The good news is that the graph wasn't showing the metric I thought it was showing, and therefore we don't have such sudden growth of unreviewed changesets.
The not so good news is that now it's clear that we have a steady growth in our backlog of unreviewed changesets: 538 in August 2013, 1079 in August 2014. 100% increase in one year. These numbers might still change a bit based on the definition of "waiting for review", but I believe the shape of the lines, the relative growth of the Gerrit queue, will not change much.
http://korma.wmflabs.org/browser/gerrit_review_queue.html
On Saturday, September 13, 2014, Jon Robson jdlrobson@gmail.com wrote:
Do we have any similar data which shows how neglected certain users are?
We might look into that, but already now we have (in the URL above) a list of repositories sorted by median age of patchsets waiting to be reviewed. It is fair to assume that maintainers are not neglecting someone systematically, but it is more likely to find repositories that are less maintained, untidy, or with clearly more contributors than reviewers.
Unsurprisingly, quite often there is a relation between repositories at the top of the list and lack of maintainers declared at https://www.mediawiki.org/wiki/Developers/Maintainers