Even though I don't find the actual ranks interesting, I find country rankings interesting because the methodology tells you what the ranker values.
E.g., the founder of Kagi created a ranking that uses 16 variables, two of which are "Medals/Capita" and "Medals/GDP". 1/8th of the ranking being based on Olympic medals, but not absolute performance, implies that it's extremely important for a country to devote a large fraction of its resources towards training people up in Olympic sports.
It's also hard for a large country to do well on this ranking, so it implies it's good for a country to be small and have a culture that's highly oriented towards a few sports it can do well at for its size.
Although there are some composite metrics, there isn't one for crime and the two crimes ranked are homicide and cocaine use, implying that these are extremely bad and other crimes are fine (or that these are very good proxies for all other crime, but that's not true).
GDP only directly appears in the denominator of Medals/GDP, so the ranking implies that having high GDP is bad.
I don't particularly care about country rankings, but I feel like this same idea applies to benchmarks. The benchmark methodology tells you what the person thinks is important, e.g., when people pass around benchmarks of terminal throughput to prove that some terminal is the best, they're effectively saying that terminal throughput is very important.
@danluu Or maybe it was an easily available metric and he didn’t think much about its implications. 😅 (This might be a common theme, both for this guy, and for benchmarks in general)
Though a large country doesn’t necessarily need to train a large part of its population to win the genetic lottery and find some extremely talented people that can benefit from a lot of resources to maximize their performance, so it might cut both ways.
@tshirtman No large country does decent on Medals/GDP. It's very difficult to sustain linear growth in medals as a country gets larger. The variance in number of medals you can win is just totally dominated by the variance in GDP. The top ranking large country in Medals/GDP is Russia, at #32
On the medals metrics in particular, I would guess that getting raw medals or weighted medals is much easier, but the author didn't like how highly that ranked the U.S. and other economic powerhouses
@danluu yeah, you are right, it helps with dominating absolute medals, but i didn’t think this through either, there is just not enough medals to get to go over the massive GDP difference between countries. Maybe more convoluted function would help, but it would certainly look like you are trying to find a metric fitting your prejudice. 😄
Add comment