[Book Reviews] Spies, damn spies, and statistics

[thewronghands]

Comments

[marzipan_pig]

Hah ha ha your 'Tactical Primer' review is killing me ded laughing :)

[thewronghands]

Sometimes bad is just bad. [grin]

[jalenstrix]

Ooh, I am reading Nate Silver's book now with my Dad! It is lovely to see your review of it. My favorite parts are definitely the real world applicability history & discussion. (I never realized how much harder earthquake prediction is, compared to weather prediction, for example. Also, power law distributions in action - so cool! I have a whole spiel I do for my language acquisition classes about Zipf's law and such distributions in language. :) ) The only minor letdown for me was the overview of Bayesian inference - clearly, Silver knows his stuff since he makes such great stat models, but it seemed like a very unintuitive way to present it ("Look, some variables that we add and divide, just because!"). I'm of course horribly biased though, since I also cover basic Bayesian inference in my classes, and (for me), the way the calculation is done corresponds so beautifully and intuitively to the concepts of prior, likelihood, posterior, and evidence, and how they relate to each other. But anyway, yay predictions of hard, messy human

[thewronghands]

What does your Dad think of it? I agree on the favorite parts... I forgot to mention that I also liked the section on earthquakes, though I knew some of it from my favorite weatherman's blog. (He talks about prediction models and weather systems a good bit too.) I didn't find the Bayesian explanation particularly challenging, but that might be because I'm in that middle ground of already understanding how it works well enough but not being a teacher of the subject as you are.

[jalenstrix]

My Dad is also quite fond of the historical overview stuff, but I think he was hoping for more in-depth coverage of the stat modeling. It's my turn to pick what we read next, so I might try to find something like a gentle introduction to Bayesian inference, or something similar. ;)

[moodyduck]

Ah, you liked Viestur's book for the same reason I did. He also seems to give credit where credit is due to luck, which many people don't...but it is also clear how his methodical and cautious nature led to success. I've read his other books but didn't like them as much. But then I'd generally rather read first-person accounts than third person assessments of other people's adventures or misadventures. You may like those more than I did.

I got Signal and the Noise for Christmas and am looking forward to reading it. Felt obligatory, for my line of work. Which includes a lot of not just doing the math but figuring out how to express it to people who can't or won't put in the effort to fully understand it.

[thewronghands]

I like the misadventure stories when they're well told from a third party perspective -- Accidents in North American Mountaineering is split between first person and third person accounts, and I still find that interesting and valuable in the Don't Do This sense. I think I apply different bias filters to first party and third party accounts... they have Venn diagram-like blind spots. Personal bias and not willing to admit to your own mistakes in the outside on one side, lack of visibility into what happened because you weren't there and can't reconstruct it on the other. So, yeah, there's the armchair generaling that often happens with third parties who feel themselves superior/invincible, but to me that's a different filter for the unknown unknowns rather than a lesser mode. I will eyeball his other books when next I come across them in a physical books bookstore, but I'm not running right out to buy absolutely everything else he ever did tomorrow. [grin

[moodyduck]

Because of your interest in the Accidents book and thing-going-wrong-in-the-wilderness in general, I thought you might like the others more than I do. I'm more interested in how people mentally deal with their challenges and how it integrates with their lives, than with the raw events themselves, hence the reduced interest in third party accounts unless they are very well done.

[thewronghands]

[nods] That part is interesting too, I agree. The integrating with their lives particularly, since so many of us don't have life or death in our day jobs.

[thewronghands]

Rory Miller! Wrote a whole mini book of drills encouraging you to do just that. [grin]