The Signpost

News and notes

24,000 votes later and community position on image filter still unclear; first index of editor satisfaction appears positive

Contribute  —  
Share this
By Tom Morris, Skomorokh, Jarry1250 and Jorgenev

Personal image filter referendum highlights polarization

The breakdown of scores for the first question in the personal image filter referendum.
A word cloud of the most frequently occurring words in participants of the image filter referendums' comments, a sample of which were assessed as 30% positive, 29% neutral, and 41% negative.
The results of the image filter referendum were released this week. In this "largest exercise of its type for the Wikimedia Foundation", some 24,000 votes were cast, more than 7000 of them accompanied by comments, which are still being analysed. The referendum featured a small number of questions, each asking voters to attach importance to a number of potential features; as such, it attracted considerable criticism for those who felt it was unclear how voters would signal their dissatisfaction with the whole idea of a filter.

The results are unlikely to calm the rhetoric on either side of the debate. With mild support shown overall—the most general question had a median result of 6 (on a scale from 0–10, where 5 was "neutral")—there is probably enough encouragement to ensure that the proposal is not abandoned altogether, and some useful results were gathered with regard to priorities. On the other hand, about 3750 respondents (16% of the sample) gave a score of zero to the broadest question, "It is important for the Wikimedia projects to offer this feature to readers", the clearest indication yet that a significant body of editors would oppose the implementation proposed by the Foundation regardless of its features. (This result looks set to be endorsed by a poll run in parallel on the German Wikipedia which currently indicates that about fourth-fifths of Wikipedians there are opposed to the measure as stated.) A third group consider the referendum to have been badly mismanaged in a way that would render the result meaningless.

As British Wikimedian Michael Peel commented, the poll probably points towards a "no consensus" result. As a result, the next move of the Foundation is unclear. In all likelihood it will choose to alter the proposed implementation to build a new consensus, since it is dubious as to whether the Foundation could now meaningfully proceed without convincing at least a small proportion of those currently skeptical to the idea. One possible compromise would be on whether or not there was a single global implementation of the filter. User:FT2 added that "enabling on some wikis and not on others" may yet be a good way to "leave more people feeling fairly satisfied".

Index of editor satisfaction released

Mani Pande, Wikimedia Head of Global Development Research, has revealed on the Foundation's blog a new metric as a tool for analysing community health: the Wikipedia editor satisfaction index (WESI). The WESI is based on the answers to two questions.

  1. Editors are asked to select two adjectives from a list of eight—four positive (Collaborative, Intelligent, Helpful, Friendly), four negative (Arrogant, Unfriendly, Rude and Dumb)— that described their perception of the editing community.
  2. Editors are asked if feedback through other editors had helped them become a better editor or not.

The two options from Question 1 are scored for one point each: +1 if the response was a positive adjective, and −1 for a negative. The response to Question 2 is scored as +2 for helpful, and −2 for feedback having been "a bad experience". Added together, they span a range from +4 to −4, which is normalized to a 0–10 scale.

The initial results of compiling the index were greeted by Pande as "encouraging": around 47% of respondents gave a score of 10/10, and about 77% of the editors surveyed scored 7.5 or higher, which she took to indicate that "the majority of our editing community is very satisfied with their experience" of the project and have "a healthy assessment of fellow editors". Delving deeper into the breakdown of the findings, Pande isolated three factors critical to determining an editor's satisfaction with their contributing experience: being offered help, enjoying the respect and recognition of their peers, and receiving adequate explanations for when their contributions are reverted. It is expected that the WESI will be established as an ongoing metric for measuring satisfaction, to yield further insights into the self-reported experiences of Wikimedians in the future.

In brief


The last week saw the following milestones among WMF-supported projects:

+ Add a comment

Discuss this story

These comments are automatically transcluded from this article's talk page. To follow comments, add the page to your watchlist. If your comment has not appeared here, you can try purging the cache.
  • Regarding the referendum result, would it not be worth mentioning that in the small, but representative, sample of comments already analysed, over 40% were negative, while only 30% were positive and another 30% neutral. It's quite a big difference that shows a significant amount of opposition... - HIGHFIELDS (TALKCONTRIBUTIONS) 23:38, 5 September 2011 (UTC)[reply]
    • Though with reference to that, it is not clear if the comments were talking in negative terms about the idea of a filter, or complaining about poor design of the referendum itself (a related but different issue). - Jarry1250 [Weasel? Discuss.] 09:06, 6 September 2011 (UTC)[reply]
      • I would observe that on the broadest image filtering question question, more people voted 10 than 0, more people voted 9 than 1, more people voted 8 than 2, more people voted 7 than 3, and more people voted 6 than 4. Discounting neutral and undetermined votes, 64.6% of 20,935 declared voters were supportive. I also find the percentage of comments which were negative is surprisingly low: discounting the likes of "Looks good" and "Why not?", the majority of comments at an RfA with similar statistics would be negative. Admittedly we are still in the realms of no consensus for short term implementation, but there is certainly a clear steer for the Foundation to find a way in which this can be implemented in a way which will have zero impact on editors who do not wish to be filtered.

        Questions over the Foundation's competence aside, I'm struggling to comprehend why there remains fierce idealogical opposition to this. In particular, it is absurd that a Muslim who wishes to work on articles as fundamental as Muhammad (or perhaps more relevantly one as controversial as Jyllands-Posten Muhammad cartoons controversy) is forced to either ignore his faith, or disable all images on a sitewide basis. If this results in some would-be Muslim editors boycotting the site – and I challenge anyone to credibly claim otherwise – we are moving away from personal reference and into the realms of systemic bias. —WFC10:49, 6 September 2011 (UTC)[reply]

        • My concern is not so much with individual editors -- though I certainly lack respect for any religion that feels that any image ought to be forbidden -- than it is with people in authority requiring these filters to be in place for their subordinates/employees/subjects/etc. Powers T 21:10, 6 September 2011 (UTC)[reply]
          • But they already do that. Filtering systems have been around for yonks. They're far more advanced than a simple category-based image filter (which you should be able to turn off?). - Jarry1250 [Weasel? Discuss.] 21:12, 6 September 2011 (UTC)[reply]
            • There is a distinct line between filtering and censorship. The key to ensuring that the former does not become the latter is to make the system opt-in for individual users, to ensure that a filtered user is not missing out on information (we should be providing alt text for the benefit of blind users anyway), and to give users the option of viewing "offending" material should they wish. —WFC21:45, 6 September 2011 (UTC)[reply]
            • There are ways around them; why provide the censors with additional tools? Powers T 23:04, 6 September 2011 (UTC)[reply]
  • I think that the community position came across quite clearly in the discussion pages following the referendum- that there is a small majority of editors and significant support from interested reader-groups for the option of a personal filter - but that there is also a large minority who are strongly and philosophically opposed to the filter and see it as "enabling censorship" (Notably from As well as a large number of editors who are still confused about what the image filter would mean. In a sense that is pretty much what we all knew before. It will be interesting to see how the foundation moves forward as I don't see the position of either camp changing (I am in the former). Ajbpearce (talk) 13:12, 6 September 2011 (UTC)[reply]
  • Also to add, there is a detailed critique of the survey process on the results discussion page that everyone should read, it lays out in very clearly the significant flaws in the whole survey processs. Ajbpearce (talk) 13:53, 6 September 2011 (UTC)[reply]
    • I'm not sure you can run the "survey wasn't well written" and "survey gives clear results" in parallel. (Most complaints about survey design do not seem to be entirely watertight to me, anyway, so there is bound to be an argument over that, too.) As I put forward above, the result is probably "no consensus". Whether it will stay like that is another matter. For example, many details about the proposed filter are yet to be worked out - e.g. not deploying it on the German Wikipedia. That could well be enough to get support behind the proposal. - Jarry1250 [Weasel? Discuss.] 14:13, 6 September 2011 (UTC)[reply]
      • Your right, what I meant was that the community position seems fairly clear to me, after reading the discussions surrounding the referendum - not that the referendum itself was much good in providing much useful information. I didn't phrase that very well, so I have updated my response.Ajbpearce (talk) 19:20, 6 September 2011 (UTC)[reply]
      • The problem is that there is no mandate for the filter. The Harris report specifically targeted that recommendation at the community and specifically not at the board. The survey explicitly begs the question. The community discussions are, by and large, against the proposal. Rich Farmbrough, 17:52, 6 September 2011 (UTC).[reply]
    • Ajbpearce, you wouldn't have meant to link to this comment? If not, I believe this comment makes some worthwhile points. Regardless of how one might think about this proposal. -- llywrch (talk) 04:29, 7 September 2011 (UTC)[reply]


Does the WESI metric seem particularly flawed to anyone else? There are many negative experiences one could have with fellow editors that don't fit into one of the four negative categories. And even so, I still wouldn't ascribe any negative quality to "the editing community" as a whole, even if a significant minority possessed that negative quality. Powers T 21:13, 6 September 2011 (UTC)[reply]

  • It does to me too. Not only is it way to community interaction focused —community is very important but not the sole thing people care about—, I think the answer to if the community has helped you grow as a Wikipedian is going to be somewhat independent of how much you like the community. Also I question the value of a scale where 50% of the scores are at a single extreme. jorgenev 23:30, 6 September 2011 (UTC)[reply]
  • Yeah, it's terrible, like a parody of survey statistics. My jaw dropped. It wouldn't be so bad if the "pick two from this set of many more positives than negatives" weren't so inherently biased. The fact that the Signpost write up is primarily about how it reached its inevitable conclusion would be charming if it wasn't so sad. Perhaps the Foundation should hire a professional survey statistician before attempting any further surveys. (talk) 09:35, 9 September 2011 (UTC)[reply]


The Signpost · written by many · served by Sinepost V0.9 · 🄯 CC-BY-SA 4.0