More interesting albeit difficult to quantify would be PG's writing style.
If you've read a PG essay and found yourself agreeing because you're gradually taken towards a conclusion yourself instead of having it preached to you, you'll know what I'm referring to.
Turns out there is a style for that.
I am no writing expert, but I feel PG's style is (intentional or otherwise) most akin to Classic Style as outlined in the book "Clear and Simple as the Truth" (overview from Robin Hanson of Overcoming Bias here: http://www.overcomingbias.com/2009/03/deceptive-writing-styl.... ).
Most telling excerpt: "Classic style is in its own view clear and simple as the truth. It adopts the stance that its purpose is presentation; its motive is disinterested truth. Successful presentation consists of aligning language with truth, and the test of this alignment is clarity and simplicity."
Paul, two questions I've always wanted to ask and now seems like the right time: Why do you use an image for your essay titles? What do you use to write and publish your essays to your site?
Both have the same answer: Yahoo Store. It was written in an era when browsers had about 3 fonts, so when sites wanted to to do anything interesting with typography the text had to be rendered as images. One day I'll move my site, but there's a bit of activation energy involved, because I'll have to write another site builder first.
Unless you want to build it from scratch, it's a matter of a two or three dozen lines of glue code between libraries.
I've recently built a custom, light CMS for a friend based on
* PHP
* Smarty for the templates [1]
* a few YAML files for the DB [2]
* Markdown and Smartypants for the text markup. One file per blog post. [3][4]
* Some .htaccess magic to create clean URLs
Sprinkle some CSS on top and you've got a new site :-).
I didn't implement any search facility, but I guess there are several libraries that cover that functionality as well.
See [5], [6] and for Common Lisp equivalents. I couldn't find a mature YAML parser but JSON is of course a decent alternative. There's [7], but the project is on hold and not documented at all.
I guess the biggest hurdle would be the translation of the Yahoo Store posts into Markdown. Note that Markdown supports embedded HTML, which should make things easier.
How does it feel to be so thoroughly analysed in a purely numerical fashion? I would feel as though things were being turned up about myself that I never even thought about.
My first reaction is curiosity, the second is a sort of flinching at the probable reaction of haters (whose reactions I've internalized by this point) to the idea that it matters what month I write things in, and the third is that I really should work harder. Especially in June.
I don't take taking advice lightly. It really takes a lot to convince me that what you are preaching is something I should thoroughly consider. Despite this fact, I find most of PG's essays to really hit the spot.
I've realized that one of the reasons why I trust PG essays, besides the fact that they are written with the general tone of discovering the truth, rather than inventing the truth, is because a lot of them are old. Well, relatively old. And quite plain. A few years ago when I first started reading them, my instinct was to think less of them for this. But that the essays I like most are over 5 years old, yet still highly relevant, only adds to their credibility. I don't think a lot of stuff I read on blogs today will stand the test of time as well, though I admittedly don't read enough.
Anyway, I realize this is slightly off topic. So, um, nice charts, and keep 'em coming Paul.
A nitpick: People in English departments would probably call this a "rhetorical analysis," since "literary analysis" usually refers to fiction.
I would say that Graham tends to have the modernist tendency to cut anything superfluous; to quote Milan Kundera in _Encounter_, "Almost all great modern artists mean to do away with 'filler,' do away with whatever comes from habit, whatever keeps them from getting directly and exclusively at the essential (the essential: the thing the artist himself, and only he, is able to say)." Kundera, like Graham, is very good at doing this in his nonfiction; his essays on the novel contain more in 200 broadly spaced pages than most tomes about literary theory that I've read.
One other thing: is very, very good at not repeating himself and not having ideas with an extreme amount of overlap. I think that's one reason he tends to say things like, "I'm not saying X" as clarification -- to avoid misreadings that might result from his terseness.
This tendency towards minimalism, by the way, is one of the same things that John Sculley talks about Steve Jobs having in terms of industrial design: http://www.cultofmac.com/john-sculley-on-steve-jobs-the-full... . Graham does it in nonfiction writing. Elmore Leonard does it in fiction; he's incredibly skilled at compacting a lot of action into a very small space. If you're looking for fiction that's as "dense" (for lack of a better term) as nonfiction, try Leonard. His subject—capers of a vaguely criminal and usually underworld nature—might not appeal to you, but the man has style.
In addition, Graham is very good at using metaphors and comparisons with history. There isn't really a good way to analyze metaphor that I know of. Some of the better attempts include Lakoff and Johnson's _Metaphors We Live By_ and Steven Pinker's _The Stuff of Thought_. But, as far as I know, no one has a complete theory of why we think associatively and metaphorically. Good writers, however, will tend to exploit the tendency to view one thing in term of another.
The problem with these kinds of generalizations is that they're very hard to make concrete. You might notice that I don't cite a lot of examples from Graham's writing because it''s harder to prove the absence of perceived filler than it is to identify filler when you see it.
That's something I'd like to do. This is really just a starting point. I think there's interesting info to be gleaned even without looking at the content of the essays.
That's also where our own thoughts (my cofndr & I) are heading to: instead of quality, let's look at quantity and distribution patterns and maybe later we'll find correlation with quality. Test if regularity of a practice breeds quality.
Because I want to know when something was first available (am I late to see it? early?), and where it fits in the context of other things that were written around that time.
The majority of my information comes in time-based feeds and streams, whereas these essays have inscrutable URLS and date formatting. I suppose that's by design but I don't find the lack of information adds significant value.
For me, what has been most noticeable is his shifting focus. When he started out, he mostly wrote about Lisp and programming languages. Now he primarily writes about startups. Since I'm a lot more interested in the former than the latter, I've gradually stopped reading his newer essays. (Usually a paragraph or two is enough.)
If you've read a PG essay and found yourself agreeing because you're gradually taken towards a conclusion yourself instead of having it preached to you, you'll know what I'm referring to.
Turns out there is a style for that.
I am no writing expert, but I feel PG's style is (intentional or otherwise) most akin to Classic Style as outlined in the book "Clear and Simple as the Truth" (overview from Robin Hanson of Overcoming Bias here: http://www.overcomingbias.com/2009/03/deceptive-writing-styl.... ).
Most telling excerpt: "Classic style is in its own view clear and simple as the truth. It adopts the stance that its purpose is presentation; its motive is disinterested truth. Successful presentation consists of aligning language with truth, and the test of this alignment is clarity and simplicity."
CaSatT homepage: http://www.classicprose.com/