The author cites the common CEPR paper [0], but missed its most interesting finding. It found that the algorithms definitively did show signs of collusive behaviour, but that their chosen equilibrium price point was far below the Nash equilibrium. That is, the researchers expected these algorithms to maximally extort the consumers, but they only modestly extorted the economy's consumers.

[0] - https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3304991