TED conferences/TED speakers challenge/Lessons learned

The TED speakers challenge was set to run from 6 June 2016 to 6 July 2016 and due to Wikimania and end-of-school vacations, probably did not get as much community attention as it might have had it taken place earlier in the year. A last-minute decision on the 6th to extend the challenge by one week was surprisingly successful, possibly because Andy & Jane were able to talk about the challenge at Wikimania.

That said, with 150 new articles spread over 45 language Wikipedias, the concept of a writing challenge to encourage multi-lingual re-use of content based on existing lists has proven its worth yet again. As a first step to summarize some findings about how this challenge was setup and how the results reflected expectations, this page tries to collect concepts and issues and how they were dealt with.

Expectations

edit

It's hard to make estimates for writing challenges, but based on previous results run internationally, an estimate of a few hundred new articles was made. The points system took some time to agree on, with the emphasis being on inclusion (Wikipedia, Wikidata, and Commons contributions could earn points) and though only 12 languages were initially included in the data donation, no effort was made to restrict the challenge to those languages and in the end three more languages were added. The tracked languages have translated labels in their TED talk items, but didn't include any data for their speaker items. Beforehand, Andy & Jane linked TED talk items to speaker items and tried to create labels for speaker items in as many languages as possible. Improvements to Wikidata speaker labels, which is needed for many languages, was not a candidate for points. Though points were included for raising articles to a certain quality level, it was already known and discussed that this is a slippery slope as quality is widely different among subjects and Wikipedia language projects.

Without stating it explicitly, rough expectations were that the numbers of new articles would be contributed according to the size of various Wikipedias, with English leading, as the prizes are books in English. On reflection, it would have made more sense to include prizes that were not books in English, such as T-shirts or coffee mugs. Also, since English already had well over a thousand articles of the 1700 or so speaker articles, it also makes sense that it would be more difficult to create new articles in that language. Beforehand lists were created per language, with all languages included that had already enabled the ListeriaBot user account, enabling easy list updates. For tracked languages that had translated TED talk labels, the Dynamic lists feature was used to enable users to browse talks on Wikidata in their language.

Results

edit
 
Birthplaces of TED speakers
 
Bubble chart of occupations of TED speakers
Some basic numbers:
New articles created = 162
Articles translated = 76
Articles improved = 12
Number of participants = 22
Number of languages = 47
Photos added = 576
Referenced statements added on Wikidata = 191

The winners are selected based on self-reporting of points. As is common in writing challenges, a healthy competition took place among lead participants. The list of winners is on the Data page. A total of 47 language Wikipedias took part and of the selected 15 languages that were included in the data donation, only 12 ended up with new articles. It is unknown whether the Listeria lists (see lists per language) were actively used or whether participants just randomly selected articles to write based on looking up their favorite TED speakers. In fact it is still unknown whether participants were TED fans or not! About a third, or 55 of the 150+ created articles were in the tracked languages. Since the tracked languages included mostly large Wikipedias it would seem that translated TED talk labels had little to no effect on participation. Also worth noting was a comment by a German speaker that German Wikipedia doesn't lend itself to writing challenges in general because of pending changes, which means the participant must wait for such changes to be approved before counting them and no one working in such a challenge wants to wait.

Regular updates on challenge progress could not take place for all facets of the competition as the points system was a bit complex. New articles, a measurement made easy through the "Site statistics" feature of the Mix-n-Match tool, was monitored and posted on the Data subpage of the competition. The updates were also posted with a link in Twitter and Facebook and this seemed to help attract participants throughout the challenge. The languages page was designed to allow participants to browse both the list of speakers as the list of talks. Participant Papuass ran a query to track down articles about female speakers with the most interwikilinks without an article in their language Wikipedia. Presumably articles with the most interwikilinks are also the most notable and therefore the easiest to create new articles for.

It is unknown how many participants saw the posted status updates or whether they were influenced by it. This feature measures the number of articles per language of a dataset, and comparing these over time with the start position led to some interesting results in terms of #new articles vs. percentage of coverage increases per language Wikipedia.

Top ten Wikipedias with new articles created during the competition (several tied for 10th place)
wiki Tracked? Language #articles 6 June #new articles Total %increase
ukwiki Ukrainian 151 23 174 15.2%
hywiki Armenian 53 20 73 37.7%
enwiki Y English 1161 17 1178 1.5%
lvwiki Latvian 60 12 72 20.0%
fawiki Y Farsi 251 11 262 4.4%
nlwiki Y Dutch 297 10 307 3.4%
ruwiki Y Russian 305 5 310 1.6%
scowiki Scots 45 5 50 11.1%
tgwiki Tajik 12 4 16 33.3%
dewiki Y German 521 3 524 0.6%
arwiki Arabic 201 3 204 1.5%
plwiki Y Polish 244 3 247 1.2%
shwiki Serbo-Croatian 75 3 78 4.0%
bnwiki Bengali 67 3 70 4.5%
Top ten Wikipedias with a starting #articles greater than 100 with the highest increase in coverage during the competition
wiki Tracked? Language #articles 6 June #new articles Total %increase
ukwiki Ukrainian 151 23 174 15.2%
fawiki Y Farsi 251 11 262 4.4%
nlwiki Y Dutch 297 10 307 3.4%
ruwiki Y Russian 305 5 310 1.6%
arwiki Arabic 201 3 204 1.5%
enwiki Y English 1161 17 1178 1.5%
plwiki Y Polish 244 3 247 1.2%
nowiki Norwegian 177 2 179 1.1%
cawiki Catalan 187 2 189 1.1%
hewiki Hebrew 212 2 214 0.9%
Top ten Wikipedias with a starting #articles less than 100 with more than 1 new article resulting in the highest increase in coverage during the competition
wiki Tracked? Language #articles 6 June #new articles Total %increase
hywiki Armenian 53 20 73 37.7%
tgwiki Tajik 12 4 16 33.3%
lvwiki Latvian 60 12 72 20.0%
scowiki Scots 45 5 50 11.1%
uzwiki Uzbek 19 2 21 10.5%
bnwiki Bengali 67 3 70 4.5%
shwiki Serbo-Croatian 75 3 78 4.0%
srwiki Serbian 78 2 80 2.6%
tawiki Tamil 81 2 83 2.5%

Pageviews of TED article per language Wikipedia

edit

Using the pageviews api for "langviews" you can see how often the article about TED itself was read during the challenge, which gives an indication of the popularity of TED as an organization overall. Using the English Wikipedia article for and looking at the dates as of 7/12/2016 results in the date range 6/23/2016 - 7/12/2016. These are the results per language, which show significantly more interest in English than in any other language:

# Language Page title Badges Pageviews Average
Totals 56 languages 38 unique titles 93,760 4,688 / day
1 en TED (conference) 39,809 1,990 / day
2 ja TED (カンファレンス) 10,164 508 / day
3 es TED 7,222 361 / day
4 fr Conférence TED 5,199 260 / day
5 de TED (Konferenz) 4,534 227 / day
6 ru TED (конференция) 4,217 211 / day
7 zh TED大会 4,033 202 / day
8 ko TED 3,384 169 / day
9 pt TED (conferência) 3,382 169 / day
10 it TED (conferenza) 2,163 108 / day
11 tr TED Konferansları 1,804 90 / day
12 fa تد 1,246 62 / day
13 ar تيد (مؤتمر) 1,152 58 / day
14 pl TED (konferencja) 1,143 57 / day
15 nl TED (conferentie) 1,126 56 / day
16 vi TED (hội thảo) 524 26 / day
17 sv TED (konferens) 407 20 / day
18 he ועידת TED 368 18 / day
19 hu TED (konferencia) 276 14 / day
20 cs TED 213 11 / day
21 uk TED (конференція) 196 10 / day
22 da TED (organisation) 163 8 / day
23 fi TED 152 8 / day
24 sk TED 84 4 / day
25 no TED (konferanse) 82 4 / day
26 ca TED 78 4 / day
27 el TED (συνέδριο) 64 3 / day
28 bg TED 51 3 / day
29 th เทคโนโลยี การบันเทิง และการออกแบบ 42 2 / day
30 ro TED (conferință) 41 2 / day
31 zh-yue TED (會議) 40 2 / day
32 ps ټېډ 30 2 / day
33 hy TED (համաժողովներ) 29 1 / day
34 hr TED (konferencija) 26 1 / day
35 simple TED (conference) 26 1 / day
36 lv TED 23 1 / day
37 sr ТЕД (конференција) 22 1 / day
38 lt TED (konferencija) 19 1 / day
39 af TED 17 1 / day
40 bo TED 16 1 / day
41 kk TED конференциясы 16 1 / day
42 hi टेड (सम्मेलन) 15 1 / day
43 mk ТЕД (конференција) 15 1 / day
44 gl TED 15 1 / day
45 bs TED (konferencija) 14 1 / day
46 sl TED (konference) 14 1 / day
47 et TED 14 1 / day
48 be-tarask TED 12 1 / day
49 eo TED (konferencoj) 12 1 / day
50 be TED, канферэнцыя 11 1 / day
51 ml ടെഡ് (സമ്മേളനം) 11 1 / day
52 scn TED (cunfirenza) 11 1 / day
53 sq TED 10 1 / day
54 is TED (ráðstefna) 9 0 / day
55 tt TED (конференция) 7 0 / day
56 ta ரெட் மாநாடு 7 0 / day
edit

Using Quarry it is possible to find links out of a Wikimedia project to an external website, in this case TED.com. Example is here Links from English Wikipedia to TED.com.

Language Count 9 March Count 21 July Query URL
en 2591 2695 http://quarry.wmflabs.org/query/7963
de 212 219 http://quarry.wmflabs.org/query/7964
es 207 218 http://quarry.wmflabs.org/query/7965
fr 266 270 http://quarry.wmflabs.org/query/7966
it 136 144 http://quarry.wmflabs.org/query/7967
nl 117 126 http://quarry.wmflabs.org/query/7968
ja 229 237 http://quarry.wmflabs.org/query/7969
pl 127 135 http://quarry.wmflabs.org/query/7970
ru 163 178 http://quarry.wmflabs.org/query/7971
sv 61 69 http://quarry.wmflabs.org/query/7973
vi 53 58 http://quarry.wmflabs.org/query/7974

Social media outlets

edit

Social media posts were added to facebook and twitter through various personal feeds using the hashtag #TEDwiki. Many more tweets and fb posts were made without the hashtag however. Specific aspects of the challenge were posted such as the Sparql query on birthplaces of TED speakers. No post was made at the official Wikipedia facebook page however, though any Wikimedian can make requests for posts there through the message facility. An announcement post made by TED was reposted to the facebook groups Wikidata+GLAM and GLAM-Wiki Global. Though the hashtag was posted on the challenge page and users were encouraged to use it in edit summaries, this was clearly not communicated well and only one user in fact used it , see Wikipedia edits by hashtag. Perhaps it is still early days for hashtag use in edit summaries?