2. I think the question should be why on earth should they not use the same operator symbol? In most other mainstream languages they are. I know the answer is "because [Char] isn't part of the Number typeclass" but that misses the point.
5. Both Text.Regex and the split library someone else mentioned operators on strings ([Char]). Data.Aeson uses ByteStrings. So not everything in Haskell-land uses Text when they should which leads to lots of annoying string packing and unpacking. Contrast this with Python (3) where you have (unicode) strings and (byte) buffers. Only. If you want more efficiency there are 3rd party lazy implementations of them, but they aren't forced upon you like in Haskell.
2. They should never use the same symbol. As the parent said, if you want the generalized notion of a monoid then use that (<> or mappend), otherwise each library should not be conflating its instance of a TypeClass with other libraries' instances of a TypeClass - it is confusing and you end up with stuff like:
import Some.Lib as L
import Some.Other.Lib as OL
(L.+) 2 3
(OL.+) "2" "3"
vs. import Data.Monoid
(Sum 2) <> (Sum 3)
"2" <> "3"
Which is inevitable for some libraries (Fay for example re-defines a lot of Prelude functions) and sometimes accidental; but it is confusing and bad design unless you define a way to generalize it (Monoids!).5. Data.Aeson uses ByteStrings for the JSON serialization, when encoding/decoding you provide your own instance defining the Types (can be Text/String/ByteString) so I don't understand your gripe here; Text.Regex shouldn't be used for Data.Text, that's what this package is for: http://hackage.haskell.org/packages/archive/text-icu/0.6.3.5...
text-icu is an experimental 3rd party binding and doesn't work like pcre anyway. It's not something you would use over Haskell's standard library regexp support, so you still have to deal with X number of different string types.
I personally haven't used the Text.ICU package but I know many, big, packages that retain the "experimental" flag. So that shouldn't necessarily deter usage; just because it's 3rd party doesn't mean it's bad either - many fundamental components of the Haskell ecosystem are 3rd party! There are also many libraries that implement bindings - I don't necessarily see why that is a bad thing either; obviously a pure Haskell approach would be nice but in some cases it does defeat the purpose to reinvent something.
What I do agree with you on, though, is the lack of clarity in what should be used. I think the Haskell Platform is a good start in that direction but there are few docs written on "this is how you deal with strings and unicode in Haskell and all the libraries we recommend for it".
The good news is, this language has the capability to serve both the research needs of academics and the practical needs of implementers; so uncovering this stuff is very good.
2. Because they are not the same operation. The fact that some languages make them the same operator is not an indication that doing the wrong thing is good. And I do not believe most mainstream languages make that mistake. Again, there is a name for the general concept of "things with an identity and some form of append operation". The name is monoid, it defines the exact operator you want: <>
5. I don't see the problem here. Of course aeson encodes to and from bytestrings, json specifies an encoding already. The entire purpose is to take data, and encode it as a bunch of bytes. You shouldn't be trying to split encoded json as if it were a string, so what is the issue? Your contrasting it with python weakens your point, as it makes it clear that haskell only has one more option than python does.
2. Why on earth would string concatenation be the same operator as addition? That isn't the same operation. I want to know if I used + that I have to have gotten a number, or else it won't compile. Getting a string would be very unhelpful. If you just want any monoid, then there is a generic operator for that, it is <> or mappend.
5. The different types are for different purposes. String is a list of chars. You use normal list functions on it. Bytestring is not a string, it is an efficient container for storing bytes. It is used for things like networking. Text is an encoding aware, efficient representation of strings. Use this for text data. There is an IsString typeclass for generic functions that operate on any type that can behave like a string.
7. I've always heard the exact opposite from everyone, and my experience has also been the opposite of yours. The only other computer related community I have seen that is as helpful as haskell is postgresql.