Did Unicode break?

[Timberwoof]

I found the German-language section and noticed that the Umlauts and Ess-Zett (ß) are broken all over. Let's see if I can reproduce what I see: 
I can make these on my keyboard and they look okay while posting: ä, ö, ü, Ä, Ö, Ü, ß look okay. But this is the sort of thing I see:  heißt er E6000. That looks like a Unicode byte pair not getting displayed as an Ess-Zett (ß). 

[Timberwoof]

Huh. My newly-entered text is okay, but what I copied and pasted looks like what it looks like. 

[Locitus[Admin]]

Older posts from a different database collation are a bit broken.

[Daetrin[Admin]]

Is there a pattern that can be used to do a find/replace?  I can run a database update on certain posts if so, e.g. older than a certain date.

[Locitus[Admin]]

Around 2015 I think it was we moved to hostgator and the recommended collation .

[Timberwoof]

Huh. The error isn't actually consistent. Some old posts have the problem; some do not. Sometime between June and September 2015 the problem went away. 

 

1-glyph substitutions: 
ü -> ü

ä -> ä

ö -> ö

ß -> ß

 

Examples: 

Kostüme -> Kostüme 

längere -> längere

eröffnen -> eröffnen

Grüße -> Grüße

Oberkörperpanzer -> Oberkörperpanzer = Upper Body Armor. Isn't that a cool word? 

 

These examples don't include any capitals but I suspect 1) a search for Ã will find them and 2) these are the 99% solution. There is a lot of relevant and useful text that contains a lot of these errors. Lucky for me I don't actually have to read this; I can get by in English.

 

Spanish also has the problem but I don't know what the substitutions are. The problem is rare as the volume is low. 

 

Ohboy. This is Bad with a capital Bad. The one thread in the Russian section looks like this: 

 

The last reply looks okay, so this may be a DNF: Do Not Fix.