Ray Bradbury said, ""

Linky links

Some interesting things I've been reading today:

Scott Aaronson's The Mouth That Does Not Bite: Why Don't Vaginas Have Teeth?

Scott Meyer's How To Share A Horrifying Experience

Bad Hemingway Parody Generator

Python Challenge Riddles

General Collection of Fermi Questions

The Ackermann Function

The Daily WTF's Tales From the Interview

Cliki: Common Lisp Wiki

Christopher Smart's For I Will Consider My Cat Jeoffry

Posted by: Tom "The Pooklekufr" Treloar at January 31, 2008 17:33 EST | Permalink | comments |
discoveries, geekery

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

More Merciless Mockery

How did I go so long without finding Attrition.org's Going Postal archives?

It is not often that you find proof that a stereotype is, in fact, accurate.

Posted by: Tom "The Pooklekufr" Treloar at January 27, 2008 15:16 EST | Permalink | comments |
humor, geekery

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

*Belly Drum*

Megan and I had a wonderful date. We met at the bookstore, had pizza, went to the pet store and gawked at kittens, watched that Cloverfield pseudo-Cthulhu movie, had some coffee and pastries, then spent an infinite length of time cuddling at my place.

Life is good.

Posted by: Tom "The Pooklekufr" Treloar at January 26, 2008 21:29 EST | Permalink | comments (3) |
life

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

Geek Humor Heaven

I've been dicking around with Bruce Schneier's Doghouse exposes of snake-oil crypto products. Commenters provide detailed analysis combined with brutal, ruthless mockery. I love it.

Posted by: Tom "The Pooklekufr" Treloar at January 26, 2008 03:14 EST | Permalink | comments (1) |
geekery

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

Encryption Made Easy

I installed FireGPG, a Firefox plugin that integrates with GnuPG to let you sign/verify/encrypt/decrypt with a right mouse-click. It works perfectly in Gmail, too.

The only problem I have, is its little "save passphrase for session" option. I won't trust that, even though I've looked at the source code to see what happens to this data. Otherwise, it works beautifully.

Yet another example of the open-source community making it that much easier to stick it to the man.

Posted by: Tom "The Pooklekufr" Treloar at January 20, 2008 23:13 EST | Permalink | comments (1) |
geekery

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

Scary!

Check this out: Paterva's Maltego Search.

Enter your name, IP address, whatever. You'll be surprised at how much data you've left lying around that can allow someone to reconstruct your online browsing habits, telephone number, email address, and more.

Posted by: Tom "The Pooklekufr" Treloar at January 20, 2008 20:00 EST | Permalink | comments (5) |
geekery

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

Woo Hoo! Books!

Bruce Schneier's Applied Cryptography arrived yesterday, full of sneaky-secrets. First line: "There are two types of cryptography in the world: cryptography that will stop your kid sister from reading your files, and cryptography that will stop major governments from reading your files. This book is about the latter." *bellydrum*

I also got Malacalypse the Younger's Discordia.

Why these books? My pineal gland only speaks to me in 4096-bit RSA encryption.

Posted by: Tom "The Pooklekufr" Treloar at January 19, 2008 04:38 EST | Permalink | comments (2) |
books, geekery

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

Overheard on the Bus

Crazy Guy: Candy! Sinful candy! Candy is luring us from the Right Way to the Sweet Way! How much sugar do you eat? Our babies are addicted to candy, worse than heroin! Young ladies putting themselves out and shaming themselves with tattoos all over they booties! Son, when was the last time you got a "treat"?! The Sweet Way is distracting us from the Right Way!

Guy, humoring him: What's the right way?

Crazy Guy: The Right Way is not the Sweet Way but the Bitter Way! Sweetness has perverted our minds- look at 9/11 and the Iraqi War 100,000 dead!

Another Crazy Guy: I told you it was all a conspiracy. It was Bush trying to-

Crazy Guy:- gone to warcrimes over candy! Snickers bars and those hollow Easter bunnies!

Guy, humoring him: You're saying Bush orchestrated 9/11 and started the Iraq War to get candy?

Crazy Guy: Why else?!

Guy, humoring him: What about oil?

Crazy Guy: It goes into candy! Look at your gum and chocolate bars and Easter bunnies! I dare you to show me candy without oil in it!

Me: [pulls a box of candybars out of my bookbag] These don't have oil products.

Crazy Guy: [screams]

Posted by: Tom "The Pooklekufr" Treloar at January 18, 2008 16:46 EST | Permalink | comments (5) |
life

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

Viterbi Hidden Markov Models

I've been playing around with Viterbi Hidden Markov Modelling. Here's a more in depth explanation of it.

Essentially, the problem is to determine a hidden parameter underlying visible behavior, both of which depend only on the previous state. From this, it is possible to determine the conditional probability of a given sequence of observations, as well as the hidden parameters.

Here's a fun example. Suppose you wish to determine whether a woman has PMS or not given her observed behaviors. All you need is: a)the proportion of time each month spent in PMS, b) the probability of each behavior while in both PMS and normal states, and c) the probability of changing from one state to another.

Here's how you'd represent these in Python (numbers guesstimated):

states = ('PMS', 'Normal')

observations = ('smile','sing', 'cuddle', 'tap_feet', 'yell', 'complain', 'clean', )

start_probability = {'PMS': 0.15, 'Normal': 0.85}

transition_probability = { 'PMS' : {'PMS': 0.4, 'Normal': 0.6}, 'Normal' : {'PMS': 0.3, 'Normal': 0.7}, }

emission_probability = { 'PMS' : {'smile': 0.01,'sing': 0.01, 'cuddle': 0.01, 'tap_feet': 0.1, 'yell': 0.25, 'complain': 0.50, 'clean': 0.21,}, 'Normal' : {'smile': 0.39,'sing': 0.04, 'cuddle': 0.3, 'tap_feet': 0.01, 'yell': 0.01, 'complain': 0.05, 'clean': 0.2,}, }



So, in English: there is an 85% chance at any given time she is not having PMS, a 60% chance she will return to normal while she's on PMS, and she is much more likely to complain while on PMS. These numbers and behaviours are pulled out of my ass, but they illustrate the point. The beauty of Markov models is that they allow you to model high-entropy signals (those which are unpredictable) such as speech; they work precisely because the signal contains information.

So, assume you've spent enough time with her to get a rough idea of behaviour patterns (a heuristic neural net could do it, but that's another post entirely). Now you plug the data into the following program:

def forward_viterbi(y, X, sp, tp, ep):
   T = {}
   for state in X:
       ##          prob.      V. path  V. prob.
       T[state] = (sp[state], [state], sp[state])
   for output in y:
       U = {}
       for next_state in X:
           total = 0
           argmax = None
           valmax = 0
           for source_state in X:
               (prob, v_path, v_prob) = T[source_state]
               p = ep[source_state][output] * tp[source_state][next_state]
               prob *= p
               v_prob *= p
               total += prob
               if v_prob > valmax:
                   argmax = v_path + [next_state]
                   valmax = v_prob
           U[next_state] = (total, argmax, valmax)
       T = U
   ## apply sum/max to the final states:
   total = 0
   argmax = None
   valmax = 0
   for state in X:
       (prob, v_path, v_prob) = T[state]
       total += prob
       if v_prob > valmax:
           argmax = v_path
           valmax = v_prob
   return (total, argmax, valmax)

##################################
#Output Function
##################################

def example():
   return forward_viterbi(observations,
                          states,
                          start_probability,
                          transition_probability,
                          emission_probability)
print example()

Bam! You've got a way of tracing her moods, predicting trends, and hopefully preventing being dismembered. This is a very simple model with no feedback mechanisms, yet when the algorithm is embedded within a larger structure it can become a very powerful means of extrapolating patterns in digital signal processing, natural language recognition, cryptanalysis, weather forecasting, etc.

I had an idea to design a program based on the Viterbi algorithm, using backpropagation neural nets, to model manic depressive cycles. Each day the user would log his or her activities, moods, diets, etc. Given this data, the program could learn underlying patterns behind mood swings, and allow the patient to predict them, plan around them, and tailor treatment. A psychiatrist in a box.

Posted by: Tom "The Pooklekufr" Treloar at January 15, 2008 19:04 EST | Permalink | comments (4) |
geekery

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

Cryptogeekery

Heh.

Posted by: Tom "The Pooklekufr" Treloar at January 14, 2008 04:38 EST | Permalink | comments |
humor, geekery

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

Blogroll Play

I'm going to play around with my blogroll and add a bunch of sites I've discovered.

Here's one I just discovered yesterday: Qwantz Dinosaur Comics. Pure genius.

Since I am a geek, I'm going to write some code to get this done. Basically, I want to grab links from my .htm bookmark list, strip them of extraneous attributes, and jam them into my template. This can be done with a perl script. But, I'm lazy enough to spend lots of time trying to make a program that will sort them and filter out links that are not already present, and then place them right where they should be.

Posted by: Tom "The Pooklekufr" Treloar at January 13, 2008 19:08 EST | Permalink | comments (1) |
blogging, discoveries, geekery

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

My Work This Weekend

I'm studying this Python neural network library. Very¹⁰⁰ cool.

Here's a clear introduction to neural nets specifically implemented in Python.

I would write more, but I'd rather get back to chugging coffee, chainsmoking, and poring over the mechanics of backpropagation. Maybe I'll post a series of dissections of the mathematical structure of artificial neurons.

Posted by: Tom "The Pooklekufr" Treloar at January 12, 2008 02:16 EST | Permalink | comments |
geekery

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

Two Slices of Life

Sir Abu Horsetassle the Infinitely Benevolent Toward Small Children and Animals:[lighthearted smalltalk]

Me: Hey, nice pants.

SAH: Say my pants are nice one more time!

Me: They look soft and somewhat fluffy, and well-tailored. If I had pants like that, I'd be unstoppable.

SAH: [pushes face near mine] Not today! You don't want any of this!

Me: What, you don't like it when someone compliments your pants?

SAH: Say it again and I'll break your goddamn face!

Me: All I said was that your pants look soft and are well tailored.

SAH: [feeble slap at head]

Me: Ok, my pants are better than yours.

SAH: [grabs my jaw, menacingly waves fist at me] One more word!

Me: You're too touchy about your pants.

SAH: [stands back, clenching jaws, with face muscles twitching]

Dostoyevskian Sadist Bookworm Coworker: Man, he went psychotic in less than a minute![almost passes out laughing]

Me: You can't have your cake and eat it too, Abu.

While clocking out:

Desk Monkey: Someone told me Sir Abu Horsetassle beat the crap out of you.

Me: [in low voice] He touched me.

SAH: I DIDN'T TOUCH YOU, MOTHERFUCKER! STOP IT WITH THAT GAY SHIT! [launches himself across room at me]

Me: You did hold my face.

SAH: [while grabbing at me] Not in a gay way!

Dostoyevskian Sadist Bookworm Coworker: [bursts out laughing, begins to choke]

Me: Homoerotic or not, you still touched me. And you're touching me right now.

SAH: [tries to throw me out the door]

[ten seconds later]

Dostoyevskian Sadist Bookworm Coworker: Man, Abu, you have manic depression. You go from giggling to white-hot rage in seconds. You gotta get some medication.

SAH: [laughs] I do have a short temper.

Babbling Coworker: One minute you're laughing and buying people donuts, the next you're lunging at their throats like a friggin ferret. You're worse than Oprah.

SAH: Heh, my girl says I'm bi-polar too. [smiles]

Dostoyevskian Sadist Bookworm Coworker: Of course she does. You're nuttier than a squirrel's asshole.

SAH: Anyone got a cigarette?

Me: Here. [offers cig] I'm like Mahatma goddamn Gandhi. You see how magnanimous I am?

Babbling Coworker: Abu, at least today you actually acted on your psychotic rage instead of bluffing like an impotent chihuahua.

SAH: I'll beat your face in! You don't want to see what Lucifer's got!

Dostoyevskian Sadist Bookworm Coworker: See, that's what we're talking about.

Me: That's perfect. When you feel the Frenzy coming on, just alert us that "Lucifer is coming!"

Posted by: Tom "The Pooklekufr" Treloar at January 11, 2008 18:28 EST | Permalink | comments (1) |
humor, scruffy work

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

General Life Planner

First, WANT. Had to get that out of my system.

Humans are notoriously bad at planning our lives. We restrain ourselves with unnecessary limitations and irrelevancies, hampering our dreams. What goals we do have despite this ingrained doubt, we further restrain in their necessary sub-goals. And that's not even including our inability to easily assimilate knowledge of our environment. We are plagued with doubt and ignorance.

I'm working on a General Life Planner, based on the General Problem Solver and Cyc, that is, a means-end analytic inference engine combined with a knowledge database, able to formulate algorithms for the optimal path to one's goals.

The idea: the program accesses a database of commonsense knowledge ("college education increases potential income" "jobs require experience") and a database of statistical knowledge ("IBM requires X years of experience and these skills"). A glance at any bookstore will indicate that plenty of such advice exists. This is actually an achievable goal for an expert system, as expert systems have proven themselves extremely capable at specialized knowledge (for instance, almost all EKG's and 10 percent of all pap smears in America are diagnosed by an AI). I'd be terribly surprised if you  actually needed to accumulate a considerable fraction of CYC's massive database to get enough information for this purpose.

The user would describe initial conditions, and the goal. Again, this ought not to be not very difficult considering the narrow domain of relevant factors. The program could then work out the necessary steps to reach that goal.

There are many ways to optimize the result, perhaps using feedback from experts in resume-building, education, real-estate, surveys of those who have reached their position, etc. At first, the program could help with only large decisions ("How do I get a job at Transmeta?"), but with more knowledge, finer-scale decisions could be delegated to it ("should I take this course or another course?").

I think this is possible, because the program only has to consider a functional representation, it wouldn't have to bother with small details. Consider the algorithm you use to take a shower. Only Monk would describe it at the level of scrub-direction and lather density. The program need not take into account everyday actions (yet) toward the goal, only general sub-goals. And as it is refined, more and more detailed  plans could be formed.

Posted by: Tom "The Pooklekufr" Treloar at January 08, 2008 00:19 EST | Permalink | comments |
blog nocturne, geekery, vinge

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

In a World Where CGI is Rampant...

I wonder. Suppose a time-dependent function C(t) of computational resources used in the movie industry. For a movie like "I Am Legend," what value of C would it represent? More CG effects than every movie made before 1990? Later? Could it actually have used more CG than even an animated movie like Shrek?

I'll look into it. How long before a "non-animated" movie requires more computing capacity than everything made before 2007? Imagine if the opening trailers alone used order of magnitude more resources, for a fraction of the cost of a Pixar movie.

Posted by: Tom "The Pooklekufr" Treloar at January 07, 2008 19:02 EST | Permalink | comments |
geekery

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

Uncomfortably Pregnant

The Isiah: Tom's leaving at 9:50 for the 10:00 bus.

Shira the Mommymonster: [eating licorice and bananas] Why are you telling me?! He's sitting right there!

The Isiah: I already told him, I thought you'd like to know what's going on.

Shira: I'm not his mother! HahahaHAHA! I don't need to know everyone's schedules! HAHAHAHA!

The Isiah: I thought you'd just filter it out, not start yelling.

Shira: I wasn't yelling. I just don't need to know our so- friend's schedule!

Me: Agh. I feel like an old man or a kid. I'm right here, you know.

Isiah: You don't want to know when he's going home?

Shira: I don't need hehehe to know! He's what, 24?

Isiah: What does his age have to do with keeping you informed?

Shira: Oh, you're the one keeping me informed about him and everything else?

Me: I'm 23.

[snip]

Isiah: And that's the Andromeda Galaxy [points to planetarium program]

Shira: [nods head, taps feet, and tries very hard not to show interest]

Posted by: Tom "The Pooklekufr" Treloar at January 07, 2008 15:30 EST | Permalink | comments |
life, conversations

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

Wee Morning Shower Thoughts: Self Aware Memes

I just had an interesting idea while in the shower: a self-aware meme.

Consider all the memes floating around the blogosphere- book memes, iPod memes, nonsense memes, story memes, etc. Ideally, they would grow exponentially: after 10 generations at least 1024 descendents should exist.

In practice, though, they die out rapidly, but not so rapidly that they are still manageable by their creators. After about three generations they become too unwieldy to track. Despite our best intentions, they grow too chaotically and little trace is left.

It ought to be possible to create a means of tracing a meme's progress, graphing it, as well as allowing it to actively propagate itself.

One way to do it would be via XFN rel attributes. Each hyperlink to predecessors would contain information capable being backtracked and mapped into a relationship topology. Each participant could use a spiderbot to determine his genealogical relationship to the originator. This would be crude, yet possibly serviceable.

A better way would involve the transmission of a spiderbot/autopinger along with the meme, in source code. Upon accepting a tag, the participant could simply add the permalink of his instance of the meme to the code, run the python or perl script, and have each descendent (or maybe only the originator) receive a trackback containing an upticked number specifying his genealogical distance. The spiderbot could then backtrace the evolution of the meme and map it out in graphical format.

A more interesting way would involve a version of a pagerank algorithm to determine the density (inbreeding) of the meme. A spiderbot implementing a pagerank algorithm could determine the range of the meme. A fine-detailed relationship topology could be constructed. With this information, it would be possible for the meme to undertake genetic programming. Simply put, also transmitted with the meme would be a perl script for "mutating" it, changing some of the questions, or the number of people to be tagged, etc, at specified values of connectedness or pre-selected generations. The meme would diverge into competing variants, competing not only in the attractiveness of their content but also in the effectiveness of their ability to mutate. This already happens, but at a slower rate, constrained by our lack of imagination. Like a true mind-virus, strains of the meme would actively spread, tailoring themselves in a dumb manner to the peculiarities of human minds and with each iteration becoming more and more irresistable to those who participate.

Cybernetic memes capable of overrunning the blogosphere. I'll add this to my projects.

I've been slowly making my way through Washington University's excellent Artificial Intelligence code archive. Not only is it an excellent example of what we can already do, it is an excellent example of just why we must work harder. I could spend years dissecting the code, and still not have found any real analog to human intelligence.

Posted by: Tom "The Pooklekufr" Treloar at January 06, 2008 04:36 EST | Permalink | comments |
memery, blog nocturne, geekery, best posts

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

General Problem Solver: Too Neat

I've been playing around with various versions of Newell & Simmons' General Problem Solver. Here's a simplified version, and here's a rather more complicated one.

These are very fun Neat experiments, but as Minsky pointed out, entirely on the wrong track. We know to a very high degree how some portions of the brain work, particularly auditory pattern recognition and sub-processes, and it does not use such "expert system" databases or crude symbolic representations.

Damn, I need to learn more machine-learning. The GPS looks awfully formidable, but it and its children arguably retarded AI research for decades. Hopefully I won't get caught in that trap.

Posted by: Tom "The Pooklekufr" Treloar at January 06, 2008 00:18 EST | Permalink | comments |
geekery

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

Greasemonkey on Motime

I've been playing around with the Firefox Greasemonkey extension (and userscripts in general); it allows you to customize webpages exactly how you'd like them to work. Almost any changes you want, you can make: remove ads, alter the function of forms, adjust content, add features, etc.

I'm going to work on some Mo'time hacks. In Opera and Konqueror, the Gecko script doesn't work, giving it no WYSIWYG editor. I'm going to build one and incorporate it; if it works well my fellow Opera/Konqueror users won't have to bother with html formatting.

Another hack I'm going to do, is to turn the digest window into a sidebar- you'd be able to see in real-time all the comments and responses you want, as well as who's online.

It would be very cool to create an onmouseover menu that displays new comments on your blog right from the home page, but I fear the volume of spam comments would make it useless.

Besides these, what else should I work on? Comment!

Update: just got Platypus extension. Woah. I can play around with website form/content by drag-n-drop.

Posted by: Tom "The Pooklekufr" Treloar at January 05, 2008 17:56 EST | Permalink | comments (1) |
geekery, hackery

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

Headachey RSA

While my migraine retreated back a bit, I played around with my encryption programs.

Here are my public keys.

You know that primal joy you used to get when playing hide-and-seek as a kid, curled up in a hiding place? Encryption gives you that same joy. Something deep within the human soul loves and embraces secrecy, privacy, and that squirrelly instinct to hide things. Like Chuck Norris' beard, it can be used for good or evil.

Posted by: Tom "The Pooklekufr" Treloar at January 04, 2008 22:31 EST | Permalink | comments (1) |
geekery, hackery

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

Not a Horn Can Beep but I Do Not Wish the Driver Dead

After spending most of the day in a dark and quiet room, I finally got the migraine under enough control to get up, make a pot of coffee, and eat something.

You know, if it weren't for the pulsating pain, a migraine would actually be fascinating. During the prodromal phase of a migraine, I experience extremely high-resolution, long-lasting after-images. It's probably the closest I can get to eidetic memory, glancing at a book, closing my eyes, and seeing the words in bright negative colors. Its prenumbra also affects my auditory sense: I lose the ability to filter/prioritize sounds, giving me an uncomfortable feeling of omniscience as every single sound passes through my consciousness, from heartbeats to distant carhorns. I wonder if that how autistics like Rainman experience the world, without any means of filtering the signal from the noise of their senses (William James' "blooming, buzzing confusion").

At the current rate of the rate of technological progress, the human brain will likely be reverse engineered to a high degree by 2030 or so. I can't wait to see how migraines actually work. What exactly is their function and cause? Why do they interfere with so much sensory processing?

Posted by: Tom "The Pooklekufr" Treloar at January 04, 2008 17:01 EST | Permalink | comments (1) |
life, introspection

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

Ugh

Migraine. Besides the annoyance of the pounding headache, sensitivity to noise, vomiting, and loss of coordination that made me go home early today, I feel so biological. When I feel better I'm going to purposely write some overloaded C++ arrays and cause a core dump out of spite.

Posted by: Tom "The Pooklekufr" Treloar at January 04, 2008 10:26 EST | Permalink | comments |

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

Late Night Mad Scientist Thought

The P300 signal is an evoked neurological potential produced involuntarily upon recognition. It's easily reproduced and highly noticeable, making it one of the most promising neurological feedback/interface signals.

Currently, cochlear implants do not work with auditory nerve damage. They cannot as yet output directly into the auditory cortex. Incidentally, the structure and function of the auditory cortex has been reverse engineered to a high degree of accuracy.

I have total congenital deafness in my right ear, and partial deafness in my left, resulting from damage to the auditory nerve. I know from long hard experience that this damage produces noise, garbling speech and forcing me to semiconsciously resolve language. The low-level speech processing regions are inadequate to the task of filtering out the noise and recognizing the signal.

Here's the mad scientist idea: suppose someone hooked you up to an EEG and fMRI and played back to you a recording of someone reciting a text in a thick accent or with a bad speech impediment. Basic digital signal processing would allow one to identify which components of speech were identified as such, which components are identified as noise, and (due to the incredible redundancy of language), filter out the noise. It would be possible to, by this means, filter out very thick accents. In effect, the machine would be taking over much of your auditory processing. This is nothing compared to its possible applications.

Think of the uses:

Speech impediments. You could apply the processing to one's own speech with appropriate feedback devices, allowing for the correction of speech impediments and voice-training in a matter of days rather than months or years. You could implant a processing device outputting directly to higher-level processing regions, eliminating even severe speech impediments.

Partial nerve deafness. You could map out which phonemes are most difficult to identify and resolve, and patch them into higher signal processing regions. You could identify the noise produced by transmission along the damaged nerve, and filter it out. The machine will be compensating for the failure of biological processing to cope with the nerve noise. Even more, we need not worry about the problems associated with speech recognition software, because there would be an objective basis for feedback necessary to heuristic algorithms.

Babelfish. Yes. There is no reason whatsoever why it would not be possible to create a feedback process which will enormously accelerate the process of acquiring a language, mapping its phenomes into existing patterns. Electronic digital signal processing, remember, is three million times faster than biological processing. Real-time translation implants are fully possible.

Now I must go to sleep.

Posted by: Tom "The Pooklekufr" Treloar at January 04, 2008 01:28 EST | Permalink | comments (2) |
blog nocturne, geekery

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

As I Walked Out the Bank

"Mister mister my baby's crying I need money for formula can you spare a dollar five dollars maybe ten I need money how about a twenty my baby's hungry," the rotten-toothed woman chanted, having somehow smelled that I had money on me.

"I've only got my busfare on me, sorry."

Whereupon she threw her arm over my shoulder and reeked something into my deaf ear.

"What did you say?" I asked, simultaneously guarding my wallet.

"Want some pussy? I'll fuck you for a dollar." she reeked into my good ear.

"No, thanks. Sorry." I, for some reason, apologized for not taking her up on her offer to give me a dollar's worth of venereal diseases.

Why the hell did I feel the need to apologize? What makes us apologize to total strangers for not being as sick as they are?

"Sorry, I don't know where Buffalo Bill's Bukkake Bar is."

"Sorry, I still need my kidneys."

"Sorry, but you can't have my wallet. Or my socks."

Posted by: Tom "The Pooklekufr" Treloar at January 03, 2008 18:34 EST | Permalink | comments (2) |
life

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

A Pascalian Wager

I've adopted a devious Pascalian Wager on the idea of a Technological Singularity. If Kurzweil is right, by the time I am in my mid-40's the world will be unrecognizable. If Kurzweil is wrong, the world will still be unrecognizable, if only a little more bloody. Whatever happens, I do know that it would be suicide not to keep up with Moore's Law. There is of course the obvious additional upshot of ensuring that I am the geekiest person in any group. After all, I really would not like to meet those geeks in person.

Posted by: Tom "The Pooklekufr" Treloar at January 02, 2008 17:22 EST | Permalink | comments |
geekery

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

Project Vinge

By the end of this year, I will have delegated at least 25 percent of my tasks to programs.

Life is too short to waste as a primate.

Here's a list of projects I'm working on:

1. Expected Value program utilizing Bayesian inference, that will allow me to decide among my various projects.

2. A weighted minimax algorithm for wikisurfing, heuristically compiling a trek with a high chance of being interesting based on previous search paths.

3. An improved meta-RSS feeder that will decrease my blogreading time and increase its breadth

4. A money-management program that will enable me to budget wisely, from small decisions to large ones.

5. A personal assistant program able to search through my output and extend my ability to recall. Think Googlechat's search feature.

6. A program that will enable me to create recipes out of available ingredients, construct shopping lists, and tailor itself to nutritional needs.

7. A test program that will sporadically test my abilities of information retention, absorption, and calculational prowess. I want to map it and see general patterns, and be able to work on my weaknesses. I want, furthermore, to be able to act on these patterns and tailor my life to optimize them (circadian rhythms, diet, coffee intake etc).

8. Increase of bandwidth. I want a program that will estimate the bandwidth I consume each day, both online and in books, and map it out. I want to find out exactly how much information I am capable of taking in each day. And double it.

9. A program capable of sifting out important information from text documents based on keywords. I want to be able to plug in a file, and have it pump out only those parts I am interested in. Then I want to make it work in realtime while I websurf.

10. A real-time program capable of running on a cellphone or other small device, that will interface with these other proects over some channel. I want to be able to program from _anywhere_.

11. An avatar. A bot that will blogsurf, log into Gmail, and select/download correspondence. A bot even capable of holding Rainman-style conversations on Gmailchat with my friends (a cockroach-intelligent analog of answering machines).

12. Acquiring encryption. I want to have a supply of anonymous digital cash, encrypted bandwidth, and encrypted memory, by the end of the year.

Posted by: Tom "The Pooklekufr" Treloar at January 01, 2008 21:07 EST | Permalink | comments (4) |
geekery, hackery, best posts, vinge

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

My Single New Year's Resolution

Is to increase the bandwidth at which I communicate and learn. Think about it a little.

A sub-resolution is that I will have, by year's end, delegated at least 25 percent of my routine mental tasks to (my own) programs.* Considering all the fun and challenge this means, my resolution kicks the ass out of your's. Hah.

This year will go by fast, and I intend on keeping up with it.

* Decision-making algorithms, mostly. I want to write applications to remedy my innate Bayesian inadequacies, and then add feedback mechanisms to allow for their use on shiftier data and larger-than-I-can-handle datasets. I hope to create a bullshit detector that can fit on a cellphone. I've more than got my work cut out for me in just learning enough algorithmics and machine-learning processes.

Ernst Mach once defined science as the art of economizing thought; science progresses by the gradual elimination of inessential reasoning. This improvement has worked wonders- a physics grad student now can understand processes that would have taken Newton decades. And with the exponential pace of advancement, Norman can understand his material better than he could possibly have only a decade before. Yet, but for a few toys, we have not yet seen this economy of thought applied to the everyday world. How much thought is expended in the average day, that could be automated? What is the limit to the automation of thought? I'll know the answer roughly by the end of the year.

Posted by: Tom "The Pooklekufr" Treloar at January 01, 2008 15:44 EST | Permalink | comments (3) |
introspection, geekery

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

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