Topic on Talk:Growth/2023

Feedback moved here from mastodon, per request of @srishakatux@wikimedia.social. Referring to en.wiki where I'm a decades-old editor.

(a) The page is random for long-time users, is it only looking at recent edits for efficiency? Where is this documented?

(b) There is no talk page and there should be, even if it's just a cross-wiki redirect to this project.

(c) The "Get shortened URL" link is misleading, since it takes the resulting link takes the dereferencer to their impact, not the minter's impact page.

(d) The page appears to be set up to compare 60 day edits with 60 day views as the two main visuals, but (i) these are over very different domains (all namespaces vs article namespace) and not really comparable, (ii) the time-series graphs slightly offset making direct comparison impossible (iii) far better use of space with be bars and lines on a single graph with a consistent timeline and (iv) you end up with situations like my current Impact, where it looks like I don't even view the pages I edit (see screenshots).

(e) It would be great to see my impact against average impact of those in my editor-cohort (i.e. other accounts of a similar age / longevity).

Screenshot of my impact page on enwiki

~~~~

Two more things:

(f) the last 60 days have included a NPP Backlog drive (see https://en.wikipedia.org/wiki/Wikipedia:New_pages_patrol/Backlog_drives/October_2023) where I made the top 10 and won two barnstars. Most of that is not in my impact, since it's mainly log actions.

(g) there are many phabricator threads about inconsistencies in counting edits, because there are a large number of intricacies in this area. Any new tool in this area needs to be very clear and open about how edit count is being measured and that other methods exist.

Hello, and thank you for your message

Overall, the project is documented at Growth/Positive reinforcement. The impact module is part of positive reinforcement tools, and, overall, it is part of the entire Growth features. Please don't forget that these features are for newcomers; we provide certain features for experienced users (such as mentorship) but the impact module is not one of them, even if it gains on popularity.

a) What do you mean by random? Only edits made at the main space (Articles) are taken into account. I checked your impact data on a few random days, and the numbers make sense. Again, it is not a feature for experienced users: we decided to highlight what matters: edit to the contents.

b) can you be more precise? Which page are you referring to? If it is Special:Impact, special pages don't have talk pages.

c) I tested the shortener on your impact module (Special:Impact/Stuartyeates) and https://w.wiki/873v sends me to your impact. https://w.wiki/873x goes to mine. However, if you don't add the username to the URL, you have the case you described. I documented it on T350932.

d) i) All impact is shown for articles.

d) ii) You'd like to have a comparison between the edits graph (in bars) and the views graph (the line)?

d) iii) We asked ourselves the question, and we were afraid of overcrowding the graphs with two sets of data at the same place. The distinction between the two graphs wasn't highlighted as an issue during the testing we performed with newcomers.

d) iv) I'm not sure to get it: the graph highlights pages views for everyone, not just the pages you visited.

e) Good idea, I documente it as T350936.

f) As I said, we have only take articles into account. Logged actions would be for a next iteration, for ex-newcomers looking for challenges. You mention barnstars: we have a few ideas around providing rewards to newcomers (and beyond), based on the number of edits.

g) The edit count is based on what we have in the database, i.e. what you have in Special:Contributions. Regarding pages views, we rely on stats.wikimedia.org.

Hope this helps, and thank you again for your feedback! :)

(a) By random I meat two things (i) that it's insane that I have edits on tens of thousands of pages but those pages were only viewed 247 times. If it's only counting views on pages I've editing in the last 60 days, it should say so. and (ii) I've been involved in AfC in the last 60 days, not all of my edits to pages that are currently in main space were in main space when I edited them.

(b) We're spending considerable effort training people to go to the talk page if they have an issue with a wikipedia page. The fact that this wikipedia page doesn't have a talk page undermines that.

(c) Where on https://en.wikipedia.org/wiki/Special:Impact does it give me the hint that I can see someone else's Impact.

(d) (ii) I'd like the red lines in my annotated screenshot to line up, so that features of the two graphs can be directly compared. BTW: is there a reporting delay on the views? if so this needs to be made obvious. [couldn't upload screenshot because that's not working for me right now.]

(d) (iv) I visited the (now) articles when they were in draft: to make the edits but maybe those views are not counted, because only edits/logs and not views are moved from draft: to mainspace?

If may seem that I'm obsessing about AfC, but there are a lot of newbies at AfC...

Hi @Stuartyeates, First of all, thank you for all the feedback! I think several of the issues you pointed out could help us improve the feature further! We've had this feature released on several pilot wikis for over six months, and I think you have provided more feedback here than we've received in those six months of testing on pilot wikis.

(a) We tried to keep the layout and information simple for new account holders, but I can see how we might want to consider further design improvements or add more clarity on how the data is populated. The Your recent activity (last 60 days) header is meant to apply to all the data underneath, including on the "Views on articles you've edited". I take it that isn't very clear though?

(b) That's a great point, I've also been thinking that there is more that we should do to ensure new editors understand and feel comfortable interacting with Talk pages, and furthermore know how and where to leave feedback. There should be an easier path to getting help or leaving feedback from where newcomers actually encounter the Impact module: the newcomer Homepage (Special:Homepage). This task has been sitting unfinished for a long time: T229869, but you can see in the comments I was just thinking about this task earlier this week. Do you think completing that task would help?

(c) That's admittedly a somewhat hidden feature currently. Are you interested in seeing the impact module of other editors? We are currently working on a task to make it easier for people to share their impact module, or potentially even embed their impact module on their user page T341911.

(d) I agree that it would be nice to eventually make these graphs more complex and even interactive. But we wanted to keep it fairly simple to start to get the the new Impact module released and get feedback, and A/B test the changes, before investing considerable time in these data visualizations. I believe there is a ~24 hour reporting delay on view counts after an edit. The data is populated in the same was as https://pageviews.wmcloud.org/pageviews/, and has the same limitation: The Wikimedia pageviews API generally takes a full 24 hours to populate, sometimes longer. In some situations you may see data missing for yesterday's date as well, which will be left blank rather than showing a count of zero views.

As Trizek (WMF) mentions, the Impact module is mainly intended for new editors, and as we don't often get extensive feedback from new account holders, so it's great to hear from you what is unclear or frustrating. And although the Growth team is mainly focused on features to help onboard new editors, I think it would be great to see the Homepage (and the Impact Module) become more usable and customizable for experienced editors.

We'll work on fitting in some of the Phab tasks Trizek (WMF) added and I mentioned, but please feel free to pass along further feedback as well! Perhaps when we add a Feedback link from the newcomer Homepage, we should also add a general Help link in which we can link to a help page with a full breakdown on how the Impact module data is populated. Do you think that's a step in the right direction?

"The Your recent activity (last 60 days) header is meant to apply to all the data underneath, including on the "Views on articles you've edited". " Does it also apply to the "Longest streak above" ? I've not written a quarry query to check, but it seems unlikely that my longest streak out of >15 years is right now.

Something else I've just realised is that "Thanks received" (which is listed) is a symmetric with "Thanks given" (which is not).

As you both pointed out, I'm a long way from your target audience.

I've just had another look, and right now https://en.wikipedia.org/wiki/Special:Impact/Stuartyeates doesn't list any

impact right now: the second graph which appeared earlier has vanished and all of the articles have clock faces next time them.

(b) Help links on talk pages is a pretty standard feature (see Special:Watchlist, Special:Contributions etc.) Adding a such link to Special:Impact could probably help – at least on desktop, because they don’t appear on mobile (T75299 – the task is exactly nine years old today!).

@Stuartyeates, you said:

If may seem that I'm obsessing about AfC, but there are a lot of newbies at AfC...

Growth is actually considering to work on article creation for new editors. We know that more than 20% of newcomers (on average, at all wikis) create an account to create an article. This project is still under definition.

Something else I've just realised is that "Thanks received" (which is listed) is a symmetric with "Thanks given" (which is not).

You suggest to have a module where we would count thanks given? Won't we have a risk of users playing with this count, by thanking every edit they can to have an higher number?

Re: the graph being broken - I can't reproduce it as I see numbers now. Sorry.

---

@Tacsipacsi, as the feature is not designed for all users but mostly for newcomers, would it be relevant to create this link? I don't think we want experienced users to be frustrated by an experience that is not designed for them. What we would do, if our mandate was going in that direction is 1/ ask experienced users what they'd like to see in the graph, 2/ work on the feature 3/ release it with the link at the right place.

You said:

You suggest to have a module where we would count thanks given? Won't we have a risk of users playing with this count, by thanking every edit they can to have an higher number?

(a) You can't thank your self for your own edits (b) I'm not sure I see excessive thanking of editors impeding the encyclopedia (c) If you create a sockpuppet to do it, this is banned on en.wiki by clause 1 of en:WP:Sockpuppetry ''Creating an illusion of support'' (d) thanking random other editors doesn't seem to be harmful to me. You said:

Re: the graph being broken - I can't reproduce it as I see numbers now. Sorry.

I'm pretty sure it's because all my recent edits at the time were AfC pages and so either in Draft: or mainspace articles too new for their page views to have appeared in the stats engine.

Exactly, the page view stats are including only mainspace article edits, and the clock icon displays if there isn't yet view count data (the Wikimedia pageviews API generally takes a full 24 hours to populate, sometimes longer). If you click on the clock icon you'll see: "Pageviews have not yet been calculated. Check back tomorrow!"

That being said, if that's unclear to you, then I imagine it might be unclear to many new editors, so thanks for the questions / feedback.

@Stuartyeates the Your recent activity (last 60 days) header does not apply to the "Longest streak" data listed above that header. However, there is the "Longest streak" is limited to your last 1,000 edits. We actually just released an update that should add clarity to the info pop-up next to the Longest streak: T338174.

We needed to place some sort of reasonable cap to ensure the query wasn't too slow or costly. Since the feature is mainly intended for newer editors, 1,000 edits seemed reasonable. Would you find that stat interesting if we had a higher cap or it was an "all time" longest streak?

@Trizek (WMF):

@Tacsipacsi, as the feature is not designed for all users but mostly for newcomers, would it be relevant to create this link? I don't think we want experienced users to be frustrated by an experience that is not designed for them.

I don’t see why it wouldn’t be relevant; it’s just the opposite: help links are mostly for users new to a given feature. I’m pretty sure the help links on the watchlist or contributions pages are predominantly used by newcomers – experienced editors don’t need help on what is a user contributions page or how to use their watchlists.

@Stuartyeates:

(a) You can't thank your self for your own edits (b) I'm not sure I see excessive thanking of editors impeding the encyclopedia (c) If you create a sockpuppet to do it, this is banned on en.wiki by clause 1 of en:WP:Sockpuppetry ''Creating an illusion of support'' (d) thanking random other editors doesn't seem to be harmful to me.

Random thanks have two problems:

• Checking who and trying to understand why thanked an edit wastes the thanked user’s time. It’s probably not more than half a minute, but if people start to get dozens of thanks every day, those half minutes accumulate.
• And, more importantly, a thank communicates that the edit is welcome. If a newcomer randomly thanks another newcomer’s edit, the thanked user may think that what they did is good, even if it’s actually a test edit, an edit against policies, or even vandalism.

By the way, I don’t think sockpuppetry would be a problem at all: if you start thanking users with a sockpuppet, you increase the sockpuppet’s counter instead of the main account’s one.

@Tacsipacsi, I see your point regarding the link, but I have a different perspective. An experienced user will see an Help link and think "it is not for me, I know how to edit". They won't think the same with a link to Impact, as they will expect to have useful data showing their involvement on the multiple facets of the wiki.

@Trizek (WMF), yes, of course it will be useful also for editors who are generally experienced but not in this field. Is it a problem? Do you want to build a tool that’s only useful for newcomers, that you actively make useless for experienced editors?

@Tacsipacsi @Trizek (WMF) If experienced editors don't come to understand something, the burden of supporting the new users to understand it will forever remain with the WMF, which doesn't seem such as win.

We designed the Growth features with newcomers and for newcomers. In the case of Special:Impact, the feedback you shared proved that it is not adapted to experienced users. Should we make this module more visible if not adapted? I don't think we should.

Another point to consider besides the lack of relevance of the current design for experienced users is trust and reputation. Making Special:Impact more visible to all users as it is will expose us to negative critiques. You came here with constructive ideas, and we really appreciate that. But we also all know how some people can react to something that is not suiting their needs. These are not a pleasant nor constructive experiences, plus it takes a lot of time to fix things afterwards. This is why we prefer to avoid these conflits by taking the time to properly design a feature that suits the majority of cases for all profiles.

It means that this can change in the future! After all, the newcomers who use Growth features now are the next generation of experienced users: we will have to provide more extended tasks that fit their needs. Your feedback confirms that some experienced users expect this to happen as well. :)

At the moment improving Special:Impact for more experienced users is not in our plans. But we can consider these tasks for the next annual plan. Its definition will start in April. It is quite soon!

Screenshot of help dialog in special:impact

This help text is wrong, the count is correct, but they're not all from my most recent 1000 edits.

@Stuartyeates Hmmm, thanks for mentioning this. Indeed it doesn't seem like the Thanks count is capped at a person's most recent 1,000 edits. Looking at the related code it sounds like it "might exclude thanks received a long time ago."

@Tgr (WMF) - sounds like the Growth team should update the info presented to users... any chance you can share more about how that metric is calculated/capped, so I can improve that explanation?

We are using ThanksQueryHelper::getThanksReceivedCount(). It's capped at 1000 thanks, not thanks for 1000 edits. (It's not limited to thanks received for edits, either.)

special:impact when not logged in

This should redirect to the login page or something.

T351920 documents it.