Karl: Postmodern Computer Technology

Directions. Let's say you need to drive to 123 North Main Street, and you've never been there before. Nowadays, you might type the full address into the computer and let a powerful web mapping utility draw you a map. Still, when you get to that block on Main Street, you will need to inspect the number posted on the front of each building, find the one that says "123", and only after you have arrived, and verified that it is indeed your destination, will you be satisfied. Then, you will probably never use that number again - you may well forget it completely. Why? Because every time you need to drive to that location, an entire bundle of emotional, associative, visual cues will flood into your consciousness and guide you there. The number will be superfluous. How would we get a computer to do that? Under the current approach, we just wouldn't. We would rely on the numerical coordinate system to derive the location "fresh" every time it is needed, and use automated Global Positioning Satellite prompts to direct our vehicle there. Historical note: At one time, not that long ago, we did not use numbers as addresses. Numbering city buildings for easy locating was introduced by a communications innovation - delivering mail.

Page numbers. Let's say you read a good book last week, or last month. If I ask you to tell me what happened in the book, is it likely you will use page numbers? Will you get the order of events in the book right? No, and yes. A computer would not easily get the order of events right, without some indexing system like page numbers, whereas we will absorb a bundle of subtle cues - "after that's" and "just then's" and "before that happened's" - which will be embedded in our emotional sense of what actions were described in the book. In fact, if the author chose to scramble the time sequence, we will readily describe the action in proper order anyway, so strong is this skill. However much we may use page numbers in certain very particular circumstances, like telling someone else where to look for a passage, they are optional to successfully understanding the events described. Indexing to achieve almost any sense of location and order is not optional in a computer program. Historical note: We did not always have page numbers. They were introduced with the breakthrough technological advance of printing - as a production page assembly technique.

We tend to think of all common numerical indexing like street addresses and page numbers as indispensible. Our computer programs can't work without them, or related indexing techniques. But they serve better as an example of how even our pre-computer technology has been historically changing us for centuries, not how we have been able to make that technology "think like we do." In our current paradigm, the meaningful - emotionally grounded - aspects of task performance are viewed as the "fluffy human stuff" that comes after the computer has done the important numerical job of "locating the address" or "finding the page requested". In any new system, the computer must model exactly those "fluffy" attributes of cognition, and produce the numbers, like we do, as arduous and secondary overlays to normal functioning, useful only when the outcome we want can't be easily "pulled out" in the primary and natural way.

Computers vs Phenomenology Part II

It's-On-The-Tip-Of-My-Tongue-ness. There may be a technical word that describes this, but I can't think of it right now. It's the feeling a person can easily get that they know the specific word to express something - for certain, the word exists and is somewhere "in my head" - but I cannot for the life of me produce it. Then, twenty minutes later, after they have started a totally new and unrelated task and the entire situation that demanded the word in question is past, they remember it. By contrast, when a computer does not produce something, you can be certain it does not have it at all, or that you have not framed the correct method of asking for it. A computer will not wait twenty minutes, change programs, and then produce the answer to a request from an earlier program, which is no longer pertinant. Only humans do that. But most importantly, only humans experience the sense that something is inside them, but cannot for whatever reason be properly dislodged so it can rise to the surface - to the tongue, to get past its tip, and be pronounced out loud.

I-Can't-Get-It-Out-Of-My-Head-ness. I listened to a musical or an album a few days ago, and now I can't get the tunes out of my head. They keep popping up into my consciousness, and before I know it, I'm humming something to myself that I had no intention of even thinking about. And then, the more I try to get rid of it, the more it comes back when I'm no longer payig attention to it. What is happening here? Are we better off because a computer would never do such a thing? Or is it an indicator that computers and their programs simply don't work on the same wavelength that we do?

It would not make sense to program a computer to either arbitrarily forget a known fact, or to recall and play back a recently stored pattern for no apparent reason, and without being asked to. Yet the human consciousness performs these actions with almost ritual regularity. The answer is clearly not that we need to add such quirks to our machine simulations, but that we need to program the computer in such a way that these distinctive habits of cognition simply "fall out" as natural consequences.

Problems with Categorization - the Grade Level Disaster

These may seem trivial as examples of categorization problems, but the the argument of this paper is that they are unavoidable given the objectifying worldview of current programming technology.

A simple database program tracks children and assigns them to a grade level - that is, the grade of the class they are assigned to, First Grade, Second Grade, etc. and not any particular grade on an assignment or test, which would be a related but separate set of problems. It is interesting to note that in practice, most such systems now use a grade level with a one-tenth precision: 3.1 for the level just after the first month or so of the Third Grade school year, 3.9 for the next to last, etc. This is arguably about the simplest assignment of a category to a person one might propose - not much more complicated than assigned an age. But it is fraught with complications, and if it can be shown to fail, it is reasonable to question all such categorization approaches, upon which much of computerization, not to mention bureauocracy, rests.

We may well object, and in at least some cases this objection will prove true, that there are children who have reading capacities indicative of a different grade level, or we might imagine teachers whose expectations (or limitations) lead them to teach at a different grade level than that assigned to their class. Of course, the routine for assigning children to grade levels is extraordinarily simple: when they have completed the next lower level they automatically advance to the next higher. Sometimes, a child skips a grade level, which is a simple exception. More rarely, we may have situations where a child's grade level is nebulous, for instance, when the child enters the school from a different school, perhaps from a different country. Now it is clear from this simple discussion that entering grade levels for children in a database will be a reliable process in most cases, but not in some others. The computer can only "know" a single thing about "grade level" in our example, despite the demonstrable fact that the human teachers and administrators who deal with the children every day have, assuming they are competent, a much richer sense of the child's "real" rather than "formal" grade level. That is, they know that Child A is at a higher grade level in math, but a lower grade level in reading, and that Child B is intellectually in Grade X but socially and emotionally in Grade Y, and that, for whatever reason, Teacher A's children can be expected to be farther along than those of Teacher B.

This is not a case where the educators themselves see such categorization, and the use of computers to record it, as somehow fundamentally flawed. Rather, they take the results of the categorization as a generic tool with limited applicability rather than as a precise tool with ultimately accurate and trustworthy results. Needless to say, this is not how computer theorists and programmers desire to be taken. Accuracy, completeness, predictability are all central to the conception of a programmable computer and its role in society. We may imagine a subtle, decades-long struggle breaking out between the human users and the technologists who program software. The push from one side is to make the logic more human, and from the other to make the humans more logical. How would these friendly-antagonistic campaigns play out? The software programmers would tend to tell the teachers to select simple, well-defined categories and enter their data accurately so the database can be a useful and reliable management tool. The teachers would tend to ask the programmers to give them software that could be better adapted to the subtle, complex discriminations they routinely need to make in assessing the grade level of their students. I claim that this is in fact the ongoing dilemma of computer use in a wide variety of cases that can, in some measure, be compared to students and grade levels, and further, that if either side of the struggle wins conclusively, it will be something of a disaster. If the programmers succeed in training the teachers to disregard all their inner hunches and nuanced evaluations over their students, they have succeeded in diminishing and retarding their customers, making of them not better teachers but better computer operators. If the teachers succeed in drawing their computer programmers into making logical assignments allowing for every possible discrimination a teacher might legitimately make, they have succeeded in setting up the programmers and their software for failure.

This last point, that computer programs fail when pushed "human-wards" to become more and more complicated, may need some explication, beyond the general proposition that if you begin with a flawed assumption (that children can be assigned a single accurate grade level), then, if you increase the assumption's intricacy, you simply push the flaw deeper and deeper. The programmer's response to the challenge framed above, that one assigned grade level is inadequate and misleading, is to add further assignments. So, instead of a single grade level, we might imagine a dozen different grade levels, one in each of several disparate categories, such as language proficiency, math skills, motor coordination, emotional maturity, etc. Further we might add variables for the teaching level of the child's past and current teachers, or additional assignments having to do with the past and current schools. Then, we might allow several "fudge factor" assignments for the teacher to contribute even finer distinctions that may not even have an official name. Then, because the new complex of grade level indicators has become its own opaque module of the database which only the programmers themselves can fully comprehend, we will likely construct an arbitrary set of rules to condense what has been exploded, perhaps back into a single averaged grade level, a rule which, even if the programmer builds it with input from teachers in the best way possible, in no way captures or conforms to the human process of synthesizing from many different factors a general assessment of a child's true grade level. Once this complicated program with its fragile set of embedded categories and rules is built and deployed, it is almost certain to fail and be discarded or replaced after a relatively brief period, possibly by a return to the earlier simplest strategy of the assignment of one single grade level, which was the programmer's instinctive choice in the first place.

Recall that what has been discussed is about the simplest situation one might choose to play out in such a thought experiment. Real database applications, which are innumerable and ubiquitous in today's world, contain literally thousands of such issues, and have problem areas that overlap and interact with each other to create extended swaths of crippled software. Because the users and the programmers are in the above mentioned struggle with each other, they both choose to see in these deficient programs partial success, and they both assume that with more work, the problems will be resolved. But given the limitation of standard technological approaches, and eschewing the two alternatives of either smashing all computers, or gradually completing the transformation of the human race into an operational adjunct to computer processes, by excising that part of human-ness incompatible with standard computer software, there never will be resolution. The situation calls for non-standard technological solutions.

A Defense of a Postmodern Computer Technology

Postmodernism: a theory that involves a radical reappraisal of modern assumptions about culture, identity, history, or language. [plus technology]

You may think, "the idea is promising, but the context into which you have embedded the idea is fatal to it." I take this as an opportunity to explore that context and why I would choose to lay it out as I have.

In my opinion, the present search for computerized human creativity (a phrase meant to extend or amend the well-established term, Artificial Intelligence) is like the search of the man who lost his wallet on the way into his house, but is convinced he lost the wallet inside the house. When his neighbors come to help him find the wallet, he insists they look inside the house. If they suggest he might have lost it before he even got into the house, he says emphatically "that's impossible."

The house is rational science. The wallet is the primordial source of our intelligence and creativity. Modernism is the view that all intelligence is grounded inside rational science. Postmodernism is the view that modernism may be fundamentally misguided, and that we must search outside it to find what we most dearly desire - our own generational philosopher's stone - a machine that makes good conversation, that cares about its own conclusions.

I did not start out a postmodernist thinker. I came to that view within the last couple years, largely through the force of arguments that seem to have been most powerfully formulated by Martin Heidegger in Being and Time. But I now see that Heidegger is just a critical focal point in a much larger pattern of great thinkers who challenge modernism, which is to say the enlightenment perspective, the "house" of Galileo, Bacon, Descartes, Newton, Hume, and Kant. Since Rene Descartes has emerged as the critical link in this crucial reconfiguration from medieval and renaissance modes of thinking, we can use his name as a shorthand to describe rational modern thought in general: Cartesian.

From childhood, I was fascinated by history, and the only reason I never pursued it was because I found no teacher who knew more than I did - a terrible experience for a child. After a rocky beginning, and the first of many academic disasters, I started out as a rather poetically and historically oriented natural scientist with grand ideals. My first love after history was ecology (which after all is a sort of history of the fabric of natural landscapes), my first goal to defend and "save" the environment from the evils of man-made pollution. My undergraduate research paper was to quantify the soil reservoir of the global carbon cycle - a project which was prescient to the Global Warming controversy. My intellectual idol was the obscure Russian scientist V.I. Vernadsky, who enunciated the idea that later became a commonplace theme of environmental sentiment - that the Earth is itself a living organism. My professional apex in the field of environmental science was in the application of bioremediation to chlorinated hydrocarbon chemical spills affecting soil and groundwater resources. These studies convinced me that natural processes are dramatically superior to human intervention in cleaning the environment of human pollutants.

It was in gathering the data needed to make such assessments that I was introduced to computer science in a serious way. I had always been a power user of spreadsheets, word processors, and graphics/mapping programs. Now I was forced to delve into the theory and practice of databases, and I found my new calling. The disposition of data points in relatable tables seemed like second nature to me, and I very quickly branched out of environmental science and into computer science, professionally speaking, and concentrated thereafter on building commercial database programs. I also delved deeply into the structured markup which enables the internet, and the object orientated programming style, which is the ruling paradigm of advanced software construction. At this point, I would say that I had become fully and consciously Cartesian, as rational as the von Neuman machine inside my computer, and I was certain that the computer could carry out any task we might ask of it, if we could just figure out the right way to program it.

Nevertheless, because I had been exposed to Thomas Kuhn's Scientific Revolutions, and because I had closely followed the introduction of chaos and complexity theories as what appeared to be foreign impositions on the body of accepted mathematical and scientific understanding, I had some localized doubts about the integrity of the positive scientific outlook. One very clear example of the limitations of science came to me viscerally in my acquaintance with the transition in the geological sciences from older geosynclinal to newer plate tectonics theories of orogenesis (mountain building). This was a process which did not start in 1920 when Wegener offered convincing evidence for continental drift. Rather the scientific establishment maintained the total fallacy of Wegnerian theories until the 1960's, when so much deep-sea evidence became available that continental movement was irrefutable, then spent the next several years in grudging transition. I was taught geosynclinal theory, and studied under professors who had earned their doctorate from defenses of it. Their induced cognitive dissonance was apparent. In pondering how a fully formed scientific discipline could so painfully and publically prove itself completely misguided, and yet quickly return to a stance of positive certainty about the reliability of its assertions, I began to carefully carve small loopholes in my overall scientific outlook.

Finally, I began to understand that the limitations of my own software was not my skill, but my assumptions. I knew how much less adept I was than the senior programmers employed by the big software companies or the legendary game developers and hackers, who not only wrote programs, but wrote books and earned millions. I had fallen into a position where I was in charge of a large database program that tracked needy children for a privatized state welfare agency. The limitations of explicit assignment of properties to these children was just so overwhelming, I could not at first grasp it's brokenness. Unlike many other items we put into databases - unlike even employee data records - needy children defy categorization. There were at least two dozen axes of categorization extending from each child - and each one remained largely empty, inconsistent, or erroneously filled. The social workers who had to deal both with needy children and a broken database program simply avoided commiting their children's future to these categories, unless they were commanded to fill them in. And that command was instigated by either legal or financial considerations - amounting to the same thing because legal sanctions consisted of fines. Therefore, filling this database was a process driven by legislation and contractual negotiations rather than by the expected results from any direct scientific model designed to optimize the needy child's chances.

But I took the next step and concluded that even if there existed such a model implemented in my database, and even if the social workers would assent to entering the appropriate data in expectation of a successful calculated course of action, the outcome would not materially assist the social worker to "do the right thing" for that child. Briefly put, the reason for this is that each needy child has unique categories and relationships that are only appropriate for them, and cannot be realistically added to a database meant to track all such children. Furthermore, the potential number of conditions, relations, observations, etc. that may be pertinent to the care and treatment of any needy child is potentially infinite, and also cannot be realistically entered, much less processed using standard algorithms. These conclusions can be illustrated by considering what the social worker does during the initial step in the needy child process - a home visit. On entering the needy child's home, the social worker very quickly absorbs a vast multitude of impressions. These impressions are correlated with the social worker's training and experience, itself a vast storehouse of learned structural models and similar but slightly different situations. Not only can a good social worker handle these cascading cross-referential multitudes of impressions and attendant notions, which I maintain must quantitatively approach infinity, she can process them into an earth-shaking conclusion - whether to remove the child from the home or not. My computer program could never expect to absorb these impressions, nor make these conclusions, not because I was unable to program it, but because the way software of any type is programmed precludes it.

These insights were rooted in my sudden turn to study Heidegger and related existentialist and postmodern thinkers. But before pursuing that direction in more detail, let me wander just a bit farther into this extended example of the brokenness of Cartesian solutions to a very practical every day problem. In my child welfare organization, it was known in the Information Technology department that the social workers who were the very best at translating their workload into the database were not necessarily the best social workers when it came to successfully discerning the needs of, and actively managing the care of, the needy children assigned to them. In other words, technologically adept workers tended to be unsuited in general for the very task we might have hoped to hand over to the technology! So it appears that in these areas, epitomized here by pointing at needy children, not only is it structurally impossible to apply technology, it is fruitless and perhaps counterproductive to apply a technological mindset.

Now I admit that it is plain common sense that computer technology such as the databases I build do in fact have many fruitful applications other than needy child care. One very obvious application is financial record keeping. The database appears very adaptable, for instance, to the structure of double entry book-keeping, and in fact to the extended corpus of financial numerology built up since the Renaissance. What software cannot do well is help explain what mysterious trends underlie the numbers. Just like needy children, moody buyers and sellers remain impenetrable to the extensive software models built specifically to predict market trends. In this particular area, where there is so much potential profit to be had from success (in stark contrast to caring for needy children), the diligent application of calculational techniques might be determinative of the likely success of current technological methodologies applied anywhere. But it would appear that the same problematic uniquenesses and infinite cascades of potential relationships noted above prevent substantial advances here too.

Back to basics. The Cartesian approach assumes that meaningful structures in the world are mirrored by meaningful structures in the mind. When we encounter structures in the world that do not match the structures in our mind, we have a problem that can be solved by reconciling these two structures. First, we have to be sure that the structure we perceive is really there, and not a figment. Then, we have to adjust the structures in our mind to properly accomodate the structure we have found in the world. Consider mathematics (and modern science) as the long-term application of this process. It is not obvious to the primitive mind that all worldly structures have measure and movement, but once that structure has been adjusted so that the world and the mind agree, the math we do in our mind and the math that determines the structure of, say, gravitational motion, comes into accord. This accord gives us the power to design bridges, skyscrapers, space rockets. The success of these endeavors in turn validates the Cartesian approach.

In Cartesian computer science, the implementation goal is to make the software mirror the structure in the mind which mirrors the structure in the world. This is the analog of making the space rocket's guidance system mirror the mathematical (mind-based) structure of gravitational (world-based) forces. The Cartesians maintain that if we stick to this approach, we will ultimately prevail on all fronts. If we can adjust the structure in our minds to fully reflect, for instance, the structure of needy children, or the structure of buyers and sellers, then we could achieve the same successes in these areas as in the others. But the Cartesian approach requires that uniqueness be distilled out of the situation by assigning discrete, abstract properties, and assumes that the structure under investigation can be bounded in finite ways.

Going back to our example, a needy child must be classified as either disruptive or non-disruptive in the computer. He cannot be disruptive under certain conditions which cannot always be well defined or predicted. If the saliency of this issue forces us to push past these initial binary choices, the Cartesian approach would then be to determine all the meaningfully discrete shades of disruptivity (highly, sometimes, rarely, etc.), and assign the child to one of these. But let's imagine the child may be highly disruptive only when he dreamed last night of an awful experience neither he nor anyone else can remember or reconstitute, and then was either awakened by shaking or ate too much sugar at breakfast. What category would that be? The child's disruptive nature is unique and cannot be abstracted; it is not limited by distinct, finite boundaries.

At this point, not only do Cartesians "give up," they insist that such a situation can safely be ignored while their program passes on to more tractable problems. Now realistically, this may be quite the pragmatic thing to do. But the critical step is to say that because you can't solve it with a software program, then it can be ignored, then it is unimportant, then it can wait. But a needy child by definition refutes these assumptions. He cannot be ignored, he cannot wait, and the outcome is important to everyone in society. To take the opposing view is unpractical (considering the very real impact of crime), cynical and self-serving, deeply anti-humanistic, not to mention spiritually darkening. One could argue that the Cartesian may recognize the importance of the task, but is simply saying that the computer is not yet ready to tackle it, and until it is, the problem is best left for humans to sort out. But I have pointed to the possibility that computer software as presently understood will never be able to replicate the achievement of a technologically naive social worker. This indicates a much deeper problem than refractory complexity. The Cartesian choice to simultaneously claim to be the only approach worth pursuing, and then to only address problems it can solve with facility, may be the deepest problem with modernity.

So the postmodern critique opposes the Cartesian self-sufficiency that enlightenment rationalism has institutionalized in modern science. Like any incipient revolt, postmodernism is chaotic and incoherent, even after well over 100 years of development. Whereas (almost) everyone agrees what makes good science, no-one agrees what makes good postmodernism, and many deny that there can even be such a thing. Many thinkers who some assess as postmodern deny it (Thomas Kuhn, for instance). They see themselves perhaps as entering a corrective adjustment to the journal of modern intellectual development, rather than demanding a break with it. The term postmodern is used in many different contexts which do not seem to bear any resemblance to issues in science and technology. Modern art, which defies any standard of ordinary evaluation, which in effect forces the viewer to create a new standard each time out, is commonly called postmodern. In literary criticism and philosophy, the term postmodern is associated with a strident relativism and multicultural perspective. Many of these postmoderns insist that nothing can be taken for granted, neither the canon of western thought represented by the Great Books (written by aged white males), nor the ordinary standards of critical assignment of value (good or bad) to any cultural artifact whatsoever. Thus, to take it to the extreme, a comparison of modern New Yorkers with Amazonian head-hunters must not pre-suppose the superiority of delicatessen over roasted man-flesh. Such a stance is bound to accrete layers of revulsion among ordinary observers who do indeed take such judgments for granted.

Vision of a Better Technology Based on Dreyfus' Critique

My objective is to use computer technology to drive back the ill effects of computers - this is a line that Dreyfus has worked much on - he has been the strongest critic of technology, but has also pointed to ways a computer could be used in a different way - a re-humanizing way, as a vehicle to review and perhaps recover marginal practices and perspectives. What is the difference between just "reading about" earlier cultures, and "experiencing" what they were like? I want to take people "into" history, different worlds of understanding, with the aid of the computer, which has the potential to make this possible.

Currently we approach technology as subject/object, and then lose ourselves - we pass beyond subject/object and become pure resources. This is, in one sense, an annihilation of ourselves and our culture. So we have to go back to initial assumptions about the computational model - Turing machines, etc. We have to find a technology that clears out a part of our world and provides an articulated background. We have to look at how our "brains" really work - the phenomenology - and try to get our technology working in a similar way. We assumed (in the 40s and 50s when we started computers) our brains were calculating machines - but they aren't.

For Dreyfus, the problem seems to be that we cannot program computers to have familiarity. Dreyfus' deepest objection is that computers do not have a body - the "embodiment problem." This stems from Merleau-Ponty, who says all our thinking is "gestural" - he means that when we think, it registers in muscle, as in what we call "body language," but even when we do not appear to move, the body is working with the brain to lay down retrievable patterns - memories. Merleau-Ponty emphasized the idea of memories being in the "now," not in the "past" stored somewhere in the brain, in what he termed a sedimented history that is present with us all the time. Computer science accepted the "in the brain" idea - that is why the computer has a "memory."

Dreyfus references the experimental neuroscientist Walter Freeman and his findings on brain dynamics extensively. Like Merleau-Ponty, Freeman claims that the brain does not "hold" memory, but rather maps it onto brain energy fluctuations - a certain stimulant will produce different energy, but the same response.

These philosophical and scientific strands would argue that our "now" is a view looking at our articulated background, sets of meaning imposed on an essentially chaotic and/or very intricately structured, stream of incoming energy from the real world out there. My plan is to superimpose our culturally orchestrated atomic bits of meaning - what I think of as ontic, semiotic, or semantic molecules - onto a chaotic background. Then, to make that background available for a user to "find" and "articulate" what that means to him or her inside a computer program.

So in my new system, there are these bits of meaning (which I assume to be our best shared undertsanding) which will map onto a chaotic (or complex) backgroud flux (or pattern). In practice, these bits would be finite lists, such as the 26 character alphabet. They could be any alphabet, or syllables, or words, or many other such things - they will map into chaos and create lots of non-sense, and just a few sections that make perfect sense - the program seeks those sections. The challenge is to reduce this idea to an algorthim set that an ordinary computer can process, with a standard graphical interface we can use as a tool.

Dreyfus' critique is based on the premise that an artificially intelligent computer must "stand alone" as an isolated system, and therefore cannot escape the embodiment problem without the highly suspect creation of artificial humanoid lifeforms. On the other hand, my idea is to have computers "mesh" with users. The embodiment can't be done solely with the computer - at best it could be simulated, but that will never capture it. I believe we need to let active users provide the body part of the equation and let the computer extend it. If users have familiarity - so will their computers. My project is based on a "pairing" idea, where a human supplies the bodily reactions and the computer does a different part of the processing. All of the Heideggerian "background" - the constituting experiences one has as a person embedded in a culture - would not be in the program. Each user would have to build up an "articulated background." We can supply some of that - like when we educate little children - but most would have to be built up. It would likely call for patience to develop a filter of sorts, unique to a particular human, which makes the interpretations.

The demonstration "work space" software I am building will not be dramatically life-like, a new and improved Second Life virtual reality, or an autonomous robotic agent, or anything like that - more like an advanced word processor. I am aiming for a very simple Merleau-Pontian program, one that builds up a "sedimented history" of past experiences. The real problem is going against 50 years of computer assumptions.

Tesselectronica

The technology to store and retrieve data in modern computers is highly developed. Modern databases store billions of items of information regarding financial, operational, and human activities. Internet search engines treat millions of world wide web pages as one vast data warehouse. Strangely, we still notice that computers do not "remember" things at all like we do, and worse, we users have given up expecting them to do so. Instead of computers becoming more like us, we become more like them - it is more accurate to say that we adapt to our computer programs than that they adapt to us. This applies even to the latest trends - to use social networking sites such as MySpace or Blogger, we first must learn how they work, or to use text messaging, we learn how to type with our thumbs, and so on. I believe the reason for this striking cultural and technological straightjacket is our standard approach to computation and software. In effect, we are wearing blinders, and computer technology is a victim of its own success. We keep trying to do what we are already doing, only faster. But, by developing software that goes "against the grain" of established practice, we may be able to develop a new generation of tools that treat information much more like we humans do. I call the outcome of this development process, which is now in progress: "Tesselectronica: The First Bio-semantic Thought Processor." This software will not "store" data directly, but rather point to where the data patterns already exist in the calculational universe. It will not classify data directly, using imposed schema, but will bind disparate core memory addresses together by subtle similarities keyed on distinct patterns and entry characteristics, including bioinformatics from the user. It will not attempt to retrieve one "correct" bit of information, but rather it will activate a "spike burst" (a dense prioritized web of connected patterns) in response to input, and allow the user to select and refine, then re-memorize or export their own unique output for that session. The patterns that will emerge into a user's workspace will span multiple formats, including written characters, sound waves, and two or three dimensional animations or graphics. The software would require, and greatly assist in, the normal stages of human learning; it would not replace humans, so much as empower their inate creativity. The first applications may well be educational, but many more specific applications will unfold from that beginning. Ultimately, using this software, a person will be able to "think" an email address onto their screen, to "remember" related words, sounds and pictures into the computer as a bundle, to have their computer surprise them with creative (or mistaken!) data connections, and to transmit core data repositories to another person and location with great efficiency.

History and Thinking Machines

So you may still be wondering why on earth I am so adamant about the "cultural" story that involves modern philosophy in a very disturbing juncture with pure, undisturbed science and technology as we know it.

It occurs to me that in our discussion we were very close to the answer to that, but maybe never really got to it. So let me make a try in this email. Remember I drew a circle-map on the whiteboard that was meant to represent that our current "T" technology world, although it seems self-sufficient, is actually embedded and adjacent to other, now lost, worlds, such as the "M" world that the enlightenment moderns (like the founding fathers) knew, and the "W" world that was the West that preserved a rather strange form of Christianity and fought the crusades, etc. and the "A" world of the ancient mediterranean greeks and romans, and the "C" world of cavemen running very fast and eating lots of meat. Of course I left a lot out. But the point is that to be in the "T" world means that we have strange and not very predictable connections - in mind, body, and soul (whatever that is) - to all these other worlds.

Now, most of us think we are past all that - history is a very minor concern to technologists, perhaps most see it as trivial. And that's the problem. The post modern philosophers (who are certainly considered trivial in computer research) have discovered a very frightening thing - that technology and science are "derivative" of the "T" world and the "M" world, that is, of our culture. That is, the technology we have is not the only technology possible, just the one that is historically presented to us by the almost accidental confluence of cultural practices like printing, or record-keeping, or clock-making, etc. (One way to imagine this is to read the "fiction" - both science fiction and fantasy - that explores imagined alternative technologies.) This view clashes with the scientific confidence that we have (finally) got it right - that our knowledge is true, and our practices are correct, and no others would be.

One reason scientists do not see themselves as derivative from a culture is that the science has itself become our culture, and generally humans cannot easily see and study their own culture. This is due to the difficulties of self-reference. Anthropologists can study other cultures using "western" methods. That works on the bushmen or the eskimo, but not on the New Yorker. Why? Because the anthropologist IS a New Yorker - he has to devise ways of studying himself, which breaks down for obvious reasons. Similarly, science cannot see itself as derivative of the culture, because it has become the culture, and this trend will only intensify into the future.

So it is this evaporation of any ability to see one's current strange new culture turning into an invisible background structure that dominates everything one does, combined with the incredible power of modern technology, that represents a real challenge. Forgetting the way in which our culture came to present us with this technology, may be one of the worse things ever. Take an example: as a modern looking back, you can see the criminal insanity of something like the crusades or the witch burnings of the middle ages. But if you had been born into it, you probably would have supported or participated in those activities.

Technology has this window of opportunity to turn itself into a solution to this whole issue of cultures arising from other cultures and forgetting themselves throughout history. Technology could itself take control - and thus responsibility - for laying out all those strange lines of influence I suggested in the first paragraph. In this way, science insures itself from the worst consequences of thinking itself supreme in total megalomaniac isolation.

A program like my thinking machine could be a start. One of its obvious applications is not just studying history, but in partially recreating historical contexts and reactions. If you can actually embed different pattern-reaction sequences into retrievable memory addresses, then you can at least simulate a large variety of different configurations, maybe one set for the "W" world of crusaders, another for the "A" world of ancients, and finally one for the "C" cavemen. And what I just described is really just very profound ways of constructing riveting games. So maybe the way to save the world is to make better games. That particular cultural current is already underway and getting stronger every day.

Retrieved from "http://www.whooshup.org/wiki/index.php/Karl:_Postmodern_Computer_Technology"

Karl: Postmodern Computer Technology

From Whooshup.org

Contents

Computers vs Phenomenology Part I

Computers vs Phenomenology Part II

Problems with Categorization - the Grade Level Disaster

A Defense of a Postmodern Computer Technology

Vision of a Better Technology Based on Dreyfus' Critique

Tesselectronica

History and Thinking Machines

Views

Personal tools

Navigation

Search

Toolbox