All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=791101

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=f640fdc0191765c262449f1bd79491c788c9ceec6e22938357fd062de0a88a0b


Bot(s) for this bug's original alert(s):

webrtc-android-tests-nexus5-kitkat

No, it really looks like a fluke. It goes back at some pretty unrelated CL. Also, I am pretty sure that the reduced message isn't happening on the android bots.

As per request by OP, reassigning to Sprang@

I've filed a couple of other strange bugs on the nexus5 device. Please assess severity and reassign to phoglund@. If the bot just wastes our time with false alerts, we might want to stop testing on it.

I've also had experienced several completely random regressions and improvements on that bot when I was a sheriff. Maybe you are right and we should retire that bot.

+phoglund, what do you think?w Maybe replacing a device would be a nice idea?

I took a quick look in the other bug, and in one of the graphs I could see five alerts for the past 5 months. 3 of them were legit (though not acted upon). I think that's an acceptable S/N ratio.

Do you have data that a particular device is especially noisy? In that case we can look at replacing it. These are consumer-grade phones, i.e. junk, so they do degrade over time :)