What would be some really good reasons to not use the Latin1_General_CI_AS collation on a system which handles English, German, Japanese, Chinese data?

I'm having a really hard time finding good sources which compare collations and answer my above question as well as the following

1. How is Japanese_CI_AS different from Latin1_General_CI_AS when it comes to sorting non-Japanese characters?
2. Does UCA sort differently that Japanese_CI_AS and Latin1_General_CI_AS?
3. What is standard industry practice for systems used globally? (For example, what does Facebook, Twitter, Google, Flickr, Baidu or Microsoft, IBM, SAP use?)
4. Does Latin1_General_CI_AS define a sort order for Japanese characters? If not, how/why does Japanese text work in Latin1_General_CI_AS?

Basically I'm trying to learn how to pick the appropriate collation :)

Thanks in advance!