Median internet speeds is a terrible metric when comparing infrastructure because it effectively only measures the cheapest to deploy urban populations internet speeds which is basically meaningless now days. 50 Mbps or 500 Mbps all are basically running on the same back haul which is why things fell apart during the pandemic when people tried to actually use their hypothetical home bandwidth.

Meanwhile millions of Americans are stuck with 1.5 Mbps DSL or really expensive Satellite internet.

I don't disagree with your point! But it seems like the same pattern is true for average speeds too (sort by average):

https://en.wikipedia.org/wiki/List_of_sovereign_states_by_In...

It’s a valid point, on the other hand average is much worse. 1 person at 1GBps makes up for huge numbers of people at 1.5Mbps vs 20 Mbps.

Not that all speeds above 20 Mbps are equal, but there are heavy diminishing returns. IMO, I would rank countries by percentage of 50+Mbps internet connections.

Median, mean and mode are all going to be pretty misleading if you're trying to work out if people outside of densely populated areas can get decent speeds.

You probably need to measure lower percentiles to get an idea of how bad it can get.

Why is it meaningless to look at the most common deployment?

The numbers got high enough the differences don’t really matter much.

Suppose you were looking at countries median car top speed. Fining one countries median is 150 MPH vs 156 just don’t tell you much when you can’t drive that speed on public roads. Or in internet terms Netflix etc isn’t sending 8k stream let alone multiple 8k streams.

On the other hand many internet connections can’t handle a single 4k or even 1080p so percentage of connections over X threshold is more impactful.

Now if people where measuring say the median miles per gallon or in internet terms say latency or bandwidth caps then the median actually matters more.

Netflix often uses local caches hosted by large ISPs, making it even more difficult to talk about Internet "speed".

I think that's a bad metric because it does not measure the actually speed of the network i.e I could get 2000 down but I only pay for 50.

Yeah - to get a decent idea you'd need to measure $/mbps as well as measuring average speed. And measure best available rather than what people have opted for personally.

E.g.

* Country/ISP A has 4 gigabit fiber for $100/month, and 1 gigabit for $50/month.

* Country/ISP B has 4 gigabit fiber for $100/month, and 1 gigabit for $90/month.

In country A a lot more people are going to opt for 1 gigabit - it's way cheaper and still good enough - therefore their average speed stat is going to be lower even though the actual options are strictly better than in country B.

Not everyone signs up to the highest speed tiers, but generally speaking 90% of the country has access to 1 gigabit fibre and some areas also have 4 or 8 gigabit fibre infrastructure.

You would not know if when you visit there. I stayed a month there and travelled all over. Internet was trash everywhere I went. I visited places that had time limits and data caps under 1gb. Even staying in AirBnBs that share internet with the owners, almost nobody had a good connection. I had one good place in an apartment in the core of Wellington where the internet was fine.