[13:32:05] <halfak>	 o/
[17:44:31] <halfak>	 o/ ragesoss 
[17:44:38] <ragesoss>	 howdy!
[17:44:46] <halfak>	 Did you happen to see the news re. ORES models for article quality
[17:45:18] <halfak>	 https://lists.wikimedia.org/pipermail/ai/2016-August/000068.html
[17:45:52] <halfak>	 We're changing modeling strategy.  It doesn't look like this will change the scores that much, but you might see a slight jump or dip in the quality-over-time graphs because of the switch.
[17:46:01] <halfak>	 So I think you might want to regenerate old data. 
[17:46:26] <ragesoss>	 halfak: cool, thanks.
[17:46:38] <halfak>	 we haven't deployed to ores.wikimedia.org
[17:46:50] <halfak>	 But the new models are like at ores.wmflabs.org
[17:46:56] <ragesoss>	 it'
[17:47:10] <halfak>	 We're looking to do the ores.wikimedia.org deployment soon -- maybe today. 
[17:47:18] <ragesoss>	 it's good timing, if it deploys soon, since it will be between terms (more or less) for us.
[17:47:47] <halfak>	 Cool.  I think i'll try to get this out today then. 
[17:47:49] <ragesoss>	 the current term is just barely started. not that we rely on the accuracy of the scores in any meaningful way.
[17:48:10] <ragesoss>	 if it's still more or less doing similar things, shouldn't cause us any problems.
[17:52:22] <halfak>	 Cool
[17:52:46] <halfak>	 Woops.  looks like the deploy window already happened, so tomorrow is likely.