Page MenuHomePhabricator

Have a metric for number of words added (or similar) per event
Closed, DuplicatePublic

Description

(From the user feedback interview.)

This is something the outreach dashboard does already.

Event Timeline

Restricted Application added a subscriber: Aklapper. · View Herald TranscriptMar 16 2018, 11:58 PM

We could use WikiWho! I know the Dashboard uses it for something, probably for this specific statistic. I don't see going through each revision one by one as an option.

This could also could go really, really slow if we have a lot of articles. We might think about where we want to draw the line in making Grant Metrics simple, versus comprehensive.

Or, we could simply get the difference of the page length from the beginning of the event to the end. That's comparatively much easier than using something like WikiWho, but the numbers may include contributions from non-participants. I think this is OK though, maybe even better... For instance it's common for new articles to be edited by patrollers (cleanup, referencing, etc.), so it would seem such contributions could still be attributed to the "success" of the event.

@MusikAnimal How hard is it to get total number of bytes added and total number of bytes removed as a statistic? I imagine not too bad?

MusikAnimal added a comment.EditedMar 20 2018, 9:04 PM

Like the "added (bytes)" column of the ArticleInfo Top Editors, except for all participants? We can probably adapt the XTools code. These numbers can be misleading, though. If I add 100 bytes, but a patroller removes 50 of them, we may not want to record it as +100. That's where WikiWho comes in, it does a much better job of determining content persistence. Their API is pretty comprehensive so I'm confident we could get some accurate numbers, but it will probably be slow. I could talk to Sage to see how they did it on the Dashboard.

Niharika triaged this task as Low priority.May 15 2018, 7:08 PM
Restricted Application added a project: Community-Tech. · View Herald TranscriptJul 20 2018, 10:05 PM

It frequently happens that people who do events need to report metrics to partner organizations or to managers. I.e., to people who are not technical. Is there a measurement that would be more understandable than bytes? I realize that different scripts require varying numbers of bytes to form each letter, so a conversion to letters or words is not simple. But I had to ask.

Is there a measurement that would be more understandable than bytes? I realize that different scripts require varying numbers of bytes to form each letter, so a conversion to letters or words is not simple. But I had to ask.

For content pages, word counting isn't that bad. XTools does this, see results for Domino Park, look under "Prose" on the right-hand side.

The issue however is doing this for thousands of pages. In that case it's considerably cheaper to run SQL queries, and there we can only measure bytes.

It frequently happens that people who do events need to report metrics to partner organizations or to managers. I.e., to people who are not technical. Is there a measurement that would be more understandable than bytes? I realize that different scripts require varying numbers of bytes to form each letter, so a conversion to letters or words is not simple. But I had to ask.

Another metric that we can compute is percentage of content added. Like if the article size is 1200 bytes and an editor added 600 bytes - they added 50% of the article content. There's potential for this to be wrong though, if an editor adds categories and templates more than content, for example. But for newbie editors in editathons, the likelihood seems less.