What Europeans do at Night | Hacker News Reader

What Europeans do at Night

asert.arbornetworks.com

54 pointsthere13 comments

Interesting, but flawed to the point that the opposite conclusion is true. The median time zone for Europe is CET, which is GMT+1, not GMT as is put in the graphs, and the median American one is probably something like GMT-6.3 or so. If you move the graphs by 2.3 hours, they look pretty similar, except the European one has a steeper drop off. This can be explained by the fact that the majority of the U.S. population lives in either GMT-5 or GMT-8, but less in between. This tends to smear out the dropoff of people going to sleep in the U.S. On the other hand, the majority of European internet users are all in the CET time zone, so there is a sharper drop as they go to sleep.

Exactly. Europe is quite narrow continuous set timezones whereas in the American graph shows the east coast and west coast nicely.

I wouldn't be so quick to assume that they don't take this into account:

"Even after we account for the multiple time zones in both Europe (3 if we exclude Russia) and the US (4 if we exclude Halifax and Alaska), European traffic really is different."

This statement suggests that they've adjusted for the different time zones. Do you have evidence that they didn't?

1) The used a wrong offset, as the parent poster indicated.

2) They assumed a simple offset could adjust for the differences.

A simple offset isn't enough:

The (by far) majority of Europeans are in a single timezone.

The majority of internet-using Americans europeans are in two timezones with a 3 hour time difference, and very few users in the timezones inbetween.

This makes a significant difference. To compare to Europe, they should split US into 2 "nations": east-coast and west-coast.

Their method was wrong. Not how they executed it.

"The used a wrong offset, as the parent poster indicated. They assumed a simple offset could adjust for the differences."

The parent poster didn't have any evidence for that statement; he just assumed it, because of the way that the data was graphed. Point out the place in the article where they say that they used a single offset.

From the section I quoted, they seem to be aware of the problem. They could have easily adjusted each data set individually for time zone, then taken the average of the sets to make the graphs shown. They do say that they're graphing average European and American traffic.

Good catch. I was going to say that's because there's stuff to do after 9:30pm in European towns, but that's unfair to the bigger US cities.

The article mentions a peak during dinner time, but many Europeans eat dinner later than Americans. In fact, in some areas, the answer to "what do Europeans do at night" (7-10PM) might be: dinner.

I always assumed that the dip in the US was people driving home. Maybe a combination of the two.

I'm in the UK, we have dinner at 6:30pm but then we have a pre-schooler. Sometimes it's closer to 8, 8:30pm but rarely that late more than once a week. In countries with a siesta I'd expect this to push back at least an hour. In short, I concur.

He completely ignores commute times. I think that's likely the cause of the dip during "dinner" time.

Before I was working, mine was very much like the European one (and I am in Europe, so...). So was the internet usage of a bunch of friends. Sleep in the morning, get on the net in the late afternoon and then go to bed somewhere around 6am - some people went earlier, though, so the 3am dropoff seems about right.

Now I get on the internet at 9am, stay on till about 6pm. Sometimes I use the internet after that, but often I don't.

If the graph data is correct, there is a very interesting intersection between European and American usage about 4 PM (First graph). What I read from this is that we align our usage with each other. Europeans stay up late to interact with Americans, which also explains the early American peak.

I guess people in Europe go out at night while in American Suburbia there is no alternative to go online.

Reply on news.ycombinator.com