Scan barcode
castle_samj's review against another edition
informative
reflective
slow-paced
3.75
I appreciate the spotlight on large-scale systems that impact the livelihood of common people. My largest takeaway is to increase my considerations for other people and groups whrn writing future algorithms - every choice makes a difference to someone. But wow this book makes it seem almost impossible to right these wrongs. Maybe if enough coders and engineers make it their goal to reduce WMDs, we will see a more equitable use of maths and sciences.
gretatimaite's review against another edition
Finally read the book. I had high hopes for it, maybe too high? Possibly it's a result of this book receiving quite a bit exposure as there's an increase in literature that deals with data science uses and misuses.
Briefly, the book deals with the harmful application of math that manifests through models. Models are biased and if we aren't careful in their implementation they can have detrimental effect on daily lives (someone might not be able to land a job because resume scanning machines are biased). I agree with the premise of this book.
But I really expected more theoretical engagement. The only time she really developed her idea of what is a WMD is the introduction and discussion on models in chapter 1. And it's a shame. It was the most interesting part of the book. The following chapters (except conclusions, ofc) was about the examples of WMD in real life: university admissions, various scoring/ranking systems, etc. They are good and needed but I soon felt bored. I need theory OK? It's not enough to tell me that this particular model fits her definitions of WMD*. The definition itself is somewhat unclear. On the other hand, I do understand that this book was intended as an intro to those who have never heard of models. BUT it had more potential, OK?
Also, a massive minus for me was referencing or the lack of it. Please, reference properly (a really weird method is used in this book). Reading O'Neil's book felt like there are no references but then at the end of it there's a bunch of notes that shows her sources and a bit of text to which it denotes. Does a reader have to guess which bit of text is referenced and constantly check the notes? NO! It's the job of a writer and editor to provide the reader with the information.
Last thing to close this review-turned-into-a-rant is that she considered the effects of WMDs to different groups of people. In other words, she didn't homogenize the victims of WMD but thought about the context in which WMD takes place. Again, it could have been better explored but hey, I'm not ranting again, OK?
Also, the quote that I think sums her book up really well (and I like it myself):
* "these are the three elements of a WMD: Opacity, Scale, and Damage." (p. 33)
Briefly, the book deals with the harmful application of math that manifests through models. Models are biased and if we aren't careful in their implementation they can have detrimental effect on daily lives (someone might not be able to land a job because resume scanning machines are biased). I agree with the premise of this book.
But I really expected more theoretical engagement. The only time she really developed her idea of what is a WMD is the introduction and discussion on models in chapter 1. And it's a shame. It was the most interesting part of the book. The following chapters (except conclusions, ofc) was about the examples of WMD in real life: university admissions, various scoring/ranking systems, etc. They are good and needed but I soon felt bored. I need theory OK? It's not enough to tell me that this particular model fits her definitions of WMD*. The definition itself is somewhat unclear. On the other hand, I do understand that this book was intended as an intro to those who have never heard of models. BUT it had more potential, OK?
Also, a massive minus for me was referencing or the lack of it. Please, reference properly (a really weird method is used in this book). Reading O'Neil's book felt like there are no references but then at the end of it there's a bunch of notes that shows her sources and a bit of text to which it denotes. Does a reader have to guess which bit of text is referenced and constantly check the notes? NO! It's the job of a writer and editor to provide the reader with the information.
Last thing to close this review-turned-into-a-rant is that she considered the effects of WMDs to different groups of people. In other words, she didn't homogenize the victims of WMD but thought about the context in which WMD takes place. Again, it could have been better explored but hey, I'm not ranting again, OK?
Also, the quote that I think sums her book up really well (and I like it myself):
"Mathematical models should be our tools, not our masters." (p.178)
* "these are the three elements of a WMD: Opacity, Scale, and Damage." (p. 33)
bootless_bandit's review against another edition
3.0
Interesting book about the power that mathematical models have in our data-based society. Certainly helped me to think about different ways in which we need to be cautious when building tools solely driven by information.
meaghanjohns's review against another edition
5.0
"A model's blind spots reflect the judgments and priorities of its creators. ... Our own values and desires influence our choices, from the data we choose to collect to the questions we ask. Models are opinions embedded in mathematics."
I cannot recommend this book enough. This is an absolute must read for anyone working with predictive models, algorithms, or big data, but it's also incredibly useful for understanding the world of algorithms that we're now living in - whether or not you work in a related field. O'Neil warns that going forward with the assumption that mathematical models are neutral and inevitable without thinking critically about their impact can cause a lot of damage to both individuals and our society as a whole. Weapons of Math Destruction is a thoughtful and well-written book with an important and relevant message.
I cannot recommend this book enough. This is an absolute must read for anyone working with predictive models, algorithms, or big data, but it's also incredibly useful for understanding the world of algorithms that we're now living in - whether or not you work in a related field. O'Neil warns that going forward with the assumption that mathematical models are neutral and inevitable without thinking critically about their impact can cause a lot of damage to both individuals and our society as a whole. Weapons of Math Destruction is a thoughtful and well-written book with an important and relevant message.
pandoramedusa's review
3.0
Informative but ar times very dense with information. I really liked the many examples that were given
colewawa's review against another edition
5.0
This book was excellent, especially because it challenged a belief that I previously held to very firmly, that natural greedy outcomes of collecting personal data would tend towards fairness. I will have to think more, though I don't think it changed my mind. More so, it convinced me that 1. the data is being used improperly and stupidly in some cases which leads to inequality, and 2. there are vital flaws in our government that cause these injustices to intensify. Because any argument, no matter how true because I usually agree, that actions are being made out of a causative relationship when it is just a vague correlation, does not support why it would be financially wise for a company to do it, since someone else could come in and offer more fair costs knowing the assessed risk is false and therefore undercut them. The real fear is that the government needs to jumpstart some of these initiatives to right the wrongs and that isn't being done. Also I wonder if many of the studies would be more or less accurate to use socio-economic class instead of race.