I am partisan of some uniformity in the results, maybe it is lazyness of which I see many advantages...
As for "Help the bookseller" I let you guess who was the translator! Maybe at that time I was a bit influenced by your saying about strings. Furthermore "my" katas" are small kata and not real applications so the form of the result is not very important. I have no boss asking for results this way or that way:-) I prefer to spend more time on the way I get a result than on the result itself.

Still, a dictionary would be much better than a string. And it's possible in almost every language, also called associative array. To quote Wikipedia (like you did ;)):

Many programming languages include associative arrays as primitive data types, and they are available in software libraries for many others.

But I'm not going to point out all the advantages of not using strings this time :P. To be honest, I simply believe that you're too lazy to change the way you're testing or don't want to use the proper types for each language. Back at Help the bookseller you at least allowed translators to use the correct type (see Clojure and Haskell), but nowadays it seems like you're using strings all the time. Funny enough, I think I've found the first comment on the problem:

I've justiced that you have the tendency to use strings in your katas. This might be fine for some small katas etc, but in real applications you usually want to use the correct data type.

Can't believe it has been 4 months. By the way, you can still improve your description.

@IVBakker:

In the description "Can you tell me how much f he could possibly have

You are not correct, it is said:

how much money f he could possibly have... ?

So f is money, isn't it? Then Each answer will be given as ["M: f", "B: b", "C: c"], M: f means that "M" is for the value "f" of 'M'oney like "B": b is for boat and "C": c for cars. Oh, yeah, I agree that is a bit cryptic. I will try to better the description.
As for

returning string is poorly designed, a list of dictionnary would be much more useful.

"poorly" is not very kind for the poor author:-) A list of dictionnary would be much more useful but "useful" to what? I had lots of discussions about strings wit @bkaes. Strings are data as another ones, common in their form to all languages (Wikipedia: Strings are such an important and useful datatype that they are implemented in nearly every programming language), easy to manipulate and allow to concentrate on the core of the function and not to the form of its result.

@g964 no I don't think anyone would understand in 5sec. I still don't get what is what.

In the description "Can you tell me how much f he could possibly have. Then further in the description Each answer will be given as ["M: f", "B: b", "C: c"] and all the answers as [ ["M: f", "B: b", "C: c"] ... ]. what is M?
Also returning string is poorly designed, a list of dictionnary would be much more useful.

[Python]
I'm also getting the timeout issue, so I'm trying to optimize my solution.

However I personally think this might have some more sense if you separate the really big test for another kata, one specifically about optimization.
I know my code works. Not so fast for enormous numbers? Sure, but the kata is about Fundamentals.

@sebastialonso: I modified the description, thanks.
@bkaes: I'll read it, thanks.

@g964: Added a section concerning stories/tasks in the guide.

Not exactly a run-on but this paragraph reads too complex.

Could you tell me how much money f he could, possibly, have (if m-n is big enough, you might have a lot of possible f...), the cost c of a "xxx" car and the cost b of a "yyy" boat in each possible case? So, I will have a better idea about his fortune.

The sentence in () is completely extra, the sentence before the parenthesis and the one after don't connect in any way, it's extremely confusing per se and the parenthesis doesn't help a bit. Get rid of the xxx and yyy descriptors, they add nothing of value to the problem. Not one comma is necessary in this paragraph.

I know this is mostly whining but there are three steps when reading a problem: seeing if I understand what's written, understanding the actual problem and then know how to solve it. It took me way more to understand what was written than the actual problem.

Please could you give an example in this kata? I see no run-on sentence in it but I could be wrong.

Your descriptions often contain run-on sentences. Consider getting rid of those.

Some guys understand in 5 seconds, somme others in 5 mn, it is a fact of nature:-)

!!! round answers to the nearest hundredth!!!

Don't. It's your responsibilty to settle floating point arithmetic differences.

Lot of issues with the testing. There's no much congruency between the pre-submit tests and the post submit tests (have to edit the code mid-tests). I don't what that is, but it happened to me.