Ray Bradbury said, ""

I Really Don't

From a Bayesian perspective, most things have no meaning, isomorphic to an expression of ignorance. In more sexy terms, they ought not to affect one's priors.

Hence, to increase your information content, you ought simply to use the short phrase "I don't know" rather than the equivalent n bits of obfuscation and phlogiston.

How do you know which statements convey no more information than "hell if I know?"?

I don't know.

Posted by: Tom "The Pooklekufr" Treloar at May 31, 2008 14:27 EST | Permalink | comments |
geekery

Comments open but moderated. I reserve permission to kill spammers on sight.

Hemingwayesque-p? t

I will try to use E-prime from now on, an English variant/mental discipline that removes from speech the syntactic object "to be" and its derivatives. It forces one to avoid passive constructions, while at the same time honing a predicate-logic rigor. Hell, it makes for better writing and has a weirdo:geek factor less than Lobjan.

If I can successfully adhere to it, I may work on "lispglish," designed for massively recursive sentence constructions and the fanatical use of slang-compression for large concepts. Would it look like Anthony Burgess had fallen into a vat of parentheses? Who knows?

Hat-tip Ilkka

Update: Yes, I know Korzybski influenced W.S. Burroughs in his craziness, that the Sapir-Whorf hypothesis has no foundation, and that General Semantics offends anyone with their rejection of the Law of the Excluded Middle. But even lunatics may stumble upon a useful method. The problem consists of keeping them from burying the gem of useful inquiry in a mire of bullshit.

Posted by: Tom "The Pooklekufr" Treloar at May 29, 2008 20:59 EST | Permalink | comments |
geekery

Comments open but moderated. I reserve permission to kill spammers on sight.

Bootstrapping my AI Secretary

One of my goals in Project Vinge is to write a LISP task manager that will monitor my progress on tasks, continuously update priorities between tasks and sub-tasks, and advise me on which best to work on.

I've finished the basic data structure and database mechanics. Now I need to implement a filestream parser that can dynamically update the :progress attribute of a specific task, based on lines written since last save. Once that function form is complete, I can move onto the interesting part: a heuristic for weighing the progress and the initial priority to arrive at a value representing the importance/easiness ("true priority") of a given task. I suspect I will have to add some form of ends-means analysis similar to the General Problem Solver, as a mere metric doesn't seem to be a flexible enough knowledge structure. One of the interesting aspects to this, is that I can bootstrap the process as I go along. Each function I add will make it that much more useful to completing it. And each time I can unify things into a macro, I will set my thighs aquiver with excitation.

I've been wondering if perhaps I could treat the various jobs as states in a Markov model, creating a transition matrix with the abovementioned heuristic values. After all, the matrix represents the possibility of moving from one state (job) to another, assuming only that each state is dependant on the previous state. Given the nature of programming, such modularization of tasks is far easier than in physical engineering. Can mechanics "stub" a piston? Someone will yell at me for even thinking about it, but why not wonder?

I will only have completed this project when, decades from now, my virtual avatar is able to sexually harass the great-great-great grand-daughter of this humble program.

Posted by: Tom "The Pooklekufr" Treloar at May 29, 2008 19:31 EST | Permalink | comments |
geekery, hackery

Comments open but moderated. I reserve permission to kill spammers on sight.

Shiny, Books

Got Greg Bear's "Blood Music", Alan Weisman's "The World Without Us", Gardner Duzois' "The Year's Best Science Fiction" and Randy Cassingham's "The True Stella Awards: Honoring Real Cases of Greedy Opportunists, Frivolous Lawsuits, and the Law Run Amok"

Posted by: Tom "The Pooklekufr" Treloar at May 29, 2008 18:24 EST | Permalink | comments |
books

Comments open but moderated. I reserve permission to kill spammers on sight.

Juxtapositions

Minkowski's ?(x), Mathematical constants, and the Online Encyclopedia of Integer Sequences

Posted by: Tom "The Pooklekufr" Treloar at May 28, 2008 20:59 EST | Permalink | comments |
geekery

Comments open but moderated. I reserve permission to kill spammers on sight.

Parentheses For Teh Win

My lisp-fu is now strong enough to implement some of the goals in Project Vinge.

Shiny.

Posted by: Tom "The Pooklekufr" Treloar at May 27, 2008 14:55 EST | Permalink | comments |
geekery, hackery

Comments open but moderated. I reserve permission to kill spammers on sight.

Brainfart: Hypercomputation (an insomniac scattershot)

Computation is defined as what a Turing Machine does. The Church-Turing Thesis in its strong physical form states that all computation can be translated to a TM program. It is generally agreed that all physics is computable, that is, capable of being simulated on a Turing Machine or similarly powerful computational medium.

If a non-computable physical process were ever discovered, that would necessitate the existence of a more powerful model of computation. To this more powerful computational model, the pressing problems of computational complexity theorists would be trivial: P vs NP would be moot, BQP would be a trivial achievement, and Omega-incompleteness would be a special case. It would be like discovering God's hotline: even if you couldn't use it, its very existence would be groundshaking.

If some such super-powerful computational model existed, as required by the existence of physical processes unable to be modelled on the current theory, then I would argue that we are almost certainly in a simulation.Why? Because the very foundation of physics would allow the possibility of easy, even trivial, creation of simulations to arbitrary perfection. What are the alternatives?

This is the kind of stuff I think about late at night when my mind is tired and prone to nightmarish neuroses. As The Isiah pointed out, geeks are precisely characterized by worrying about things that the rest of the population doesn't even consider important.

Posted by: Tom "The Pooklekufr" Treloar at May 25, 2008 23:32 EST | Permalink | comments |
geekery

Comments open but moderated. I reserve permission to kill spammers on sight.

Juxtapositions

Bell's Theorem and the Church-Turing Thesis

Quantum computation and Scott Aaronson's "Shor, I'll Do it"

Loki Jorgensen and Peter Borwein's Visible Structures in Number Theory and Champernowne's Constant

Laplace's Demon and Seth Lloyd's The Computational Capacity of the Universe(PDF)

Posted by: Tom "The Pooklekufr" Treloar at May 25, 2008 16:10 EST | Permalink | comments |
geekery

Comments open but moderated. I reserve permission to kill spammers on sight.

A brainfart

Idea: a "myopic VM environment" plugged into Emacs, that will isolate a function/method, stub called functions, then perform algorithmic analysis. Recursively it can descend to called functions, then their called functions, mapping out the entire functional tree and identifying bottlenecks, even suggesting refactorization techniques. I wonder how this could be done.

I will have to look up opensource deep-program analysis projects. Static program analysis

Posted by: Tom "The Pooklekufr" Treloar at May 25, 2008 12:46 EST | Permalink | comments |
geekery, hackery

Comments open but moderated. I reserve permission to kill spammers on sight.

Where is everybody?

Behold The Fermi Paradox. Consider it theodicy for geeks. In fact, it is orders of magnitude more essential and puzzling than the existence of evil, for it is obvious that intelligent life is necessary for the theologians' hand-wringing to even start.

In that wiki article is almost every scenario you're likely to come up with to explain our silent universe. The sole consolation is that the existence of other sentience is a falsifiable hypothesis. However, the search for such sentience is emminently non-falsifiable ("we just haven't searched the proper frequencies! We need more funding!").

For myself, I am inclined to agree with Ray Kurzweil. Given the explosive growth of technology we have already seen, in a few decades or centuries we will have unimaginable means of communication and transportation. It is highly probable that, unless we kill ourselves off, we will one way or another conquer space in far less time than that envisioned by most science fiction writers. Therefore, were any other intelligent life to exist at all, it would have in a cosmic blink of an eye hurtled itself at the rest of the universe. In less than geological time, a civilization would have progressed to the point where such feats are trivial. Therefore, we would not be having this debate: the evidence would be overwhelming (likely, our system would long since have been nano or pico-reconstructed for computation). So either we are alone, or every single other civilization has progressed no farther than we have. Literally- no more than five decades or so of progress would allow them the means. Why would that be? Which is more likely- a universe empty of all intelligence but us, or a universe in which the laws of physics admit of such a limited exploitation by technology?

If the laws of physics are such that technology is inherently limited to roughly our current level, the implications are enormous. For one thing, those limits ought not to exist. There does not seem to be any physical reason for the impossibility of nanotechnology and quantum computation, and what physical limits do exist seem hardly capable of preventing unimaginable technological progress. Instead of "where are they," one would have to ask, "why does the universe not allow higher technology?" This question is almost certain to lead to metaphysical speculation about our being lab rats in a simulation or some such. At least there would be the consolation that such empirical limits would be rather falsifiable and evident.

What's your take?

Posted by: Tom "The Pooklekufr" Treloar at May 24, 2008 12:12 EST | Permalink | comments |
geekery

Comments open but moderated. I reserve permission to kill spammers on sight.

Juxtapositions

Hasse Diagrams and Low Complexity Art

LP Dual Problem and Expectation-Maximization Algorithm

Raymond Smullyan's theodicical Is God a Taoist and Pascal's Wager

Daniel Keyes' "Flowers for Algernon" and Ted Chiang's Understand

Rene Descartes' Discourse on Method and Daniel Dennett's The Unimaginable Preposterousness of Zombies

Posted by: Tom "The Pooklekufr" Treloar at May 22, 2008 18:13 EST | Permalink | comments |
geekery

Comments open but moderated. I reserve permission to kill spammers on sight.

That's Bullshit

Ernst Mach, the founder of modern axiomatic mechanics now known only for his epynomous speed limit, defined science as the economy of thought. Science was that process by which humans were able to accomplish more with less thinking, less getting their hands dirty with experience, and scarcely any thousand-year intervals of repeated stupidity.

Science accomplishes this economy of thought in two ways: filtering out the nonessential, and using hierarchical levels of explanation.

In an example stolen from Carl Sagan's Demon-Haunted Darkness, imagine the indigenous use of willow-grass as an analgesic. For thousands of years, a wizened old man would paint himself, have a fire lit, brew a potion consisting of dried willow-grass leaves and dung and anything else laying around, and have a sufferer drink it while running clockwise around the fire while being scourged with branches. Miraculously, the sufferer's headache would disappear. Magic!

Consider, now, how much information each medicine man had to learn. He had to be patiently taught the various herbs to select, the proper type of dung to use, the spirit signs in the clouds that determine propitious days for healing, the words of an elaborate supplication to the gods, the minerals to extract for pigment to be used on his face, and myriads other bits of information just to be a general practictioner medicine man. After years of apprenticeship, he was ready to help the sufferings of his tribe members.

What happened if the headache remained? The medine man would have to analyse each element of the ritual for error. Perhaps he put his left foot in when he should have taken it out and shaken it all about. Perhaps he had said a prayer wrong and angered the gods. Any number of components could have failed, and he would be hard-pressed (luckily for his reputation) to identify which.

Science begins with laziness. The medicine man has another appointment; he skips a prayer and the man still feels better. He runs out of a herb or two, and the man still gets better. He lets his apprentice do the ritual, the apprentice screws every other word up, doesn't even hit the sufferer the right number of times as he runs about the fire; the man still gets better. This is almost science. All it lacks is the active application of laziness.

Someone wonders, how little of the ritual is actually necessary? Could I use a pared-down version myself at home and spare the expense of an appointment? Laziness practically compells the person to spend a lot of hard work reducing the ritual to its bare essentials. This lazy observer decimates the ritual, removing not only the half-hour prayers but even the funky drum solo in the middle. He scraps the face paint, the cloud patterns, and wonders why he should bother running around so much. Eventually he discovers that all that is necessary is the potion. With a little more laziness, he discovers he only needs to find one ingredient, the willow-grass, and can skip over the ground eagle-egg powder and buffalo testicle. This is science. A lot of hard work reduced years of education as a medicine man and an elaborate ritual, to perhaps a single sentence such as "find so and so herb, grind it up, and drink it as a tea to feel better." The sexy way of saying this, is that the algorithmic complexity of achieving this goal has been reduced by orders of magnitude.

And so science advances. All it takes, initially, is a particularly lazy or impatient observer; he doesn't even have to be much more lazy than his fellows. Rapidly we discover the world is simpler than we thought. When objects fall, their relative weights don't matter, only their relative air-resistance. Eclipses occur at regular intervals and don't give a damn about wars or rulers. Kill rats and remove the dead to end a plague, no need for sacrificing virgins to appease the gods. Wait out storms and droughts, they'll end even if you don't begin the usual "rip out hearts to funky chanting" ritual. As more bullshit is weeded out, the process accelerates as laziness pays off more and more. Exactly how little effort is required to heal, build a tool, irrigate a farm, or predict an eclipse?

Thus the application of laziness to our world simplifies things. Whole swathes of explanations that once required years of experience or education, now are reduced to a few statements. There is only one problem: the world is big. Even with a huge amount of bullshit ignored, there is still alot of things to take into account.

This is where the second element of science comes in. It is possible, of this reduced set of statements necessary to understand our world, to still treat most of them as irrelevant for any specific purpose. Why bother understanding how this lens works, when you're really only interested in what you're seeing through it? Who cares what new-fangled ideas explain the elements, when you're only interested in finding out why this dye helps a fever? The same laziness and impatience that led to a drastic simplification of the picture of the world, also leads the lazy man to state, "that's someone else's problem." If he wants to know how something works in more detail, he can easily find someone interested in that problem. One man discovers that willow-grass is the only important thing in the ritual, another man discovers that salicylic acid is the only important part of the willow-grass, and another man discovers that chemistry has at least one use when his wild-haired chemist friend explains to him how he can make salicylic acid without foraging.

Over time, willy nilly, this process becomes a culturally inherited mindset. A scientist is someone who, for some reason, views most of the world as irrelevant to whatever he is looking at. The engineer takes this to an extreme, and views even the scientist's explanation as mostly unnecessary bullshit. Over thousands of years we discover that lodestones are not gods and that sparks are easy to make; James Clerk Maxwell discovers that only a page of statements can explain all of this interesting behavior; and an electrical engineering major discovers he can discard all of that for two or three equations small enough to write on a stick of gum the night before an exam.

This way of viewing the world is contagious, not the least reason for which is the proven addictiveness of showing that our teachers did more work than they really had to. Laziness compells people to spend decades of their lives proving that an explanation only requires the study of a few pages, and then compells students to spend nights without sleep to prove that even sleep-deprived automatons can do the exact same task that once required a lifetime of study and a rich patron, as long as they have enough coffee in their system.

In a way, the whole history of science is a reduction of the world to terms that are so simple even a liberal arts student can pretend to understand after having read an article about them in National Geographic or Scientific American.

Posted by: Tom "The Pooklekufr" Treloar at May 19, 2008 04:21 EST | Permalink | comments |
geekery, best posts

Comments open but moderated. I reserve permission to kill spammers on sight.

Flu and Zombies

I've got a flu, so instead of doing anything productive I've been occupying myself with zombie-apocalyptic reading. Max Brooks' "World War Z," A Will to Act, Rise, etc. What better way to wait out a flu, than to immerse oneself in the anarchic hell of a zombie apocalypse? There is an element of schadenfreude, yes. "I may be coughing up phlegm, but this guy just imagined the entire North American continent a ravaged wasteland."

Google "survival tips," and you will find many forums for truly freakish survivalists who seriously contemplate the necessity of hunting stray cats for food. These guys are serious. You know what's even scarier? Google "zombie survival tips." Some of these guys are serious too. If you have written a detailed plan for barricading yourself in the nearest CostCo in the event of a zombie pandemic, it might be time to up your dosage, or at least look into the exciting possibilities of saner obsessions such as bigfoot observations and Elvis conspiracy theories.

Oh noes!

Posted by: Tom "The Pooklekufr" Treloar at May 17, 2008 23:11 EST | Permalink | comments (3) |
life, blog nocturne, geekery

Comments open but moderated. I reserve permission to kill spammers on sight.

A leisurely Sunday

Reading Steven Tanimoto's Elements of Artificial Intelligence using LISP Examples.

[Insert obligatory remark about those not understanding computer science being the slaves of those who do]

For a while now, I've been wondering exactly how I hear. Being half deaf, I am in the interesting position that some of my auditory processing is conscious; but I cannot quite examine the process. I wonder... iterative hidden markov model algorithms operating on the phoneme and word levels, outputting to higher-order pattern recognizers? Would the higher-order markov model be able to reach down and affect the parameters of the lower order one? Clearly something like this occurs: I do piece together fragments of words according to their most probable combinations, then reject those phonetic clusters which yield nonsensical words given previous words. I wonder how I could simulate something like this. Could I at least write a program as good as I am ("Could you pass the sugar?" turning, often, into "wooden passenger?" prior to filtering mechanism)? Does look-ahead exist?

One of the most interesting things about the study of machine learning, is that you are in large part trying to reverse engineer your own mind.

Posted by: Tom "The Pooklekufr" Treloar at May 11, 2008 10:43 EST | Permalink | comments |
geekery

Comments open but moderated. I reserve permission to kill spammers on sight.

This a a post about a self-referential story

This is a link to a highly self-referential story. This is not a link, but the second sentence of this post. If this were the third sentence of the post, it would contain a link to Raymond Smullyan's Planet Without Laughter.

This is the last sentence of this post, and this is the last clause of the last sentence whose last word is not this word, but this word. This sentence wishes to point out that the last sentence was a bit drunk and, as usual, too boastful about its place in this post. This is the sentence preceeding the last, donning a fake mustache and hat, strangling the preceeding sentence. This is. A swarm. Of. Sentence fragments. Which are bothering this sentence for details about the murder. This sentence tries to brush up against a particularly attractive sentence fragment. And fails. Entirely.

Posted by: Tom "The Pooklekufr" Treloar at May 10, 2008 15:12 EST | Permalink | comments |
humor, geekery

Comments open but moderated. I reserve permission to kill spammers on sight.

What's on my mind

Instead of a real post, I will fill my blog vacuum with a brief list of things at the top of my mind lately.

Meg "Teh Kitty" Scottish-name
Algorithmic information theory
Information Theory
Computational complexity
Computability theory
Algorithmic analysis
Machine learning

While the hecticity of life has interfered with programming, I've been studying Smalltalk, Erlang, and Haskell; all languages requiring wildly varying mindsets.

Reading: Donald Knuth's "Art of Programming", Shaun Hedman's "Computability and Complexity", Mikhail Atallah's "Algorithms and Theory of Computation Handbook", Sanjeed Arora's "Computational Complexity: a Modern Approach", F.D. Lewis' "Essentials of Theoretical Computer Science", Raymond Smullyan's "Recursion Theory for Metamathematics", Corman Et Al.'s "Introduction to Algorithms", and the surprisingly good wiki articles on algorithms from the Bellman-Ford to Dijsktra to Quiksort.

Greg Chaitin, laying the groundwork for algorithmic information theory, formulated the information contained in strings as a function of the smallest program necessary to produce them as output. Ever play with ELIZA or Jabberwacky? I recommend you not try to determine how easily one may reproduce the responses you receive in everyday conversations; the results will sadden you. Note to self: perhaps the drastically reduced information content of the elderly ("hot weather, today!") is in linear and inverse proportion to age. If so, should the Turing Test exclude pathological AI's that simulate the elderly, on the grounds that simulating pathological minds is surprisingly easy?

Interesting article by Scott Aaronson: The Mind is Finite.

Information/computability theory is a nodal science, touching on every other discipline, with the advantage over set theory that it does not turn one into Georg Cantor.

Information/computability theory is a nodal science, touching on every other discipline, with the disadvantage that it causes one to, late at night, try to reverse engineer the algorithms underlying the thought processes that cause one to try to reverse engineer one's thought processes about reverse engineering one's thought processes.

What would this sentence be like if it were a kitten?

Posted by: Tom "The Pooklekufr" Treloar at May 06, 2008 21:25 EST | Permalink | comments |
geekery

Comments open but moderated. I reserve permission to kill spammers on sight.

This site is a member of WebRing. To browse visit Here.