I'm not sure this is a worst practice for a couple of reasons. I tend to agree that a complete DB is a bad idea in almost all cases, especially since SQL 2000 allows single tables or columns. However, if you code to a db that requires this, you may have no choice.
1. Searching is difficult, but we can handle this in code (search "SQL" & "sql" & "Sql" & ..), but it is complicated and will impact performance. Of course, we could always us the VB approach and UPPER() everything, however, I tend to agree it is unnecessary. Users (most users) who are not programmers or techical users, will enter data and assume that "Steve Jones" == "STEVE JONES", even in other languages. If not, they will become confused and have problems. However, it is OUR job to code to handle this.
2. Case Sensitivity is good for data quality. It ensures that names are entered correctly.
3. Performance. definitely lower, but probably not noticable in 99.9% of servers. If it is, you probably need a server upgrade.
As far as programming languages go, I don't think that it is a big deal. ALL languages should force you to declare variables. If you declare @Counter, @COUNTER, @counter and use them in the same module, that's your fault.
I agree it makes things more complicated, but disagree that it affects the quality of code. People write crappy code no matter what language. That's the programmers fault. Case sensitivity doesn't affect that.
Of course, time to completion is an important factor. I like the flexibility of having constants in UPPER CASE and variables in lower case. I'd rather have more tools in the development environment to catch errors and allow me to decide if it's an error rather than blanketly handling things in one way or another.
Steve Jones