I'm joining in disappointment too. Extreme coupling to an young platform like wikidata created a lot of trouble for making anything with the project, and kill all momentum and made it look bad.
I feel like this kind of micro-contributions are the only effective way we are going to be able to engage mobile users in improving wikipedia, and the micro contributions itself as infrastructure could've been something really empowering.
We should've designed and tested lots of different micro-contributions experiments and mechanisms and measured results, without worrying about where the underlying data will end up (maybe wikipedia, maybe wikidata, maybe other projects). With how the project was managed we only have data for a Question - Yes/No/MultipleAnswer type of micro-contribution, which shouldn't be enough to judge the whole underlying idea.
Now there's a lot of skepticism towards the idea in any of its forms :(
I would love to see a *lessons learned* (not just from the data but from the struggles of the project), we should get the most of it while it is still fresh on our memories.
On Tue, Jun 2, 2015 at 4:13 AM, Oliver Keyes okeyes@wikimedia.org wrote:
Totally; as I think my email made clear, I was aware that the limiting factor here was the sheer cost of building out the infrastructure. The core question, though, was what the project would be replaced with - what those "highest return-on-investment" projects were.
On 1 June 2015 at 20:45, Jon Katz jkatz@wikimedia.org wrote:
Hi Oliver, Thanks for sharing your disappointment. I do not think you are alone in wanting to see wikigrok continue and grow. I would clarify that the 'success rates' you allude to were for reader engagement and accuracy,
not
in actually improving our projects by filling in important gaps in
wikidata.
A great deal of work would be required to build out in order for this project to have a scalable impact on wikidata.
I am not saying that casual contributions are going away, simply that we
are
going to recognize our resource limitations and evaluate opportunities
for
them based on highest return-on-investment. We currently have 5
developers
working on readership for the entire web (due to some temporary leaves)
and
there might be smaller wins using casual contributions that work towards
our
end goal, but don't require the heavy upfront investment. This doesn't
mean
we don't take on big thorny problems, just that we take a step back and
see
if there are ways to subdivide them into smaller projects along the way.
Best,
Jon
[1]
https://www.wikidata.org/wiki/Wikidata:Requests_for_permissions/Bot/WikiGrok
On Mon, Jun 1, 2015 at 12:05 PM, Oliver Keyes okeyes@wikimedia.org
wrote:
I'm personally incredibly disappointed; this was the most successful intervention I'd seen anyone try in a long while, if ever, and the results blow me away. My question would be "what interventions with similarly high success rates are going to be worked on instead?" - I assume that we're not working on them because we can achieve the same outcome through easier-to-implement interventions. I would be interested to hear what those interventions are.
On 1 June 2015 at 14:57, Jon Katz jkatz@wikimedia.org wrote:
Hi,
TLDR: Wikigrok proved that readers are interested in and capable of making casual, mobile contributions to Wikipedia. We are putting continued development of the 'Wikigrok' casual contribution feature on hold
until
we have a plan for optimally harnessing this interest/capability.
Background Given the growth of mobile traffic on wikipedia and the challenges inherent to traditional editing on a mobile device, Wikigrok was proposed as a way to test if regular wikipedia readers would be interested in making
smaller,
more casual contributions to wikimedia projects while reading
Wikipedia
on a mobile device.
Results By early 2015, the results were in: readers were relatively interested in engaging with the feature[1]. Some oft-quoted comparisons include:
3x the number of unique responders as mobile editors during test
period
(4.5K editors, 12.3K WikiGrokkers), even with WG on sample of
articles &
users 1.5x better clickthrough than 2014 Fundraising full-screen mobile
banner
(I actually do not have references for these, as they are borrowed quotes) Furthermore, we found that the quality of responses was rather high [2,3].
Future The original thought was to use these responses to fill in gaps in Wikidata and our initial test results (2 weeks worth) were successfully ported over in late April [4]. However, in order to production-ize the system, we would have to:
scale and develop queries against the new wikidata query service create an article parser to identify potential multiple choice answers for each question create a system for attributing aggregated results to the specific contributors (per Wikidata bot request discussion[5])
None of these are unsurpassable, but we have learned a great deal and, at this stage, we believe that further effort should be devoted to evaluating areas of need and fit before we commit additional efforts to specifically porting information into Wikidata.
Please feel free to reach out if you have any questions or concerns about this decision. Best,
Jon
[1] https://meta.wikimedia.org/wiki/Research:WikiGrok/Test2 [2] Quality of responses, version A:
https://www.wikidata.org/wiki/File:All_Campagins,_Scatterplot,_version_(a).p...
[3] Quality of responses, version B:
https://www.wikidata.org/wiki/File:All_Campaigns,_Scatterplot,_version_(b).p...
[4]
https://www.wikidata.org/wiki/Special:Contributions/WikiGrok?limit=500
[5]
https://www.wikidata.org/wiki/Wikidata:Requests_for_permissions/Bot/WikiGrok
Mobile-l mailing list Mobile-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/mobile-l
-- Oliver Keyes Research Analyst Wikimedia Foundation
-- Oliver Keyes Research Analyst Wikimedia Foundation
reading-wmf mailing list reading-wmf@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/reading-wmf