ah, Eastern Europe is a homogeneous slop, good to know. The 25th is only celebrated as a national holiday in like 90% of Eastern European countries. Russia and Serbia celebrate it on January 7th instead, and even that is only due to their use of the old Julian calendar. But sure, normal day in a very narrow and obscure definition of Eastern Europe.
I loved the article, the translation is pretty good, too.
Can you share a link to the full Chinese text with me, please? My wife is also from the Gobi desert and I think she would love to read the original version.
edit: I just found 我在404长大 [1] on douban, it seems the be the original version
I tried to open the link in 2 browsers, both times the step "Establishing connection" resulted in "Dashboard Load Failed
Reset operation timed out after 60 seconds"
Ugh sorry about that - was trying to bundle the duckdb packages in the docker image and accidentally included them in the prod deploy, and my static hosting is too slow to server those in time. (works in github CI though, of course - I'll need an extra post-deploy validation on the serve latency).
Anyway, if the researchers are not blinded there are many possible sources of errors.
Perhaps they do the first test in the morning, the sound just before lunch and the second test in the afternoon is made by another person that is more/less friendly to the rats, or the rats has the stomach more full/empty.
After changing a program and running benchmark, I sometimes run it again if the new program is not faster as I expected. I even gave a second chance to deterministic test, that is as useful as it sounds. It's possible that if the rat does not collaborate the researchers hit's the equivalent of Ctr-F5 just to be sure.
It's hard to be 100% neutral, so a method is to not know to ensure all rats have exactly the same test conditions.
