Algorithms, at the heart of computer systems such as search engines, can be racist or at least reinforce racism. That is the claim of this book’s author.

It is difficult to state the validity of the argument, but it is an interesting and considered look at a possible problem in any case. The author cites by way of an example a different type of search result received when you search for ‘black girls’ rather than ‘white girls’, but to my mind fails to accept that a search engine can only reveal what it has indexed, especially in the absence of hard evidence to suggest that results are being manipulated with wrongful intent. My own quick Google search didn’t come to the same conclusion as the author in any case.

Perhaps there is a problem at-large with some of the content on the Internet, but that is not the fault of the search engine provider. What can a search engine operator do? Should they or could they somehow positively affect a supposedly neutral algorithm to amend any biases that may exist? The author certainly argues that search engine operators are not offering an equal playing field for ‘all forms of ideas, identities, and activities’ due to a ‘biased set of search algorithms that privilege whiteness and discriminate against people of colour, specifically women of colour.’ The author also claims that there is a ‘culture of racism and sexism in the way discoverability is created online’ but maybe there is merit in respectfully saying that a neutral intermediary should remain neutral.

Clearly, a search engine operator may be able to put some safeguards in place, but part of a search engine’s ranking are various signals such as link strength, authority and other user-generated activities. It is not such a black and white issue, if you pardon the possibly inappropriate idiom. As a reviewer I disagree with some of the author’s viewpoints, although respect their position and understand where they are coming from. There are real problems with a limited number of search engine providers, just as there can be issues surrounding any near-monopoly behaviour with companies, but implied racism seems, so far, to be unproven.

In any case, the author promotes and argues their case well. It is far from an ‘internet rant’. Even if you disagree with all, or part, of the author’s claims, it still can be an interesting and worthy read. The author states that this book is the start of trying to make such issues visible, and there is a validity in making all algorithms truly free of any actual bias, but where I differ from the author is the possible way this can be achieved and the actions that may be reasonably expected of providers. It can be a very fine line but neither does it signify one approves of harmful or wrongful framing and descriptions either. If anything can be done to make the algorithms more intelligent without blindly filtering out content that is to be, of course, encouraged.

This turned out to be an interesting, challenging and informative read on something I may not ordinarily have invested so much time on. It may be something you should consider for a bit of intellectual stimulation if you are not otherwise connected to the issues of race, equality and society.

Algorithms of Oppression Book Cover Algorithms of Oppression
Safiya Umoja Noble
Social Science
NYU Press
2018
256