement of the knowledge graph in Europe vs. the US, this trend will go on, I think. http://www.seolytics.com/blog/20130315/new-analyze-the-knowledge-graph-with-seolytics/

  • Cla68

    Somehow Google is tweaking their knowledge graphs to present WP article info in a more aesthetic way. For example, the WP article for Kanu Sanyal has a “citation needed” tag in the intro. However, the knowledge graph blurb for that article that shows up when you Google “Kanu Sanyal” does not include that tag.

    • Radiant Orchid

      More aesthetic or less complete? In either case, the Wikimedia API allows you to extract all or part of an article with or without various markup included.

  • [...] Google’s Knowledge Graph Boxes: killing Wikipedia? [...]

  • [...] Google’s Knowledge Graph Boxes: killing Wikipedia? — Here is another example of how Google embraces free and open to build its market share, and then uses the resulting dominance to close off competitors. “The sad irony is that Google is very much for-profit, while Wikipedia is non-profit. Google has donated several million tax-exempt dollars to support Wikipedia in the past, and the Wikimedia Foundation thought that was a lovely series of gifts at the time. But now, Google has figured out a way to take that same Wikipedia content and “import” it directly into Google’s own Knowledge Graph space, where it can be surrounded by advertisements that put money back in Google’s pocket. And if these recent Wikipedia traffic statistics are to be trusted, the shift of Wikipedia “knowledge” over to Google may be exactly what is simultaneously robbing Wikipedia of its readers.” [...]

  • [...] berichtet es das Blog Wikipediocracy. Gleichzeitig sieht es auch den Verursacher – den Google Knowledge Graph. Wie im obigen Bild [...]

  • [...] közösségi enciklopédiát egyébként is hevesen támadó Wikipediocracy.com szerzője, Gregory Kohs arra hívja fel a figyelmet, hogy a Google Knowledge Graph bevezetése óta látványosan csökken [...]

  • Paul van Poppelen

    interesting discussion in an interesting forum which is highly critical of the quality of at least some of the information on Wikipedia. Frankly I love Wikipedia as it more often than not leads me into new directions previously not considered or thought of. I am critical of some of the content too, but as a critical reader I know (at least in my fields) when to take things as they are and when to start verifying… actually that brings me to my main gripe, the number of dead links or links hidden behind a pay wall in cited references.. the latter we can’t do much about but the former… if only those things were cached somewhere so I could go back to them. As it stands a ‘dead citation’ is no better than a non-referenced statement of (dubious or disputed) fact and in keeping with ‘policy’ should really be removed. But I digress… what is lacking in this discussion is the fact that the Google side bar (or whatever they want to call it)suffers from the same problem as what Wikipedia is alleged to suffer: unverifiable statements and facts, but this time in a highly condensed form, making it even more difficult to eek out truth from fiction. In our criticism of Wikipedia and obvious delight from some who are keen for its demise, let us not forget that a sometimes “dumb” Wikipedia is still infinitely preferable to a “dumbed down” Wikipedia which is what the Google sidebar in reality is and probably will remain unless Google spends massive resources verifying its content rigorously which I somehow can’t see happening..

    • HRIP7

      Google could always start a Googlepedia fork, list it higher than Wikipedia on their search results pages, and put ads in it. The Wikipedia licence explicitly allows that. Google could earn billions a year from the ads, which theoretically would enable them to employ staff to verify the content, thus differentiating Googlepedia from Wikipedia.

      If that ever happens, you read it here first! :)

  • [...] but not least een artikel van Gregory Kohs. Inmiddels kennen we allemaal de Knowlegde graph van Google die ons handige informative voor onze [...]

  • And what’s to stop Google from going the next step and showing to, say, me, a “different” version of the da Vinci knowledge graph from the one it shows to, say, you? Just slightly different, let’s say, “suggestively” different. Does the WP license have anything to say about using its data for test marketing purposes? How to make me click? Which KG style makes my click to the next page take oh-so-just-a-little-bit longer? Maybe my da Vinci KG slightly stresses his work in, say, theater design to me, because of my interest in that subject, but to you, there’s a slight focus on Leo’s mirror writing because of your interest in internet security (an oxymoron, okay, but this is just an example). If I take the bait, do I start getting more ads for Broadway show ticket sites? If I “take the bait” five times over the next two years, how much smarter Google will be! Note to NSA (off topic, just being polite): Hi Guys/Gals at NSA, happy MLK Day!

  • [...] Google’s Knowledge Graph Boxes: killing Wikipedia? [...]

  • wikipediafan

    I’m not surprised to that this trend has occurred. I have desired to help edit wikipedia and even became an editor to make some minor edits, starting out. As I joined editor discussions, it became clear to me that many in the English version were questioning the decision to make all links “no follow”. On the one hand, this has reduced editors whose sole goal was to create SEO links. On the other hand, it exposes that the majority of editors were not attracted to editing by the mere intellectual stimulus. I would submit to this discussion that while Google’s Knowledge Graph may contribute to the trend of loss of Wikipedia editors, it is not the greatest contributing factor. Links that pass no juice is the reason. And if readership is falling off, it isn’t only because the graph either. The broad scope of Wikipedia English versions’ “no follow” means no link juice is being passed from internal linking, not just externally. Again, a number of editors have debated that this is not a good practice for Wikipedia. And readership number reductions, from this perspective, seems to underscore their concerns as valid.

  • [...] גוגל-ויקיפדיה, שגרמה לאנציקלופדיה ירידה של כ-20% בתנועה במהלך 2013, כפי שניתן לראות מן הדו"חות של ויקיפדיה עצמה, היא חלק [...]

  • [...] More and more material from Wikipedia is displayed on Google’s own search results pages, thanks to the Google Knowledge Graph panel and Google’s new snippet overlay. No wonder that there is such great interest from the most [...]

  • [...] Kohs, “Google’s Knowledge Graph Boxes: killing Wikipedia?” (Wikipediocracy, January 2014). Kohs looks at a new feature of Google that means less [...]

  • What do you think is the best? get informations directly or having to pass by wikipedia to get the information. The best for the user is to get the info as fast as possible. So the Knowledge graph is usefull according to me.

  • [...] “Google’s Knowledge Graph Boxes: killing Wikipedia?” (Gregory Kohs, Wikipediocracy) [...]

Leave a Reply

  

  

  

You can use these HTML tags

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>