I don't think it was a symptom of routing synchronization specifically, but I'd be curious to know if it was a case of unexpected and undesired synchronization.

I don't think it was a symptom of routing synchronization specifically, but I'd be curious to know if it was a case of unexpected and undesired synchronization. (E.G. An independent and random cluster of blocks suddenly updated; the network was saturated; it pulled in more updates; ...) And yes, the paper talked about randomization. It also pointed out the magnitude of randomization required was larger than expected.