The idea that every author on a paper should "be accountable for all aspects of the work” is not reasonable and in practice hardly ever true. That the senior/last author should be accountable is a good idea.
Labor issues underlie a lot of this. The market is saturated for scientists and the golden age POV (which admittedly I never heard anyone actually espouse) that one would labor 5-7 years for ~1 first author paper leaves recent phd's at a large disadvantage in the job market.
The reasoning for this is quite well explained in the article.
Most universities measure staff, both current and applying, by various metrics involves authorship (number of papers, number of citations (either total, or h-index)). None of these measures care about number of authors on a paper, and it is low cost to add an extra author. Generally speaking, it is always better for n authors to be on n papers with n authors, than each write a solo paper.
Although, if you are a Phd-student or postdoc, you really need to be first author on some papers if you want to make a further career in science, at least in the biomedical fields.
The monster author lists seem to be most common in physics though, where you have these large collaborative projects, like the LHC. I guess that makes it a bit different, and that it therefore is acceptable for a Phd student or a postdoc to be listed in the author list of these papers, and not necessarily be first author.
Labor issues underlie a lot of this. The market is saturated for scientists and the golden age POV (which admittedly I never heard anyone actually espouse) that one would labor 5-7 years for ~1 first author paper leaves recent phd's at a large disadvantage in the job market.