A user at en.wp reports that reverted edits remain visible on the impact module. These edits shouldn't be visible in the impact module, since they are not considered as constructive ones. I can't test this live, since it requires to edit the main space.
Description
Description
| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Open | None | T316699 [EPIC] Growth: Positive reinforcement - Iteration 2 | |||
| Declined | None | T281975 Impact module shows reverted edits |
Event Timeline
Comment Actions
I can confirm this by reading the code, it doesn't have a check for reverted edits AFAICS (Impact::queryArticleEdits never mentions tags or ChangeTags::TAG_REVERTED, which is how you can tell if an edit is reverted).
Comment Actions
Yeah, that code was written well before ChangeTags::TAG_REVERTED existed.
There's not really a quick fix for this, that I can see. If the user makes a single edit and we don't show it because it's reverted, they would see
which would be confusing. Also, a revert is not always a value judgment -- it just means the text you added isn't there anymore, which can happen for different reasons.
I'd propose we revisit this when we get started on T248330: [EPIC] Growth: Positive reinforcement
Comment Actions
I discuss about it with Marshall, and we conclude that it needs more thinking that just a fix. So I agree that we can work on it when time for Positive reinforcement comes.
Comment Actions
Oops, we forgot to consider this one. Cc @KStoller-WMF if this is something you'd like to see addressed while we still have the new impact module in mind.
Comment Actions
We definitely should have this soon. If we display 5 edits made, while all of them have been reverted, it is not an insightful feedback. At least, we should go with a fix that removes reverted edits from the counter, and then think of a better improvement.
Comment Actions
I'm in agreement with the general sentiment: reverted edits don't result in positive impact. But I want @RHo to chime in on the UX before we consider this task, so I've added this task to T222310 so we don't forget about it.
My concerns:
- I'm not sure we should remove reverted edits from the edit count: do we do this anywhere else? That's a slightly different definition than "Total edits", more like "Total Live Mainspace Edits".
- If an edit is counted in Total edits, but then not shown in the article list, then won't that seem like a bug? Maybe that's fine, but it might be something we should be more transparent about.
Comment Actions
Note that with the default Echo settings, users don't necessarily learn that they are reverted. Excluding reverts from total edits seems potentially confusing, without adding much value.
Not showing in the article list makes sense conceptually - we are using the impact the user made; they made no impact on that article. Also potentially confusing though. (Maybe a good time to reconsider the decision about not telling new users about reverts?)
Comment Actions
Excluding reverts from total edits seems potentially confusing, without adding much value.
Not showing in the article list makes sense conceptually - we are using the impact the user made; they made no impact on that article. Also potentially confusing though.
This is my general sense as well.
I'm moving this to T316699: [EPIC] Growth: Positive reinforcement - Iteration 2 since I think it's still worth considering, but my general sense is there isn't a great way to do this without introducing complexity and confusion.
(Maybe a good time to reconsider the decision about not telling new users about reverts?)
@Trizek-WMF Do you think this is a decision best left to communities? It seems like this might be nice to include in Community Configuration, however I suppose this is a notification that is outside of the scope of the Growth extension, so perhaps that doesn't make sense?
Comment Actions
It would be possible but complicated.
On the other hand, it seems like only a little work is needed to make it configurable in the traditional way, see T273787: Enable edit revert notif by default on Wikipedia FR and especially T273787#7768815
(Somewhat related: T91909: Revert notification could use a more constructive language)
Comment Actions
Tgr provided the existing context: some communities want to turn it on and it is not possible in an easy way.
It could be on our scope as we could refine this revert notification, which is very violent to get, even for an experienced user.
Comment Actions
I discussed this further with @RHo today and I think we should close this task and instead consider these tasks:
T326354: Community Configuration for Revert notification: improve language and allow communities to configure defaults
T91909: Revert notification could use a more constructive language
Reasoning:
- Removing reverted edits from the Impact Module adds complexity: do we remove reverts from all metrics and data visualizations or just the article list?
- Removing reverts will likely seem odd or buggy for new editors: "why are some of my edited articles showing up, but not others?"
- A edit that removed manually (entirely or partially) by another editor will show up in the impact module, event though it's essentially reverted. It seems odd to treat those cases totally differently.
- The Impact module should remain positive: it's meant to serve as positive reinforcement for new editors and isn't meant to contain exact metrics.
If we are concerned about newcomers making bad edits and yet incorrectly assuming that they are doing well, then communities should be able to decide to notify newcomers about reverts: T326354: Community Configuration for Revert notification: improve language and allow communities to configure defaults or perhaps we should consider adding some sort of "quality" metric to the newcomer homepage? Perhaps showing the percentage of edits that remain (like xtool's "Live edits" percentage would be a more positive way to communicate reverts?