Topic on Talk:Talk pages project/Replying/Flow

T250295: Should we display a note about automatically resolving an edit conflict?

I forget if I've brought this up somewhere before, but edit conflicts are an issue with this tool that needs to be resolved before it goes out of beta. When you've been writing out a comment and someone else edits the article, before when you tried to click publish you'd get the edit conflict screen. Sometimes that'd be annoying, but other times it'd allow you to adjust what you write to take in their comment, or at least to add the edit conflict template to your comment to indicate that you wrote it without knowledge of theirs. However, the new reply tool just powers through and posts the comment, without any tag.

This could result in some awkward social situations (let me know if you're having trouble imagining them and I'd be happy to concoct examples). I'd recommend that the tool either learn to apply the EC template or display a heads up confirmation that lets you read the new comment before deciding to publish. Cheers,

Ha! Now that you bring this up, I realise it's been happening to me. Several times my edit has gone through, messing up in the process an edit that someone else had made minutes before. I couldn't understand why I had been (as I thought) editing a older version of the page.

@Sdkb: we appreciate you bringing attention to the experiences you've had with the Reply Tool and the way that it currently goes about automatically resolving edit conflicts.

A couple of resulting questions for you below and then one more that I'm going to post a separate comment...

This could result in some awkward social situations (let me know if you're having trouble imagining them and I'd be happy to concoct examples)...

Examples would be great...can you please share them?

...needs to be resolved before it goes out of beta.

Can you say more here? What is leading you to perceive making adjustments to the way the Reply Tool automatically resolves edit conflicts as rising to the level of blocking the deployment of the Reply Tool as a default-on feature? Note: I appreciate it might be difficult to answer this question in a way that feels "complete," but an attempt will help us better understand the impact it's having on you.

@MichaelMaggs it's helpful to hear the way the Reply Tool automatically resolves edit conflicts is causing you trouble...we value you sharing this feedback with us.

A resulting question for you and @Sdkb: What – if anything – about the "need statement" below do you think needs changing?

Need Statement: "As someone who is drafting a comment using the Reply Tool, I want to know when someone publishes a comment in the same section I am drafting a comment within and what they have said, so that I can decide whether or not to make changes to the comment I'm drafting prior to publishing it."

Here's a specific example. When replying to a recent comment in an active thread it's embarrasing, to say the least, to find on posting it that the comment you are replying to has while you were writing been significantly altered by its author (as of course they are perfectly entitled to do before a reply has been posted).

I think that the Need Statement is excellent.

Yep, great example. Often, that'll happen when someone says something uncivil, and as another editor is writing a comment chiding them for it, they cool off enough to think better of what they wrote and change it.

Examples would be where someone makes a point, and then I make the exact same point, which comes across as a violation of w:WP:READ. Or in an AfD, I !vote delete on the basis of no reliable sources provided, and then the editor who just provided such sources (that I didn't see) asks a little annoyedly why theirs aren't good enough.

On importance, a more precise way of putting this might be "this should be resolved before work on this feature stops". It's certainly an issue but not something that needs to impede wider deployment if you're getting close to that. The biggest issue for me is still that it adds a signature sometimes when I don't want it, e.g. for Barnstars and other templates that sign for you within a box. Best,

@MichaelMaggs + @Sdkb: the examples you shared^{[1] [2] [3]} are helping us to clearly visualize/understand the issue the current experience is perpetuating...thank you for sharing them.

With these example in mind, we've started "sketching" what the user experience could look like to fulfill the "Need Statement" we talked about above.

You can review the proposed user experience by reading T250295's "Requirements" section. If reading it brings any thoughts/concerns/questions to mind, we'd value you sharing them with us here, or on the ticket.

Something else that might be useful to know: How often does this happen to you? Do you encounter an embarrassing situation once a day, once a week, once a month, once a year?

The team can pull stats on how often an edit conflict gets resolved automatically, but what they need editors for is to tell them how often that automatic action creates enough of a problem that you self-revert or substantially change your reply.

I expect you can extract stats on how often I've used the tool, but overall I've had two situations in the last two months that I'd say were difficult/embarrassing. In both cases I spent quite some time trying to work out what had gone wrong, and I think in one I apologised to another editor on the basis that I must accidentally have been editing an old version of the page.

I'd be wary about using stats in this way to inform how much priority you give to this. Nobody knows, for example, how many instances they've had when they didn't notice anything. But an unnoticed error resulting in what appears to be an irrelevant, stupid or unnecessarily aggressive reply will remain forever in that editor's history, and when it's dug out again later for the purpose of criticism or even objection to an RFA, who can then tell that it really wasn't the editor's fault?

Most important in avoiding major issues, of course, is to make sure the editor is aware of any changes to the specific comment that they are replying to. Changes to other comments in the same section, while useful to know about, are generally going to be less critical.

Hmm, that's a little hard to say, and it depends entirely on how active an editor is. Personally, just earlier today, I encountered a situation where I mildly feared there would be an unhelpful edit conflict when I replied to a new post at the Teahouse and worried another user might have beat me to it, creating an unhelpful duplication of information. Luckily it didn't happen for me this time, but for another user who replied a minute after me, it might've.

I think the risk depends on how much time you spend on active discussion pages. I've probably used the Reply tool about 400 times during the last month. I think I've regretted the automatic resolution of edit conflicts twice during that time period. So perhaps the risk for busier pages is about 1%?