In the past few months I have been challenged to defend computational philosophy, particularly philosophical modeling. Philosophical modeling, like scientific modeling, is a formalization process. However, instead of capturing real-world phenomena, philosophical modeling captures thought experiments.
The process of encoding a thought experiment in a formal system is, itself, beneficial in the same way as standard conceptual analysis: hidden assumptions are unburied and seemingly simple ideas yield refined notions. But, in a way, encoding is more honest—the process is not satisfied until you reach a syntactic, algorithmic level of explicitness and, once our intuitions are encoded, further light may be cast during runtime. Will your thought experiment crash, loop endlessly, or churn away at intractable problems? Do your assumptions actually lead to your conclusions? Do they yield more than expected? Less? As any programmer knows, what we program to happen and what actually happens can be quite different.
Philosophers often use formal systems to reinforce their arguments, but they are typically piecemeal and the most significant assumptions (often unintentionally) remain outside the formalism. Computers are well suited to explore certain features of complex systems like thought experiments. With a computer, philosophers may encode their thought experiments and systematically explore their features. This cannot be matched by armchair speculations. We are prone to errors in inference and unconscious biases, and simply do not have the time or energy to consider the complex interactions of philosophical presuppositions.
I should note that this methodology is not completely foreign to philosophers. Here are a few examples:
*Daniel Dennett uses cellular automata as a toy world to test our thoughts about evolution, model making, and free will (Darwin’s Dangerous Idea, ‘Real Patterns’ in Brainchildren, Freedom Evolves).
*Grim, Marr and St. Dennis programmed logical and game theoretic systems to illuminate standard philosophical problems in The Philosophical Computer.
*Lastly, in John Pollock’s essay ‘Procedural Epistemology’ (in The Digital Phoenix) he reports about OSCAR, an epistemological agent, to demonstrate his philosophical theories.
In the last essay we are encouraged to “put our model where our mouth is” when it comes to our epistemological theories. Indeed we should, in epistemology and elsewhere.
Posted by John at January 26, 2004 10:40 PMObviously, I'm in complete sympathy with this. But it's worth noting that part of the push for formal systems, originally, was that it was supposed to rescue philosophers from their penchant for "mere thinking" (in Russell's phrase). In fact, Clark Glymour likes to claim that Carnap's Logical Construction of the World was one of the first times anyone had been concerned with explicitly showing that certain effects could be achieved through mechanical, effective procedures. So when you're making the case for computation to your fellow philosophers, you need to explain what it adds that formal systems don't. For that, I think you want to hit the "complex interactions of philosophical presuppositions" note loud, hard and often.
Posted by: Cosma at January 27, 2004 06:37 AMCosma,
Thanks for the advice. Getting anyone to respond to this programme has been difficult.
I think you are right about making my case by, as you say, hitting the complexity note “hard and often”. I have been trying to find a diplomatic way to say "So, all that that armchair reasoning, no matter how creative, just won't cut it-- the problems are to damned complex, or ill-posed". I am a temperamentally moderate person and like to guide people to the answer in a way that will make them understand that it is not a radical departure, but merely an “extension of analytic philosophy by other means” (I think Glymour said that, actually). Perhaps segueing from formal methods to computational methods will show contiguity with a respectable tradition, while also demonstrating important results that imply the need for computational methods?
Posted by: John at January 27, 2004 08:08 AM> Perhaps segueing from formal methods to computational
> methods will show contiguity with a respectable tradition,
> while also demonstrating important results that imply the
> need for computational methods?
It's worth a shot, especially if you can at least talk about
addressing some of the commonly-perceived faults of
formal methods. "See, this lets you go in the
direction you wanted to, even though you thought the way
was blocked." And it'll help if you put things as "there are
some problems which are just too hard even for
you" rather than "your puny brain is no match for k-SAT".
Oh, and read Michael Nielsen's "networking on the network"
posts, and the essay by Phil Agre he's commenting on.
I appreciate something in this post that may have been otherwise over looked.
I appreciate the fact that you used quotation marks “ ” instead of the standard inch marks "".
Just thought you should know. I hate it when people use inch marks. not so much on computers because we are somewhat forced into it. But I saw it on a sign today and it drove me nuts, and when I see it in print I get pissed off that I don't have the persons fucking job.
Anyway. That is all.
Time for another vodka tonic; part of my sleep regiment.
I used to boycott the semicolon. Now I try to use it when appropriate. I'm not sure if I have the hang of it yet.
Posted by: Elex at January 29, 2004 11:18 PMBy the way, I mean absolutely no offense to anyone who uses inch marks on the internet or on computer applications. For one, to most people, there isn't any difference. (I was first exposed to the difference by my boss, a seasoned graphic designer.) For two, the quotation marks don't appear on the keyboard, so how the fuck is anybody going to type them?
I'm just acknowledging John’s use of them. and my appreciation. I’ll stop bastardizing his post by bringing up mundane non-sequitur asides.
Posted by: Elex at January 29, 2004 11:26 PM