I can't really tell what is wrong with this kata , but the dynamic data makes it really hard to troubleshoot my code. Data constantly changes in every trial, making it impossible for me to compare my results to the expected results. Plus, one of the requirements is about the expected number of rows (also changes in every trial). How can a kata be so speficific when dealing with messy text data, which contains problems that could be solved with a certain probability. Names are not meant to be used for %100 linkage.