Dear Smileous,

The logic of your algorithm is quite right, it would generally work absolutely the same way as kanonical BWT, the difference is in the fact that canonical BWT should be expected to work not with letters of text, but with bytes and their values where character 'a' and 'A' are quite different letters having different ACII code values, while your algorithm implement some default C# comparison method with specific comparison logic.

In order for you to fix your code you shall try to adjust default for C# sort order and to ensure that texts are sorted based on ACII codes of relevant characters.

I assume that remark regarding sort order would be useful in KATA Description, but it is absolutely not mandatory, and would be not fare to consider it to be KATA issue.