I can't blame Louis for that, as he simply does what works for him. So who or what is responsible for that mess?! The actual tools that were used to create most of the content - the so-called WYSIWYG HTML editors! Many editors of this kind (i.e. good ol' MS Word) produce extremely ugly and beyond-bloated HTML document structure, and sadly that's exactly what we are dealing with here. In some cases the documents are filled with hundreds of lines that consist of nothing but useless metadata. This extra data is there only to allow for more accurate reverse-conversion of the document. Even if we ignore the metadata the documents are stuffed with redundant and/or nonsensical HTML constructions - constructions no sane human being would ever create. To make the situation even worse, the document structure deteriorates over time with all the incremental updates and modifications. This often leads to formatting problems and all kind of strange inconsistencies. While modern browsers and modern computers typically won't have too much trouble parsing and displaying this mess, you may run into trouble if you access the pages from some older machine that doesn't have many CPU cycles to spare (like any of our beloved PS/2s), and/or from a browser that isn't smart enough to cope with the poor document structure.
Another downside of the WYSIWYG approach is that it makes it significantly more difficult to compare different versions of the same document when merging changes from different sources (i.e. from me and from Louis). This is because the entire document gets re-generated with every change, and unfortunately the generator doesn't seems to be very stable (or perhaps multiple different tools or settings are being used) and the result differs greatly even if only a very minor change was made. This makes it much more difficult to automate the merging process. Manual diff/merge tools must be used in many cases to review and merge the changes, which is - of course - slower and more tedious.
I have managed to clean up many pages and remove a huge amount of the auto-generated bloat, but there is much more to go. No WYSIWYG editors or similar tools are used to create or edit pages of the MAD Edition, only plain-text editors (Notepad++ and Sublime Text). Additionally any larger external changes are first checked and polished before they get merged to this branch of the Tool. The pages that already received this special treatment are easier to edit, there are far less formatting problems and they have more consistent look.