Algorithms are practically everywhere now, running evaluation programs, deciding about search results or job applications. But they are far from the objective tool people would like it to be, and often they are harmful instead of helpful. O’Neil examines algorithms in this book and how they go from simple mathematics to Weapons of Math Destruction.
Weapons of Math Destruction is a very readable and very comprehensive look at the effect of algorithms, tracing both the areas in which they are used in a general way, and personalizing the effects with stories of particular people. It’s definitely a good primer on the subject.
O’Neil is a mathematician herself, but the book doesn’t explore the math of algorithms. Instead it covers their impact, succinctly laying out the many areas in which they are already in use – and what can make them so very destructive. For her, an algorithm becomes a Weapon of Math Destruction when it is used on a big scale, with opaque criteria and a (potentially) harmful outcome.
In a combination of personal anecdotes and systematic analysis of the algorithms already in use, she gives us an excellent overview over the situation in the USA. As a European, I was wondering how the situation looks like here, whether Europe is similarly affected or has more restrictions in place about which data can be used. But O’Neil’s book is not the place to get this info (this is not a criticism, just a statement of fact – I didn’t expect this information from it). There are certain parallels for sure.
I did go into reading the book knowing a little bit about the subject already and for the most part, it was a systematic confirmation of my knowledge with some more in-depth observations. But I feel that it would work for absolute newbies of the topic as well. O’Neil explains things well and strikes a good balance between anecdotes and facts.
It makes the book really easy to read, while also feeling rather comprehensive. The key take-away for me, in addition to the WMD criteria, was the fact that it all boils down to the goals: why are algorithms used for something? And in capitalism it’s usually to save money or to make more money for certain (already rich) people, instead of helping or counteracting power imbalances. If that’s what they were used for, algorithms could be great. At least according to O’Neil. I do believe that algorithms themselves could be treated more critically than that.
What is certain though is that at the moment they appear to do more harm than good. And they are certainly not a solution, but a tool. And that picture becomes very clear after reading the book.
Summarizing: a well done primer.