Apr 07, 2015

Reading list Q1 2015

Books I read in the first quarter of 2015: [1]

Discourse on the Method by René Descartes
Two summers ago, as I was struggling to conceive of a topic for my senior thesis, a professor of mine recommended reading the Discourse on the Method to settle some of the more existential bits of the process. I got around to reading it this January.

I have just a couple things to say about the Discourse now:

I don't at all understand how Descartes is able to confidently move from "I think, therefore I am" to "God exists." The line of thinking might go something like:


  1. I think,
  2. therefore I exist.
  3. I determined step 2. using a process of reason.
  4. Because I'm confident in the movement from 1. to 2., I can be confident
  in this process of reason.
  5. I didn't create this process of reason that I'm using.
  6. Because I'm confident in a process that I didn't create, it must have
  been created by a being separate from myself and more capable than myself.
  7. We call this being "God".
  

But there is something fishy (or circular) in step 4 – "I'm confident in the results of this process, therefore I can be confident in this process generally." It's not clear why I am confident in the results of the process (though I am), and it's not clear why being confident in the results of a process should mean that we can be confident in the process overall.[2] But enough from the armchair.

I hear Wittgenstein has interesting things to say about this – specifically, how could a disembodied Cartesian agent (i.e. an agent confident only in the fact that ze exists) formulate thoughts? How could ze be confident that zir language use was meaningful? How would ze learn language in the first place? This line of thinking seems to go strongly against Cartesian skepticism.

There's obviously a lot going on in this part of the Discourse, so I'm reluctantly to pragmatically dismiss it after curt examination here. I'd like to revisit it at some point, and pick through his reasoning with a little more care.

The Murders in the Rue Morgue: The Dupin Tales by Edgar Allen Poe
Consists of three stories: the title story, the Mystery of Marie Rogêt, and the Purloined Letter. I enjoyed them, but as with most mysteries, didn't quite understand what all the fuss is about. Reading the stories reminded me of this adaptation, one of the highlights of my childhood.

The Years of Lyndon Johnson: The Path to Power by Robert A. Caro
Part one of arguably the greatest biography yet written. Reads like a novel. A novel centered on an amoral, recklessly driven, ruthlessly ambitious political climber.

Mr. Penumbra's 24-Hour Bookstore by Robin Sloan
What a romp. Any story which features both fictional ancient secret orders and real-world tech mammoths is enjoyable by default. Getting to cruise around the Bay with your likable, mild-mannered protagonist is just icing.
Plus, I've gotten to hang out with the author a couple of times, which was great fun – he's a cool guy.

The Years of Lyndon Johnson: Means of Ascent by Robert A. Caro
Part two of arguably the greatest biography yet written. Our hero arrives in elected office and begins pulling levers.

Portfolios of the Poor by Daryl Collins et al.
When a team of economists tries to track every financial transaction that a couple dozen very poor families make, what do they learn? That very poor people think a lot about money, and have sophisticated methods of handling it. Millions of people live on less than \$2 a day, but they're not living hand-to-mouth. First chapter is online for free here.

Understanding Power by Noam Chomsky
I previously proposed a project for this book. I haven't started it yet. But that's okay. I'm not on a deadline.

The Short and Tragic Life of Robert Peace by Jeff Hobbs
As the title promises, this is a tragic story. It's also a beautiful story. It made me further appreciate the determining effect that environment, upbringing, and social standing has on your life.

Superintelligence: Paths, Dangers, Strategies by Nick Bostrom
Artificial Intelligence is a subject with a powerful draw on me. AI sits at some intersection of my favorite topics – precise thinking, epistemology, ethics, and machines that can compute. Yet I'm very much on the outside of the thing. I can read things and then think "well, that's interesting," but I don't have the skills to contribute to the project. Happily, skills can be gained with a bit of elbow grease.

But that's an aside. In my read of it, Superintelligence focuses on two topics:

  1. Potential ways in which a superintelligent general AI (i.e. a computer that is better than humans at most tasks that humans do) could arise, and the implications of these pathways.
  2. How to go about teaching morality to an AI, so that the AI doesn't decide on some very silly goal and then proceed to execute this goal with extreme efficiency.

The treatment of topic 2. was very interesting. It really is the question. If we are able to endow an AI with a robust ethical framework, all our speculations, preparations, and precautions are unimportant. (granted that it may be impossible to endow an AI with a framework of sufficient robustness, and even if this were achieved, it may be impossible to know if the framework was sufficiently robust before letting the AI go about its business. And in these cases, all the speculations, preparations, and precautions would be very valuable.)

An Unquiet Mind by Kay Redfield Jamison
Powerful memoir of manic-depression. A fast read that paints a strong impression of the condition – both the utter drudgery of depression and the alluring, empowering, careening drive of mania.

The Shadow of the Sun by Ryszard Kapuscinski
Probably the best book I've yet read about Africa. For decades and decades, Kapuscinski was the Africa correspondent for socialist Poland. If his memoir is any guide, he has been everywhere on the continent and done most everything. And he is able to write about it in beautiful prose.

Poor Economics by Abhijit Banerjee and Esther Duflo
Very good at driving home three lessons:

  1. Trying to develop the third world is a very complicated problem.
  2. Very complicated problems are not amenable to sweeping theories of change.
  3. Well-designed empirical tests can lend some insight into what will actually help.

[1]: Excludes online media, academic articles, books I am currently reading but have not finished, and books I started and now have little intention of completing.
[2]: I'm using a logical, step-by-step argument to examine the nature of logic. Yes, it's both problematic and meta.

[rereads: 3, edits: minor tweaks, formatting edits to the Cartesian proof, further formatting edits]