On 9/3/15, Brian Wolff <bawolff(a)gmail.com> wrote:
This sounds a lot like PageTriage, which at best was a mixed success.
I hope the team is able to extract lessons from that extension and
apply them to whatever they intend to work on.
"at best was a mixed success" speaking as someone who has used it
extensively, that is not the case. If you mean there are lessons to be
learned around ensuring we have generalisable solutions to specific
workflows, rather than project-specific solutions to project-specific
workflows, I share your hope.
The fact that its almost impossible to use outside enwiki, would be
one of the main factors of why I would call its success mixed (One of
the goals even was "allow expansion and modification of the system to
support different backend systems and logic screens."). I do hope we
do not repeat that limitation with new systems.
The other reason I would call it mixed, is that the majority of
patrolling even on enwiki, still takes place through the other
interface (I believe. I'm not very familiar with patrolling on enwiki,
so I may be misunderstanding something here)
Stats for this August:
MariaDB [enwiki_p]> select log_type, log_action, count(*) from logging
where log_type in ( 'pagetriage-curation', 'pagetriage-deletion',
'patrol' ) and log_timestamp > '20150800000000' and log_timestamp
<
'20150899000000' group by log_type, log_action order by count(*);
+---------------------+------------+----------+
| log_type | log_action | count(*) |
+---------------------+------------+----------+
| patrol | NULL | 4 |
| pagetriage-curation | unreviewed | 134 |
| pagetriage-deletion | delete | 1132 |
| pagetriage-curation | delete | 1132 |
| pagetriage-curation | tag | 2170 |
| pagetriage-curation | reviewed | 10913 |
| patrol | patrol | 184269 |
+---------------------+------------+----------+
7 rows in set (3.75 sec)
I'm not sure what level of popularity was targeted, but 6% market
share (4.6% if you count by unique users instead of total number of
patrol actions) doesn't seem like a run away success.
Its certainly not a failure. But I'm not sure I would call it a full
success either due to its lack of flexibility and luke-warm adoption
numbers.
--
bawolff
I just realized I was counting autopatrol entries in with the other
patrol log entries, which is rather unfair to PageTriage. If you
discount autopatrol, then you get about 34% of patrol actions being
done via page triage (18% if you count by number of unique users
instead of total patrol actions). Well perhaps still not taking over
the patrolling world, that's a much more respectable percentage of the
market share.
I still hope lessons, both positive and negative, can be extracted
from PageTriage, when embarking on similar projects, and in
particular, I hope that future solutions can be used by all wikis
where applicable, not just enwiki.
--
bawolff
p.s. For reference, in case anyone wants to dispute my numbers.
MariaDB [enwiki_p]> select log_type, log_action, count(distinct
log_user) from logging where log_type in ( 'pagetriage-curation',
'pagetriage-deletion', 'patrol' ) and log_timestamp >
'20150800000000'
and log_timestamp < '20150899000000' and log_params not like
'%auto";i:1;%' group by log_type, log_action;
+---------------------+------------+--------------------------+
| log_type | log_action | count(distinct log_user) |
+---------------------+------------+--------------------------+
| pagetriage-curation | delete | 98 |
| pagetriage-curation | reviewed | 230 |
| pagetriage-curation | tag | 123 |
| pagetriage-curation | unreviewed | 53 |
| pagetriage-deletion | delete | 98 |
| patrol | patrol | 1252 |
+---------------------+------------+--------------------------+
6 rows in set (1.05 sec)
MariaDB [enwiki_p]> select log_type, log_action, count(*) from logging
where log_type in ( 'pagetriage-curation', 'pagetriage-deletion',
'patrol' ) and log_timestamp > '20150800000000' and log_timestamp <
'20150899000000' and log_params not like '%auto";i:1;%' group by
log_type, log_action;
+---------------------+------------+----------+
| log_type | log_action | count(*) |
+---------------------+------------+----------+
| pagetriage-curation | delete | 1132 |
| pagetriage-curation | reviewed | 10913 |
| pagetriage-curation | tag | 2170 |
| pagetriage-curation | unreviewed | 134 |
| pagetriage-deletion | delete | 1132 |
| patrol | patrol | 31802 |
+---------------------+------------+----------+
6 rows in set (5.78 sec)