clach na ben (![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
clachnaben) wrote2019-01-28 05:05 pm
clachnaben) wrote2019-01-28 05:05 pm
Entry tags:
Book Log: Signal and Noise
I finally finished Nate Silver's The Signal and the Noise, aporximately 20 million years since it was in fashion. It was one of my new year's resolutions to read some of the unread non-fiction books I keep around the house, so at least one of them is done!
I tend to donate or pass on books I've finished that I don't think I'll read again, and I doubt that about this one, so if you'd like it and we have some liklihood of meeting in person/you live in the UK, let me know and I'll set it aside for you.
On the topic of the book, in essense, I can think of no more white boy a thought that "I can predict baseball, why can't I predict ~~THE PRESIDENT". Anyway, it turns out Silver couldn't even predict baseball that well, and a lot of the book is about how predicting usually fails. Extremely confusingly, he spends most of the book extremely optimistic about political predictions, describes politics as a data-rich environment with high data quality (to compare it to other fields he talks about in the book, like national secuirity, and earthquakes), and seemingly trusts polls a lot. And we all know how well that went for him.
I think the first half of the book is the best - these are the chapters on baseball, earthquakes, hurricanes and the weather. He derives some "best practice" for predicting that makes sense, and then completely throws them out the window for the final chapters, where he suggests terrorist attacks obey the same statistical rules as earthquakes (??), and includes an entirely incomprehensible chapter on climate change & global warming. Not a great end.
Overall, it's a decent read for non-fiction. It's clear, he's a good writer, but I think he doesn't interrogate his sources very hard, likes experts too much, and doesn't think about data quality enough. I would have loved more on how we collect data, and how poorly collected data is mostly unusable.
My next book is record of a spaceborn few, which I'm a little bit in. It started super sad so I had to nope out, but I'm making another run at it. Is it all like the start, i.e. basically just ficlets? I liked Becky Chambers other stuff, so I'm sure I'll enjoy it. After that, I have to pick another non-fiction book!
I tend to donate or pass on books I've finished that I don't think I'll read again, and I doubt that about this one, so if you'd like it and we have some liklihood of meeting in person/you live in the UK, let me know and I'll set it aside for you.
On the topic of the book, in essense, I can think of no more white boy a thought that "I can predict baseball, why can't I predict ~~THE PRESIDENT". Anyway, it turns out Silver couldn't even predict baseball that well, and a lot of the book is about how predicting usually fails. Extremely confusingly, he spends most of the book extremely optimistic about political predictions, describes politics as a data-rich environment with high data quality (to compare it to other fields he talks about in the book, like national secuirity, and earthquakes), and seemingly trusts polls a lot. And we all know how well that went for him.
I think the first half of the book is the best - these are the chapters on baseball, earthquakes, hurricanes and the weather. He derives some "best practice" for predicting that makes sense, and then completely throws them out the window for the final chapters, where he suggests terrorist attacks obey the same statistical rules as earthquakes (??), and includes an entirely incomprehensible chapter on climate change & global warming. Not a great end.
Overall, it's a decent read for non-fiction. It's clear, he's a good writer, but I think he doesn't interrogate his sources very hard, likes experts too much, and doesn't think about data quality enough. I would have loved more on how we collect data, and how poorly collected data is mostly unusable.
My next book is record of a spaceborn few, which I'm a little bit in. It started super sad so I had to nope out, but I'm making another run at it. Is it all like the start, i.e. basically just ficlets? I liked Becky Chambers other stuff, so I'm sure I'll enjoy it. After that, I have to pick another non-fiction book!