A bullet list of quality thoughts

Some quality points raised by FT2 and various comments by others.

1. Newcomers - there is a pyramid effect (fewer expert editors, many new/inexperienced editors. Observations:

   Guidance and basics for newcomers is significant (affects a significant proportion of users and edits, and may improve the efficiency of established users)

   Piotrus says "Addressing newcomers (with tools such as the article creation wizard) is important as we do need more editors, and I think at this point we have tapped the pool of those who can and want to master the current wiki syntax".

2. Established editors - obseervations:

   Personal improvement and development is self-initiated and self-guided. There are no quality related classes, courses, or masterclasses in anything. It's "pick it up for yourself". Users can edit for years and not be exposed to some ideas. Exposure to better working methods or (good) new ideas used by others, would probably be well received and adopted, and improve quality considerably.

   Also we need to ensure editing stays enjoyable for these users, by looking at ways it can cease to be fun. As in the real world, it's easier to retain people who like you, than attract newcomers, and they have a higher motive to work to a high quality. If it stays enjoyable, they'll mostly continue editing (subject to real-world constraints).

   Piotrus says "Ensuring that editing stays enjoyable, by looking at ways it can cease to be fun" is a KEY ISSUE [emphasis in original].

3. Patrolling - an important quality tool. The same principle could apply wider; if we could generate some kind of feed or list of that. This might be the most effective way to cover missing content, for example automatic or user-originated feeds for "stuff to add", "facts to reference", or "sections and topics to write", in small doses so patrolling is interesting, or encourage patrolling generally.

   Piotrus says bots can ultimately only do a few things better than people.

4. Interwiki efficiency/mobility - If content was quality rated, could editors on an article/topic on one wiki be kept informed if the same article/topic on another wiki is better written or referenced in some areas, so wikis can cherrypick the best of other projects' work, and so that good work or updated information can migrate easily between wikis.

   Woodwalker is in favour of exchange of ideas, but cautions against reducing diversity, and fragmenting the experts and content discussions on a topic. Some projects don't want rating systems - nlwiki recently deleted and forbad the rating of articles. [Note: find out more]

   Yaroslav Blanter would not translate all content, but in some cases (eg an article on a US state copied from enwiki or Paris metro from frwiki) it is an obvious approach. But users should be aware that even so their quality is not always good.

   Piotrus adds information on this from the Polish Wikipedia: "I know pl wiki for example has a way to add mini-templates to interwiki links to indicate articles in other languages have been FAed or GAed. This could and should be improved to reflect the entire spectrum of quality changes, from stub to FA, and possibly a flag to indicate something like 'this article has been recently edited'."

5. Ratchet effect - it's easy for edits to reduce good quality rather than improve it. Ideally we want a kind of ratchet effect, where if something's decent quality then its next step is better quality (eg peer reviewed against older versions). Can we achieve this?

   Woodwalker agrees and notes "flagged revisions" is a way to do this, but more needs to be known before any use since it has potential drawbacks too: "There are some other similar systems too, like patrolled revisions, which is used at wp-nl a couple of years now, or the systems used at external projects like Citizendium. It's not simply a choice between two options, there's a whole spectrum of systems available."

   Piotrus says "perhaps" but it's too close to flagged revisions, which he doesn't endorse.

6. Other areas - problem editing (deliberate or wanton poor editing), social efficiency (disputes, arguments, resolving differences of opinion, and other matters that arise between editors), automation (where we can better use automation or technology to help editors).

   Woodwalker notes flagged revisions would solve problem editing.

   Piotrus says problem editing "seems under contol" but social efficiency is important; the key issues are enforcing civility and a good editing environment.

Looking at quality as a whole:

1. Newcomers: We can expect a pyramid effect, many newcomers, fewer regular and experienced users. So guidance and basics for newcomers may be significant in improving average quality:
   • Affects a large proportion of users
   • Newcomer editing quality is a significant proportion of edits
   • Improving newcomer information may improve the efficiency of actions, of more established users
2. Established editors: For established editors the issues differ:
   • Most endeavors have some kind of way to improve oneself. But on our projects this is self-guided. There are no quality related classes, courses, or masterclasses in anything. It's "pick it up for yourself". Users can edit for years and not be exposed to some ideas. Exposure to better working methods or (good) new ideas used by others, would probably be well received and adopted and improve quality considerably.
   • Ensuring that editing stays enjoyable, by looking at ways it can cease to be fun. As in the real world, it's easier to retain people who like you, than attract newcomers. And they have a higher motive to work to a high quality. If it does stay enjoyable, they'll probably edit, subject to real-world constraints.

3. Patrolling: Patrolling is an important way we improve quality of coverage. New page patrols and recent changes patrols are well known, but the same principle applies to "obvious areas of coverage that aren't covered", if we could generate some kind of feed or list of that. This might be the most effective way to cover missing content.

   Is there a way to create automatic or user-originated feeds for "stuff to add", "facts to reference", or "sections and topics to write", in small enough doses that someone can easily enjoy patrolling such a feed? Is there a way to encourage patrolling generally?

4. Interwiki efficiency: At a point where topic quality rating is available across all projects, could there be a way to automatically let editors on an article/topic on one wiki know that the same article/topic on another wiki is better written or referenced in some areas, and provide them a rough machine translation of it? So that wikis gradually get used to cherrypicking the best of other projects' work, and so that good work or updated information can migrate easily between wikis?
5. Ratchet effect: As quality of an item improves, it gets easier for edits to reduce it rather than improve it. Ideally we want a kind of ratchet effect, where if something's decent quality then its next step is better quality. A simple means (using flagged revisions) would be that for a revision to be flagged, it has to be peer reviewed in comparison to its previous flagged version. This is easy enough ("Good Article" reviews are fairly quick) but would create a powerful effect on quality.
6. Other: There are three areas separate to the above:
   • Problem editing: How we handle deliberate or wanton poor editing (vandalism, pov pushing, etc) that degrades the quality of information. Not all users want "quality" in the sense we hope for.
   • Social efficiency: How to keep the communities efficient (or make them more efficient). Social efficiency would cover things like disputes, arguments, resolving differences of opinion, and other matters that arise between editors and cause the projects to lose out.
   • Automation: Where we can better use automation or technology to help editors.

@nr 4: Though I am in favour of more communication between projects and exchange of ideas and data where guidelines differ, I am not sure about cherry-picking for translation. Simply having copies of the big projects (especially wp-en) everywhere reduces diversity. However I have to admit that I often translate content myself. Another point of concern is that the discussion about the content will become fragmented once a translation is made. There will then potentially be two groups of expert users discussing separately at different projects. For the quality it would be better to have groups of experts discuss matters at one place, for example at the metawiki. Another problem is that not all projects want a quality rating system. The Dutch Wikipedia recently voted to delete the project page on quality rating and forbid the rating of articles, even on TPs.

@nr 5: I totally agree. FR is potentially a way to increase the quality of content, and I've always been (cautiously) in favour of its implementation. Yet its effect on quality should be known better before we can draw a conclusion, since there are some potentially negative effects (what is its effect on the influx of new users for example?). Flagged revisions can be implemented in many different ways. There are some other similar systems too, like patrolled revisions, which is used at wp-nl a couple of years now, or the systems used at external projects like Citizendium. It's not simply a choice between two options, there's a whole spectrum of systems available.

@nr 6a: If FR is used, edits of problematic users can simply not be flagged as trusted.

In my opinion, translating all content is stupid. On the other hand, there is some content which is clearly project-specific. For instance, I recently needed to create an article on the history of Georgia (US state). The obvious route I have chosen was to translate an English article first, and then to add the content I find in addition. I think it is kind of obvious that articles on en.wp on the history of US are of good quality. As another example, articles on Paris metro stations on fr.wp are not fantastic, but they are still a good basis to improve on after the translation; on the other hand, en.wp articles on the same topics are often substandard and not really recommended for translation in other projects.

@1: Addressing newcomers (with tools such at the article creation wizard) is important as we do need more editors, and I think at this point we have tapped the pool of those who can and want to master the current wiki syntax.

@2: Established editors: Ensuring that editing stays enjoyable, by looking at ways it can cease to be fun.. KEY POINT, KEY POINT. Our pool of editors is in fact shrinking ([1]). Addressing this is a major issue, because if we loose too many editors, quality improvement efforts will inevitably suffer (I am familiar with complains "we need more editors" from various places - copyeditor people, copyvio people, thematic wikiproject people... In fact, here's a simple recommendation to improve quality: get more dedicated editors (reach out to those who left?).

@3: Automatization is nice, and all efforts towards that end should be supported. But in the end, bots can do only a few things better than humans.

@4: An interesting point. I know pl wiki for example has a way to add mini-templates to interwiki links to indicate articles in other languages have been FAed or GAed. This could and should be improved to reflect the entire spectrum of quality changes, from stub to FA, and possibly a flag to indicate something like: "this article has been recently edited".

@5: Perhaps, but this smacks to close to flagged revisions and other stuff I don't fully endorse.

@6: Problem editing: seems to be under control, no need to change it substantially (why fix what's not broken)? Social efficiency: another important point. I have dedicated quite a lot of thought to this issue and perhaps I will write something here in more detail later. For now, I'd say the key issues are related to enforcing civility and creating good editing atmosphere. Automation: covered above.

--Piotrus 04:07, 24 November 2009 (UTC)