Why seventeen seconds of film "only" comes out to about two pages

Last night, I sat down and picked apart one of the more popular character moments in the Avengers movie. The clip was seventeen seconds long, and the analysis was several times longer than what was actually written in the screenplay.

Apparently this amazes people. What's amazing to me is that I managed to keep it down to just a couple of pages. Human interactions are fractal in nature. Everything people say and do is supported by personal experience and cultural context, which in turn is supported by cultural and familial history, which in turn is supported by the development of civilization. To drastically oversimplify: Everything is related, somehow, to everything else.

The problem with getting this down on paper is that after a certain point, it starts to lose linear structure. The trails of context start to shoot off in a lot of different directions, and rapidly diverge to the point where trying to hit them all in one explanation would make me sound somewhere between schizophrenic and horribly ADHD. The information is laid out spatially in my head -- planar for the context attached to each person, volumetrically when you include the various layers of subtitles -- but since I have no way to write blog entries holographically, and you have no way to decode them, it would be virtually impossible to keep it all straight.

Other information is stored similarly in my head, which by normal-person standards apparently makes it a right mess in there. There is also the phenomenon that Eyes calls "paralleling" to take into account. She handles profiling for criminal cases and media figures, and her references to it are strictly in the context of recognizing that someone she's analyzing is similar enough to someone she already knows for her to use the familiar person's behavior to predict the behavior of the unfamiliar one.

I do it with pretty much everything. I consider their "shadows" when comparing and contrasting them. In physics, when you have to draw a graph of something that happens in n dimensions when is greater than the number of dimensions your medium makes available to you (two, in the case of paper or chalkboards), what you have to do is essentially flatten the image. You pick a point of view and graph one of its dimensions against a second dimension, ignoring the others. (Abbott's infamous book Flatland has extensive explanations of this, within the framework of a story, if you need an extended treatment of it.)

The trick is to choose your viewpoint so as to make your life as easy as possible. A 2-dimensional projection of a sphere is always a circle. A 2-dimensional projection of a cylinder can be either a rectangle or a circle, depending on where you choose to "stand" when you look at it. From the proper point of view, retrovirology bears a striking resemblance to memetics, to me; knitting patterns look like an analog television signal; and people usually cast subtle variations of a fairly small number of distinctive shadows.

The other major part of it is simply observing enough detail in order to construct an accurate projection from any given POV. I honestly don't know to what extent this can be learned. I know that my ability to put together discrete pieces of data is an acquired skill, but my ability to fit things together quickly enough to do it while I'm having a conversation depends very heavily on the fact that my head is packed full of otherwise completely useless, interconnected cruft -- so far the only thing it's done for me other than let me pretend to be Sherlock is get me banned from playing Trivial Pursuit with more than one group of people, on the grounds that it's not fair to the other team. I've never not been able to do this, although I used to get myself into ungodly amounts of trouble by mistrusting my people-reading and assuming that whatever I thought was automatically wrong.

There may be a hard limit on bandwidth somewhere, but I have no idea what it is. So far, it's been limited mostly by how fast I can spit these things out. (Presuming I'm sober; when I'm sloshed I'm worse for obvious reasons, although using vodka to knock 30 points off my IQ still leaves me the ability to do calculus and read JAMA papers.) I'm certainly not perfect at it, but I enjoy involving a certain amount of showmanship, and confirmation bias mostly works to bolster my reputation.


  1. So, how do knitting patterns resemble analog television signals?

    Oh, and hi, it's good to hear you - I've been reading but not commenting, mea culpa.


    1. You're building a picture up line by line out of a finite number of non-square pixels. Admittedly, TV uses a horizontal retrace gap and knitting progresses boustrophedonically, but the comparison is non-trivially helpful when working out new patterns, especially if you used to occupy yourself during your hideously boring computer classes by doing pixel art in Paint.

      As it happens, the height-width proportion of stockinette stitches are about the same as those used by old Atari game consoles. I have a pattern for Space Invaders socks that works out surprisingly well.

    2. We also have, in analog television systems, RF interference manifest itself as a herring-bone pattern.

      Now then, in deciding what image to portray to a viewer, a lot more than just 'point of view' is taken into consideration. Field of view, rate of change of that field, shutter angle, focus, apparent depth-of-field, the size of the recorded image, the relative angle and/or curvature of the recorded image, and a whole slew of other parameters (really, I could probably go on for a while just listing them) change the image that the viewer is presented with in interesting ways. To see a good example of a variety of parameters changing at once (and producing a cool effect in the process), look for a video of a "Hitchcock zoom".

    3. I'm familiar with the aforementioned A/V stuff. I have about three-fourths of an electronic media production degree that I had to quit due to lack of money and the general idiocy of the university (huge new building which was so badly designed that no wireless signals, WiFi OR cell phones, worked anywhere within it, half the computer labs weren't running, classes cancelled right and left -- right mess, that was). One of my sociological focuses is also media and technology, i.e., I have successfully gotten research credit for writing about video games. Another thing I'm notorious for doing is watching things and pointing out 'that was shot on film', 'that was done digital', 'that was animated in Flash', 'the restoration team missed a lot of Quad scratching', 'the BBC had crappy vidicon cameras in the '70s', 'the transfer is interlaced and has a shitload of flicker', etc. I try to keep a sock in it most of the time, but one of my roommates is in the bad habit of encouraging me.

      AFAIK, in this case, Avengers was shot 35mm with proper cinematic lenses, with a digital intermediate for SFX work. The 3D version was done in post. (A lot of people hated that. They couldn't do 3D realistically, so they chose to go stylized, to what I thought was excellent comic-panel effect.) The clip I linked to looks like it's been through multiple layers of questionable MPEG encoding before it was bunged through Flash, hence why it has all the picture fidelity of a bootleg three or four VCRs down the distro chain. tumblr is apparently in possession of a clean if lo-res copy somewhere, judging from their plethora of animated GIFs.


Post a Comment