Broken validation statistics

classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|

Broken validation statistics

Gregory Maxwell
Is anyone working on fixing the broken output from
http://en.wikipedia.org/wiki/Special:ValidationStatistics ?

I brought this up on IRC a week-ish ago and there was some speculation
as to the cause but it wasn't clear to me if anyone was working on
fixing it.

_______________________________________________
Wikitech-l mailing list
[hidden email]
https://lists.wikimedia.org/mailman/listinfo/wikitech-l
Reply | Threaded
Open this post in threaded view
|

Re: Broken validation statistics

Thomas Dalton
What's broken about it? It seems very odd to me that the mean is an
order of magnitude greater than the 95th percentile, but otherwise it
all looks fine. I suspect there are a few invalid data points messing
with the mean - perhaps pending changes is being turned off on
articles while there are unreviewed edits and they are counting as
being unreviewed for ages? (Or perhaps only if PC is turned back on
again for that article and they are eventually reviewed days after
being made.)

If that is the problem, then I would suggest disallowing turning off
PC on an article with revisions still pending. Alternatively, turning
off PC could automatically approve any pending changes.

On 27 June 2010 20:19, Gregory Maxwell <[hidden email]> wrote:

> Is anyone working on fixing the broken output from
> http://en.wikipedia.org/wiki/Special:ValidationStatistics ?
>
> I brought this up on IRC a week-ish ago and there was some speculation
> as to the cause but it wasn't clear to me if anyone was working on
> fixing it.
>
> _______________________________________________
> Wikitech-l mailing list
> [hidden email]
> https://lists.wikimedia.org/mailman/listinfo/wikitech-l
>

_______________________________________________
Wikitech-l mailing list
[hidden email]
https://lists.wikimedia.org/mailman/listinfo/wikitech-l
Reply | Threaded
Open this post in threaded view
|

Re: Broken validation statistics

Gregory Maxwell
On Sun, Jun 27, 2010 at 3:24 PM, Thomas Dalton <[hidden email]> wrote:

> What's broken about it? It seems very odd to me that the mean is an
> order of magnitude greater than the 95th percentile, but otherwise it
> all looks fine. I suspect there are a few invalid data points messing
> with the mean - perhaps pending changes is being turned off on
> articles while there are unreviewed edits and they are counting as
> being unreviewed for ages? (Or perhaps only if PC is turned back on
> again for that article and they are eventually reviewed days after
> being made.)
>
> If that is the problem, then I would suggest disallowing turning off
> PC on an article with revisions still pending. Alternatively, turning
> off PC could automatically approve any pending changes.

Reviewing the logs I am unable to find even a single article with a
wait anywhere near that.

Can you find one?


By day two or so it was showing an average of several days during some
hours. Some people speculated that in cases where no edits had been
made since PC was activated it was simply taking the time between the
prior two versions or something like that.

_______________________________________________
Wikitech-l mailing list
[hidden email]
https://lists.wikimedia.org/mailman/listinfo/wikitech-l
Reply | Threaded
Open this post in threaded view
|

Re: Broken validation statistics

Thomas Dalton
On 27 June 2010 20:34, Gregory Maxwell <[hidden email]> wrote:
> Reviewing the logs I am unable to find even a single article with a
> wait anywhere near that.
>
> Can you find one?

I'm not sure which logs to review. The "Advanced Review Log" doesn't
distinguish between edits by new registered users and edits by anons,
and only the latter are included in the statistics (why is that, by
the way?). There also isn't an easy way to see how long it took to
review (you have to calculate it manually for every row).

_______________________________________________
Wikitech-l mailing list
[hidden email]
https://lists.wikimedia.org/mailman/listinfo/wikitech-l
Reply | Threaded
Open this post in threaded view
|

Re: Broken validation statistics

Aaron Schulz
In reply to this post by Gregory Maxwell
I'd ignore the review time stats. They are disabled for PC (but not regular flaggedrevs) on /trunk.

Gregory Maxwell wrote
Is anyone working on fixing the broken output from
http://en.wikipedia.org/wiki/Special:ValidationStatistics ?

I brought this up on IRC a week-ish ago and there was some speculation
as to the cause but it wasn't clear to me if anyone was working on
fixing it.

_______________________________________________
Wikitech-l mailing list
Wikitech-l@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikitech-l