Hi folks,

the use of AllOurIdeas for prioritizing community requests came up in a conversation about the Community Tech wishlist this morning. I realized I'd never followed up on this thread to report the results of the informal poll I set up in March.

As of this morning, nearly 1,000 votes have been cast. This is the full ranked list of ideas that were submitted and evaluated: http://allourideas.org/wikidata/results

I copy here the top 10 results and their score*:
  1. Wiktionary data 89
  2. Use of data on Wikipedia 83
  3. Support in-context editing of wikidata content shown on Wikipedia articles so that 'edit-this-page' keeps working seamlessly 75
  4. Statistics of re-use of data from Wikidata 72
  5. Automatically generated list pages on Wikipedia 64
  6. Federation: Re-use of wikidata items and properties on other Wikibase instances. 64
  7. Unit conversion 64
  8. Wizard-style dialog for entering references 64
  9. Tools to monitor the changes in a query 63
  10. Automatically suggest references from high quality sources that we as a community have compiled 57

(* "the score of an idea is the estimated chance that it will win against a randomly chosen idea. For example, a score of 100 means the idea is predicted to win every time and a score of 0 means the idea is predicted to lose every time.")

The tool is still accepting new votes and new ideas, so I encourage you to check it out if you haven't yet: http://allourideas.org/wikidata


On Sat, Mar 12, 2016 at 4:16 PM, Dario Taraborelli <dtaraborelli@wikimedia.org> wrote:
Hey all,

Lydia (or somebody operating the @Wikidata handle :) posted this question on Twitter and a few great ideas started trickling in

I went ahead and created an AllOurIdeas poll, seeded with the first ideas posted on Twitter, to crowdsource the generation of new ideas and produce a robust ranking.

If you're unfamiliar with AllOurIdeas, it's an open consultation engine allowing people to choose which idea they like best via pairwise comparisons (I am cc'ing Matt Salganik, the project lead). It's very simple on the surface but it uses algorithms such as the Condorcet method to test how strongly each idea performs against another, reducing the weighing of the oldest ideas to create a level playing field for newly created ideas and preventing gaming or self-promotion of one's own ideas.

Try it out or post new ideas: the more votes it gets, the higher the confidence of the ranking. Real-time results and statistics are here



Dario Taraborelli  Head of Research, Wikimedia Foundation
wikimediafoundation.org • nitens.org • @readermeter