Talking to Europeans? — And Asians?

- I agree CEST would have been more representative of European traffic. GMT was just simpler since our native data collection infrastructure uses UTC and I could skip a conversion step.

- Having fond memories of late night tapas in Madrid and more than a few nights in a Cork pub (as well as a varied array of other cities and countries in between), I understand the challenge in viewing Europe as single, cultural monolith — it clearly is not.

- Unfortunately, our data granularity is mostly at the level of continent — without a significant amount of work, we cannot easily tease out traffic patterns by individual country.

- The traffic is the sum of outbound and inbound. We explicitly excluded tier1 transit networks to avoid bias from Europe / Asia traffic going to US based content / servers.

Seriously, all the above arguments are cancelled out. The dinner time drop of North American traffic suggests that the influence of other regions routed through there is small. The cultural variation of different dinner times in Europe would be enough not to expect a European dip, and there is none. Thus the observation remains: Europe has shorter Internet day. None of the above gives an explanation for that (except for the suggestion that Europeans use home internet less — which would explain it, but I think it’s bogus), and I do not have a clue either.

Is it possible at all to get country level data? Would that even make technological sense?

In any case, brilliant question!