Page MenuHomePhabricator

Find out why the visual editor is so used at the Portuguese Wikipedia
Closed, DeclinedPublic

Description

It's not as colorful as the ipê-amarelo flower. It doesn't taste like Ginjinha. Then, why do Portuguese-speaking editors like the visual editor so much?

Its uptake at pt.wiki doesn't seem to have evident reasons, and it's so different from the average value that deserves a little investigation as to why it's so successful there. We may be particularly interested in why IPs prefer it to wikitext editing, but let's not forget that earlier this year we also found out that 27 registered editors there were among the top 500 VE users at the time.

Here are a few facts/data points/remarks.

  • VE has the same config there as on most other big wikis (primary tab, opt-out);
  • There are not so many templates with TemplateData (around 130, although this includes common Cite templates, Infobox person);
  • Citoid is enabled, and only provides basic info with pt news sites;
  • There is no local documentation about the product (only a redirect to mediawiki.org's one);
  • The guides on mw.org are not entirely translated (currently, the user manual basically is - although it's at 76% for pt-br; when you look at all the VE-related documentation, pt is at 58% and pt-br at 75%);
  • The interface itself is translated almost entirely;
  • VE is not mentioned or promoted in local help pages (except for a few tutorial pages with less than a dozen views in 90 days each);
  • VE is not mentioned or promoted in interface messages (i.e., after looking for an article which doesn't exist);
  • the local feedback page isn't very active and never really hosted conversations around the product (although, let's thank again @He7d3r for all his efforts there);
  • CLs haven't engaged in a particular way with this community and haven't held specific initiatives;
  • A couple of instructors involved in Education programs at pt.wp seem to be using it (cc @FloorKoudijs: maybe they're recommending it in class?). Details on these initiatives (and others which may be initiated by community/chapters/user groups etc. like edit-a-thons) are much welcome;
  • Finally, VE is also much used at pt.books, where it's in opt-out mode (95 out of 500 edits made with it to mainspace pages as of a random check minutes ago);
  • pt.wiki is "mostly edited from Global South, in contrast to earlier concerns the VE would be only usable for GN people with state of the art computers".

If we could use T89970 or T104439, we could survey editors directly in a quick and easy way...

Event Timeline

Elitre raised the priority of this task from to Needs Triage.
Elitre updated the task description. (Show Details)

Of course, can be re-assigned as needed.

nshahquinn-wmf set Security to None.

Is there a wildly popular site in the Portuguese-speaking world which VE's interface may remind people of?
Which are the most used word processors in Portugal and Brazil?

Probably Microsoft Word and/or LibreOffice Writer. I'm not aware of any popular site which might be similar to VE's interface.

nshahquinn-wmf lowered the priority of this task from Medium to Low.Dec 7 2015, 11:46 PM

I just noticed that https://pt.wikipedia.org/wiki/Wikip%C3%A9dia:Editor_Visual/Coment%C3%A1rios also contains more feedback left from within VE than other wikis (could be just an obvious consequence of higher usage, of course).

(Sometimes at night, when I can't sleep, I stare at the ceiling and think about this task.)
The only "explanation" I could give myself so far is that pt.wiki may be at a point where it is more common for editors to make a certain kind of edits, and that VE may be particularly convenient for that specific workflow (for example, there may be an ongoing focus on adding sources, and using Citoid makes that just easier; or maybe the articles are generally very complete and users mostly need to make "aestethic" changes, like fixing punctuation, etc.).
If the community could help with this, an analysis of sample edits could help us moving forward.
There may be tools quite ready for that (see https://phabricator.wikimedia.org/T90894 and https://en.wikipedia.org/wiki/Wikipedia:Labels/VE_experiment_edits ).

We may be particularly interested in why IPs prefer it to wikitext editing

Maybe because for newcomers "Editar código-fonte" sounds almost like programming software, and in comparison "Editar" might sound almost not scary? I'm not joking.

Also note that in Portuguese "Editar" is at the left, which feels more like first option (in Spanish is at the right, and "Editar código" feels more like first option). That might also help.

Is it a community which runs many workshops and editathons, teaching VisualEditor first? ca.wiki do, VE all the way, and that also counts (I wonder what are the stats of ca.wiki).

We may be particularly interested in why IPs prefer it to wikitext editing

Maybe because for newcomers "Editar código-fonte" sounds almost like programming software, and in comparison "Editar" might sound almost not scary? I'm not joking.

Sure, but that should be true for several other Wikipedias as well (also, why would it scare Portuguese-speaking IPs more? :) )

Also note that in Portuguese "Editar" is at the left, which feels more like first option (in Spanish is at the right, and "Editar código" feels more like first option). That might also help.

Again, that's actually the default setting for Wikipedias where the visual editor is the primary editor (es.wiki being an exception).

Is it a community which runs many workshops and editathons, teaching VisualEditor first? ca.wiki do, VE all the way, and that also counts (I wonder what are the stats of ca.wiki).

Yeah, this is mainly my third last point above (plus others though: evidence suggested that there was no particular teaching content available, at least online). Feel free to change the description if you feel like your ideas belong there, ty.
(For ca.wiki, I'll take a look at the visual editor tag there to make at least a very rough estimate - it's the only thing we got.)

Added some ptwiki users who might have ideas on this (or might just be interested in future results).

(FYI'ng Amir, who's also interested in similar subjects.)

(FYI'ng Amir, who's also interested in similar subjects.)

Thank you!

We may be particularly interested in why IPs prefer it to wikitext editing

Maybe because for newcomers "Editar código-fonte" sounds almost like programming software, and in comparison "Editar" might sound almost not scary? I'm not joking.

Sure, but that should be true for several other Wikipedias as well (also, why would it scare Portuguese-speaking IPs more? :) )

This is a curious and not improbable theory: A weird label that makes an otherwise popular and well-known feature less used :)

The following lists of translations to all languages may be somewhat helpful for comparison:

Some of them are completely cryptic to me, but I can kinda understand Romance and Slavic languages... and I see that some say "source", some say "source code" and some say something like "wiki text" or "wiki syntax". May be worth comparing the condition there.

Another thing to check is the survival rate of VE edits there, as well as things like appearance of <nowiki> tags. See example in Hebrew (all the relevant parts are translated to English).

(Sometimes at night, when I can't sleep, I stare at the ceiling and think about this task.)

It takes many lives till we succeed
To clear the debts of many hundreds years,
That's why we are here!

Mobile editing doesn't seem to be playing a role either. Out of 500 VEdits I just checked, only 31 were made through mobile site/device. (I'll note that users from two countries where Portuguese is an official language, Angola and East Timor, can edit WP from their mobile devices at no cost thanks to Wikipedia Zero.)

We announced the beta test on the Watchlist from June to October 2013 https://pt.wikipedia.org/w/index.php?title=MediaWiki%3AWatchlist-details&type=revision&diff=37047797&oldid=36119894, but I don't know if that has anything to do with the popularity of the feature among registered users. We did not see much resistance to its use as on many other languages, but again I can not explain why. Portuguese Wikipedia is also not as techy-driven as other language versions. There is only a handful of editors with programming skills. I personally started using it mostly because of citoid and my aversion of filling out citation templates by hand.

One thing that nobody mentioned yet for some reason: Maybe it just has less bots running?

Not sure I understand how that would be related.

In other news, from pt.wiki we seem to be getting most of the comments posted via the internal tool which publishes VE feedback at mw.org.

It's a tad complicated to filter bots out of the editing stats because the revision table doesn't store this info and recent changes data is deleted after some time.

@Amire80, that's a good hypothesis. Since we now have the editor month dataset available, it'll be quick to check it out.

First, let me see if there are fewer bots running on ptwiki by looking at the percentage of edits made by bots so far this year:

select wiki, (sum(if(bot is true, edits, 0))/sum(edits) * 100) as bot_edits_percentage
from staging.editor_month
where 
month >= "2016-01-01" and 
wiki in ("enwiki", "frwiki", "eswiki", "ruwiki", "zhwiki", "arwiki", "ptwiki")
group by wiki;

wiki	bot_edits_percentage
arwiki	66.9569
enwiki	16.4292
eswiki	7.2240
frwiki	20.2926
ptwiki	5.3710
ruwiki	15.3820
zhwiki	19.5088

Wow, I wasn't expecting that. Only 5% of ptwiki's edits come from bots, which is quite low compared to other wikis. So it looks like Amir's intuition was completely correct. It also deserves mentioning that a huge 67% of arwiki's edits come from bots.

Let me see how the rates of VE use look when we exclude bots:

select wiki, (sum(visual_edits)/sum(edits) * 100)
from staging.editor_month
where
month >= "2016-01-01" and 
wiki in ("enwiki", "frwiki", "eswiki", "ruwiki", "zhwiki", "arwiki", "ptwiki") and
bot is false
group by wiki;

arwiki	2.9301
enwiki	4.1624
eswiki	7.2653
frwiki	12.1690
ptwiki	17.5178
ruwiki	10.9085
zhwiki	1.6305

So even when we correct for the low number of bots, ptwiki is still an outlier. VE is not fully available at enwiki, eswiki, arwiki, and zhwiki, but even comparing with ruwiki and frwiki where it is fully available, ptwiki is still considerably higher (17.5% compared with 12% and 11%).

Also, let me just check if excluding mobile edits (where VE isn't an option) changes anything:

select wiki, (sum(visual_edits)/sum(edits - mobile_app_edits - mobile_web_edits) * 100) as non_mobile_ve_percentage
from staging.editor_month
where
month >= "2016-01-01" and 
wiki in ("enwiki", "frwiki", "eswiki", "ruwiki", "zhwiki", "arwiki", "ptwiki") and
bot is false
group by wiki;

wiki	non_mobile_ve_percentage
arwiki	3.3021
enwiki	4.4691
eswiki	8.1926
frwiki	12.6547
ptwiki	19.2702
ruwiki	11.3542
zhwiki	1.7825

No, it increases the VE percentage slightly for all the wikis but it doesn't change their ranking. So the question asked by this task is still open!

nshahquinn-wmf subscribed.

Unassigning myself because I don't have any plans to work on this right now. The task is still on the Contributors-Analysis backlog, meaning we still think this is worth doing if we can find the time.

In the 9 years since this task was filed, the visual editor has become a well-accepted tool across Wikimedia. Understanding why editors do or don't accept it is a far lower priority, and I'd go so far as to say this isn't really worth doing at all.