@Samwalton9-WMF

Here is the data (Except "All Edits", everything is within main namespace only)

In T352026#9445377, @KCVelaga_WMF wrote:

@Samwalton9-WMF

Here is the data (Except "All Edits", everything is within main namespace only)

Thanks! Could you explain the differences between 'Page creations' 'redirect creations' and 'non redirect creations'?

@Samwalton9-WMF Page creations is all page creations (within main namespace), which is technically equal to redirects + non-redirects. However, non-redirects count include both articles and disambiguation pages.

In the table above, the numbers are the averages of their own individual columns, but not sum of the averages. For example, "Page Creations" is calculated as "average number of page creations / day", but not as sum of "average number of redirect creations / day" + "average number of non-direct creations / day". So page creations may not be the exact sum of redirect and non-redirect columns, but should be close. I have verified, redirects + non-redirects = page creations , holds true for yearly totals.

In T352026#9449745, @KCVelaga_WMF wrote:

@Samwalton9-WMF Page creations is all page creations (within main namespace), which is technically equal to redirects + non-redirects. However, non-redirects count include both articles and disambiguation pages.

In the table above, the numbers are the averages of their own individual columns, but not sum of the averages. For example, "Page Creations" is calculated as "average number of page creations / day", but not as sum of "average number of redirect creations / day" + "average number of non-direct creations / day". So page creations may not be the exact sum of redirect and non-redirect columns, but should be close. I have verified, redirects + non-redirects = page creations , holds true for yearly totals.

Oh I see, this makes sense. Can you help me understand why the sum of Page Creations, Bot Edits, Sysop Edits, and Self Reverts might sum to more than the total of Main Namespace Edits? Is it because they're independently counted and might be duplicated?

@Samwalton9-WMF Sorry missed this. I only see that sum of Page Creations, Bot Edits, Sysop Edits, and Self Reverts is greater than Main Namespace Edits for arwiki, unless I missing something. This can happen as some of these columns are non-exclusive of each other:

• Sysop edits and Bot edits also include page creations.
• Similarly with reverts, the reverts column counts both self and non-self reverts. Example, for enwiki, of 7160 reverts, 934 are self-reverts, and some of these self-reverts could have been by admins as well.

Let me know if it would be helpful to isolate any of these for exclusive counts (for example, sysop edits excluding reverts and page creations).

Gotcha - thanks for explaining! From this I found that of all mainspace edits on the projects we investigated here, on average:

• ~4% are page creations
• ~23% are bot edits (there are notable outliers here which bump up the % - ar.wiki at 65% and a few around 35%)
• ~10% are sysop edits (ar.wiki is again a notable outlier at 45%)
• ~0.7% are self-reverts

These percentages can't be added together directly since, per the last comment, they're not exclusive of each other, but this gives a rough scale of how many edits we're skipping entirely given our current configuration and model limits.

@Samwalton9-WMF yes, that's a perfect summary.