Each college uses a different ranking system, utilizing different factors and different measuring systems, and even measuring different things. So using the average ranking does not make sense. You can’t average between #1 for undergraduate teaching, #3 for research spending, and #6 for reputation among donors, and claim that the average ranking of the college is 3.33.
Then there is the fact that the ranking systems of these different entities are driven by different political and social agendas. Some have a vested interest in making sure that the wealthiest and most connected colleges always rank highest, others are interested in awarding publication numbers and research dollars, others want to make sure that the present power structure in academia remains as it is, while others are trying to pull attention to non-USA or non-Western colleges.
Then there is the fact that rankings like CSRankings are themselves composite, and, depending on the field within CS that interests somebody, they change. So why would a person who is interested in Machine Learning choose a colleges which has a high compasite ranking over a colleges which is highly ranked in machine learning, but has a lower composite ranking.
Finally, aside from USNews and Niche, most of these rankings do not weigh undergraduate teaching very highly, or do not include it at all, so they are not all that relevant for parents and kids who are looking for a good undergraduate education.
USNews does not have a ranking of LACs by their CS undergraduate program, so that 20 score for HMC is kind of meaningless.
As for Niche - their top factors are always the income of the students’ families, the endowment of the school, and the prestigiousity of the school…
So using a problematic metanalysis of weak or irrelevant data is hardly a good way to come to any meaningful conclusions.