Tue 12 Sep 2006
What I Believe But Cannot Prove
Posted by anaglyph under Art, Philosophy, Religion, Science
[23] Comments
I’ve just finished reading a very interesting book called What We Believe But Cannot Prove, in which prominent scientists and scientific thinkers speculate around their disciplines on things that they feel are true but for which there is as yet no scientific substantiation. I really recommend this book as an insight into the way scientists think, but also as an insight into the way in which they explore their intuition.
It should be made clear that the speculations in this book are distinctly different from just ‘wishes’ or irrational beliefs, but are hunches that scientific thinkers have based on their work, or on deep and methodical thought about their particular subjects.
Reading the book prompted me to think about what kinds of things I believe, but can’t prove. The speculations that follow are also based on some sensible science, I hope, particularly on my work with mathematics and image generation.
So, for your perusal:
What I believe to be true, but cannot prove, is that all life is the result of complexity that emerges from the interaction of very simple rules. In other words, I believe that, although we may never fully understand how it happens (indeed, it may just be unfathomable to our brains), sets of simple descriptions linked together by circumstance, chemistry and time contain all the information necessary to describe all the lifeforms we see today and that we will ever see.
This is not to belittle the majesty which is life. It is a truly remarkable outcome from humble beginnings, and is so extraordinary that many people feel the need to evoke a Creator to explain it. But I don’t, and I’ll try to give you some idea of why that is.
Some of you know that one of my many interests is the creation of images from mathematical constructs. Almost unfailingly, when uninformed people see my work they leap to the conclusion that it is something to do with fractals, even though mostly they will not have a clue what that actually entails. But my work has little to do with fractals strictly speaking.
The software I use* allows me to take algorithms that make simple descriptions of Cartesian (or flat geometrical) space and chain them together in such a way that aspects of one simple description contribute to the outcomes of all the other simple descriptions. In other words, each simple description speaks to every other simple description so that altering the aspects of one has effects on all the others. It is, to give a considered analogy, the genome of my artwork.
It really is a breathtakingly uncomplicated idea; I might have one algorithm that describes a circle, and another that describes a modulated distortion of that circle. Yet another describes colour, another describes how many times that circle is drawn. Another balances the shading applied to the image. As I start with these very easily comprehended descriptions, everything seems pretty basic. I can predict, in systems with three or four algorithms, pretty much what tweaking any of the variables in the equations will do to the final image. The results of this rendered in image form are pretty much what you would expect – fairly prosaic geometric figures and bright colours. But then, after about seven or eight algorithms have been connected together a very eerie and profound thing takes place: the outcome of changing any one of the thirty or so variables that have now been introduced is simply not predictable. And the images start to take on very special unspecified qualities. The system has become so interdependent on all its many changeable pieces that a level of extraordinary complexity has been reached. For the works that I am now creating, my systems typically contain between fifty and eighty algorithms, encompassing something like 240 individual interconnected parameters and feedback loops.
This is for all practical purposes an entirely unnavigable artistic space. Even if I had a billion lifetimes I could only explore a small fraction of the possible results. It would be a mind-numbingly daunting prospect, were it not for some other tools at my disposal, notably a way of applying rule-based mutation to the equations and observing generations of outcomes.
When mutations are applied, many of the outcomes, in fact most of them, deliver uninteresting or even ugly results. But every now and then something truly surprising and beautiful springs forth, and I can, using other tools, refine my image in countless ways. (I must emphasize that this is not a random process†; there are strict rules in play here. It is just an unfathomably complex process that can seem random without inside information). Using a combination of mathematical knowledge, the mutation tools, and artistic intuition, I now have a workable method for usefully exploring these vast numerically-defined landscapes.‡
Of course, some will argue that in the process I’ve described above I am playing God by making decisions and crafting the direction my creations take. It is my will alone that decides what is valuable and what is not. Why is it different in the real world? That’s a fair point.
If we extrapolate my dalliances in mathematical invention into the real world, is there some persuasive natural force that obviates a need for a Divine Hand and would guide the best choices of the available descriptive options in such a way to provide the kinds of outcomes we see in life all around us? I believe so, and I think that force is Natural Selection.
In other words, what I believe is that Natural Selection and the fundamental mathematical rules of the material world work in concert to efficiently, explosively and inexorably produce the kinds of life that we know. Mysteries that exist in current evolutionary theory, such as the incredible rapidity in which life has evolved on our planet, the extreme flourishing of diversity to which the fossil record gives witness, and the sheer determination of life to arise anywhere it can, have puzzled scientists for many years. If, however, we factor in an ability for some kind of information coding system (like DNA) to exploit mathematical physical laws that already govern basic matter (things like surface tension, viscosity, crystallization), stir in some (I believe inevitable) emergent complexity, and add some time and some kind of filter process (Natural Selection), I think answers to these mysteries present themselves readily.
I can’t prove I’m right about any of this. But my artistic intuition, combined with the small amount of science and mathematics that I have informally picked up over the years makes this feel right to me. I’m trained as an artist rather than a scientist, so I get a bit of license to lean more heavily on intuition than your average diligent scientist. As far as the science goes, though, there is recent thought from scientists much cleverer than I am that tends toward support for my hunches, though I’ve not heard anyone voice these exact ideas. Mathematician Stephen Wolfram in his extraordinary work A New Kind of Science suggests that cellular automata, or simple generative rule systems, are the basis for the structure of living organisms (amongst other things), and author and mathematician Ian Stewart in his book Life’s Other Secret advances an hypothesis that proposes that DNA might exploit natural physical properties. I have riffed heavily on their ideas. I have also taken inspiration from Philip Ball’s The Self-Made Tapestry and Paul Davies’ The Fifth Miracle.
If I’m even close to being in the right ballpark with any of this I believe I can make a few predictions stemming from the ramifications of my hunch. The thing about the net is that this is here for posterity now, so I will look damn stupid if I’m proved wrong. Nevertheless, I have the courage of my convictions. Here we go:
1: I predict that when we finally examine the genome comprehensively we will not necessarily be able to make direct correlations between particular genes and what those genes ‘do’. Indeed, Information Theory already hints at this outcome.** In other words, we won’t be able to point at a particular gene and say “#4123 specifically codes for eye colour”. I believe it will turn out to be far more complex than that. That is to say, #4123 may indeed have an effect on eye colour, but I’m going to speculate that if you tinker with that gene to get brown eyes rather than blue, you will also alter the balance of many other aspects of the organism in complex and unpredictable ways. I think we will discover that all the genetic therapies we are currently undertaking will in time prove to be clumsy and indirect and possibly even detrimental in some cases. Of course, if past behaviour is anything to go by, the pharmaceutical companies and Big Business behind gene research will go some way toward obfuscating any results that point to this, because the implication is that there is no direct cause/effect correlation, and business hates that kind of thing.
2: As a result, I predict that knowing how to effectively mold the results of genetic manipulation will become a matter of intuition and inspiration; the domain of artists rather than businessmen. Therefore, I predict that the design of new drugs, prostheses and even life-forms will become an artistic endeavour rather than a technical one.
3: I predict that the principles at work in emergence will also be shown to exist at the core of other complex systems such as societal behaviour, economic systems and neurological processes. I think we will learn that ‘consciousness’ is an emergent effect from mechanical brain function.
4: I predict that the internet will start to exhibit this emergence in an overt manner (I think there is already a level of emergence in the net, but that as yet it is merely phenomenological; I am speaking about something deeper, and probably to us, spookier, than this).
5. I predict that when we do find life elsewhere in the universe (because I am very sure we will, maybe even in our solar system), we will find it abundantly and in spectacular variety. I don’t believe we will just find a little microbe hiding under a rock, but entire complex ecosystems. And I think that we will find them anywhere that they are technically possible.
I apologize for this lengthy, serious and technical post. I know this is not good blogging practice. Regrettably, you will be obliged to put up with this kind of thing from time to time. Normal programming will resume presently.
___________________________________________________________________________
*The application is called ArtMatic, developed by Eric Wenger and Edward Spiegel of U&I Software, designers of unique creative enabler tools. I’ve been priveleged to have been in on the development of the ArtMatic project from the ground floor nearly eight years ago. It is no secret that I am not fond of the name of the application, it being rather too redolent of ideas of ‘art’ being produced ‘automatically’ which is about as far from the truth as you could get. C’est la vie.
†Although Randomness is another tool that I can bring to bear if I wish. It’s also a tool that nature has at her disposal.
‡One of the most astonishing things is that the beautiful results I am able to generate are often evocative of things that we see in the natural real world. I don’t explicitly make them like that, I just set up the right conditions – the forms themselves are emergent from the maths. I can control them comprehensively for sure, but ultimately they exist somewhere in the numbers. Occasionally this really spooks me out… I don’t believe this echoing of natural forms in pure maths is at all coincidental, but that’s material for another post, perhaps.
**Simply put, it is possible to show theoretically that DNA does not contain enough explicit descriptive information to detail the complex outcome of, say, a jellyfish. Consider: there are something like 20,000 genes in the human genome (the exact number is still unknown). It’s really not that many. If you allocate tasks directly to parts of the gene it’s just not enough ‘bits’ to describe all the complexity of a human being (even taking into account contributing factors from ‘junk’ DNA or Regulatory Sequences and other gene ‘enhancers’ that seem to be in play ‘controlling’ gene expression). If you don’t do that, you have to come up with an alternative information-carrying concept, which is tricky (not to say messy). Some contemporary Christian thinkers have leapt on all this to say “Aha! There’s God at work”, but of course they would.
___________________________________________________________________________
Click here to vote for this post in The Philosophy Blog War!
___________________________________________________________________________
23 Responses to “ What I Believe But Cannot Prove ”
Trackbacks & Pingbacks:
-
[…] Home of The Cowlex™ Vibrator (oh, yeah, and Watches) « What I Believe But Cannot Prove […]
Fabulous.
If I am not mistaken, the book you have been reading is from The Edge, Question of the Year 2005, Yes? I read a bunch of those responses, and try to go back there now and again to read more.
I predict that your predictions will come off pretty well. I would describe some of it differently: I think that this intuitive/artistic way of thinking you are trying to describe is at the heart of the issue. That science can interfere with understanding, primarily because of its reductionistic quality, and that we will start to understand things a whole lot better when (if) we re-learn how to think intuitively/systemically/ metaphorically.
Funny, once again I am reading an article that very much touches on some of the ideas in my latest writing.
Cheers!
tp
Interesting. I think you’re not far off in your predictions, mostly. Not because I think your thinking is at fault in anyway, I differ with your opinions in a few instances.
I think when the Internet makes some sort of emergence, it will be pretty anticlimactic. I don’t really like to join the X-Files and bad haircut crowd by calling any sort of emergent behavior the Internet exhibits as “life”, but certain requirements are needed for independent thought. The first is informational rescource, which is no problem, but the other is energy rescource. The Internet probably won’t be HAL anytime soon, or even in the future, but it will be incorperated more and more into everyday life until it is something like fire, a resource that is so ingrained into our species that it will almost be an earmark of our species. I would equate the Idependently Intelligent Internet to the emergence of an oxygen rich atmosphere in the precambrian. Not life, just a new byproduct visible without the eyes of metaphysics.
Scientifically only one thing can happen from there.
The new Independently Intelligent Internet (III) will start a war that spreads all over the world, aptly titled the War of the World Against the Independently Intelligent Internet (WW III), so that it can use our livers for fuel. The war will reach a climax forseen by the prophet Daniel back in the fifth century B.C.
Furthermore, I think that God will come back and prove me right and you wrong by rewarding us in eternity with paradise or torture, respectively. Then he will say something terrible your mother for raising someone who doubts Usher’s Chronology.
Sinner.
tp: Yes, you’re right – the book I mentioned is a compilation of some of the best Edge Questions from that series.
I don’t agree that science must necessarily be reductionist. I believe there can be gestalt approaches to scientific reasoning, and I think we already see those kinds of things (especially in physics). This is what I’m trying, perhaps clumsily, to explain in this post: that even though the rules might be simple, the process and the outcomes are profoundly not. So much so that they may in fact be beyond our capacity to understand in any reductionist manner, simply because we are unable to apprehend the huge numbers of variables at play. Think about the configuration of a chess game – 64 squares, 32 pieces and a handful of rules. Only a few brains on the planet can truly get good at handling the numbers of possibilities available with these combinations. Now multiply that by a few billion… Frustratingly, it’s much easier to explain this by showing someone how I work – I can get people to understand the basic concepts in a few minutes. It’s a lot harder to get it across just by writing about it.
One thing I would add, is that I also try not to make distinct delineations between science and art. The essays in this book show scientists using thinking that is often relegated to the realm of artists. I think most good scientists would agree that this kind of thinking is entirely necessary for making headway in their field.
Casey: I don’t think the internet will suddenly develop intelligence either, and I like your oxygen analogy. What I think will happen is that there will be major emergent phenomena that are not explicitly coded (that is, not intentionally designed). I think these phenomenae will be an inextricable meld of net culture and human ingenuity. We’re already seeing prototype examples of this: Google Earth map mashups is one. In this case, people are using data from one system, say police pedophile databases, and merging them with Google Earth to make maps of pedophile activity accessible by anyone. This ‘meta-function’ allows a utility not present in either Google Earth, or the police database alone. This is only a small case, but I think we’ll see some remarkable development of this idea. And no-one plans it – it just happens…
The God thing? I believe in God, but I’m also a firm believer in science and logical thought. I think that there is a God, but that this God is not necessarily like the representation of God in any of the major religions. I tihnk that they hold bits of truth, but the overall picture is not correct.
Thinking.
No need to apologize for such posts. I enjoy reading your philosophical musings very much. I am writing in simple sentences right now because I am truly not used to waking up at 5:45 am and remaining awake. Also, I am in a dungeon of sorts with no natural light.
I wish I could express all of the thoughts and processes that you sparked in my mind as I read this, but I lost track of them as I explored each strand of thought and returned to find my brain glucose levels alarmingly lower each time. Sufficed to say, it has been quite a trip, I have enjoyed it immensely, and of course I will return as further thoughts occur to me.
I do agree on the whole with your predictions, and look forward to a clearer (read: less sleep-deprived) state of mind in which I will be able to articulate where my opinions diverge or merely, I believe, invite a dicussion of semantics.
Anaglyph,
Yes, this is where we disagree. I do think that science is inherantly reductionistic. Without reduction it becomes difficult and eventually impossible to quantify. And science depends on that quantification.
Heck, look at all the effort being expended on the TOE. I think that systemically understanding the relationships of the forces of our world, where they meet up and interrelate, how we get from here to there, etc is fascinating, But I saw a physicist on the Colbert Report from last month (don’t recall her name) that described the search for the TOE as a search for one single, simple equation that explains everything.
One equation? Possible. Simple? I don’t think so.
And looking for a theory that can be squeezed into one equation? Probably an indication that systems thinking is not on her list of real interests.
Maybe I’m projecting, but there you have it.
tp
I really want to focus on this intuitive leap idea because it has come up before in one of Evenstar’s posts and it’s something I’m very interested in. You said:
“2: As a result, I predict that knowing how to effectively mold the results of genetic manipulation will become a matter of intuition and inspiration; the domain of artists rather than businessmen”
Now I’m assuming here that the concept behind the ability to make intuitive leaps is that while we are incapable of consciously (*looks around nervously because she used the ‘c’ word*) processing the steps required for this kind of creation, at some level we possess the necessary information and ability to manipulate it. Otherwise it’s not intuition it’s just random guessing.
So then, while we can never know how we got there in the first place, isn’t it true that after a certain number of “correct” intuitive leaps we could begin to reverse engineer the process and gain an awareness of the patterns? If that’s true then it might be more cyclical.. with creation being selective, then directed, then selective again to the next “stage” and so on?
I’m not at all sure if anything I just said is either intelligible or relevant but I’m certain one of you will be able to explain it to me :) That’s why I love yas.
tp: Well, there are scientists, and there are scientists. It is my opinion (and this is a highly divisive issue amongst scientists) that you can practise science without believing that the human brain is capable of comprehending everything. The approach that says we must be able to understand how the universe works is completely hubristic I think, and any good scientist must accept that there is a possibility that we may not be able to do so.
If you accept that, then science becomes the pursuit of trying to understand as much as possible about nature as we can through observation. I don’t see anything wrong with this, and I don’t see it as a necessarily reductionist pursuit.
I fear that I have not been at all successful in making the main point of my concept. I think it is entirely possible that we may be able to understand the rules but I emphasise very strongly that this is not the same as being able to comprehend the outcome. This idea is radically different to the 19th century view of science that is still our legacy.
To get close to grokking it you must abandon reductionist thinking. Reductionism says that we will be able to understand how things work by disassembling until they reveal their simplest working parts; that is until they can be reduced no further. My speculation is that there are simple rules, but that they cannot necessarily be deduced by dismantling the system; that is, what we see as life is an emergent property of simple structures. That more comes out of the rules than is actually encoded by them. This is a deeply unintuitive idea and it immediately evokes the idea of the supernatural but it is explicitly not necessary to call on God. It is entirely possible to encompass what’s going on in scientific terms . But here’s the rub: you must completely revise any belief you have that it is (necessarily) comprehensible.
The next step is the navigation of this space – is there a useful way of approaching these concepts? I believe there is, but that it involves gestalt thinking. It is, if you like, the very opposite of reductionism. This might appear as a dualist concept, but it is not – the gestalt is the only useful way of navigating this complexity. It is simply impossible to do it in any other way. This complex space is emergent from, and interleaved with, the rule system. The two things are one.
Which leads me to Bean’s question “Could we intuitively gain awareness of the patterns?”
I think that’s all we would gain awareness of. Shifting patterns in a complex space. But again, the complexity is so vast, that other possibilities spring up immediately. One of those is that there may be many, maybe even infinitely many ways to get to a certain place. One exceptional biological example of this is the eye; the many kinds of ‘light-detecting’ apparatus that have evolved over time to exploit a biologically useful phenomenon. The eye is often given by Creationists as the proof that God has designed us. “How could something so complete arise in stages – it’s either useful or not!” Without going into the easy demolition of this argument, looking at the phenomenon of the eye itself tells us a great deal about evolution – an eye is useful. Consequently, variations on ‘eyes’ have arisen many times through many independent means. So the ‘eye’ pattern is embodied somewhere in the rule system. Not just our eyes but the concept of ‘eye’. Evolution finds paths to this place because it makes sense.
Trying to find these paths in a concrete sense though is futile. You can trace back one path of eye evolution and it tells you nothing about why or how ‘eyes’ in general develop. Breaking this down into small pieces will not help – if you have ten different evolutionary paths that lead to eyes, you’ll have ten different rule systems that go to make up those paths. And the rules themselves will have nothing in common other than that they led to ‘eyes’. This is a very difficult concept to grasp, but it is the core of what I’m trying to get across. I don’t think we’ll ever comprehend this fully. But we can make large gestalt-type intuitions.
It is important to understand that these kinds of intuitions are fundamentally different from the kind of intuition that says you should buy a lottery ticket. In fact, I would class this kind of intuitive speculation based on knowledge as its own class of thinking. And it gets very close to the way artists work to make meaningful gestures about the world.
:( I’m sorry if I missed the point. I guess I was comparing it with encryption, which is something I’m a bit more familiar with. In many kinds of encryption (including the popular md5) a word will always encrypt to the same characters but those characters can then be decrypted into many different words, which is what makes it “difficult”. Naturally though, these kinds of encryption are becoming easier to break when there is any sort of context available. I know it’s silly to compare something like simple encryption with genetic evolution but it was the first thing that came to mind. Back to the drawing board!
Encryption and compression also spring to mind when you talk about the jellyfish not containing enough data in his genetic code – I imagine (because I’m not really capable of intellectually comprehending it) that encryption/compression systems could apply to genetic data the same way they do to digital.
Hehe, I think I should stick to sensuality.
The difficult part about this idea is emergence. As I said, it really counter-intuitive. The idea that you create very simple understandable rules that generate things that are incomprehensible defies common sense. It seems logical that if you can describe the ‘bits’ you should be able to describe the whole.
It’s taken me years to get my head around this, and I think I can only do it because I have this artistic avenue into it. When I’m making my works, I can literally see this happening in front of me. When I started out I found it extremely spooky at times. I still do!
You have me wondering whether it would be possible to make an encryption system based on emergence, but my intuition tells me probably not. The end results are so complex, I’m not sure whether you could find a way back through it.
Encryption is a form of one-stream math designed specifically for back-tracking (if you have the keys). When the result is ’emergent’ you haven’t expected the results (by definition!)
Interesting post. Supposedly the human mind is only capable of imagining four or five interacting functions. The visual space you explored is thus far beyond the human brain’s direct ability to predict. However, computers can simulate systems with thousands of interacting functions, e.g., weather simulators.
“1: I predict that when we finally examine the genome comprehensively we will not be able to make direct correlations between particular genes and what those genes ‘do’.â€
Your speculation has merit but I predict you will be wrong. More specifically…
Many low-level traits will be directly associated with specific proteins and hence to the DNA that codes for those proteins and the regulatory DNA that controls the amount of those proteins in a cell.
As for high-level traits…
Some traits will depend on many genes and some genes will affect many traits. Such relationships are common in scientific modeling. Gene engineering will be complex but not beyond computer simulation. Biological simulation experts, not artists, will be the important players.
There was an evolutionary constraint for some traits to be adaptable. Animals that could rapidly change such traits had an evolutionary edge. How else could an animal rapidly adapt to new diseases or changing food sources? Such traits will have simplified, relatively independent controlling genes. Changing such genes won’t significantly affect other important biological traits.
There was also an evolutionary constraint to keep some traits stable in the face of genetic variation. This is canalization. The vole is an example of an animal with great genetic variation but little phenotype variation. There are many mechanisms by which canalization occurs. E.g., there might be several genes that make functionally equivalent proteins. Regulatory mechanisms control the proper protein expression level. The result is that breaking or changing one gene needn’t change the phenotype.
A gene engineer could use compartmentalization. A new gene would only be turned on during a specific developmental stage or in specific tissues or when a trigger compound is present. Thus reducing unwanted, secondary phenotype affects.
Fly: Thanks for dropping by.
I appreciate you taking the time to not only carefully read my post, but to reply intelligently.
Well, as I said, my credentials in science are only those of an amateur. But I feel I can take you up on a few points. Complex modelling is one thing. Emergent complexity is another. I understand that we are able to model simulations of weather. This is significantly and fundamentally different from modelling what’s going on in actual weather. A computer model gives insight to the gross processes, but it doesn’t tell us what is happening in reality. My suspicion is that we will never be able to make computations of that complexity (what you’re asking is that we get to the point where we plug in all the data and get a simulation that shows a hurricane of a certain force hitting New Orleans at a particular time – I’m highly skeptical that we’ll ever get that kind of utility out of computers, no matter how many calculations they can do. Emergence dictates a certain level of complete unpredictability.)
This is not to say that this insight is not useful in understanding real weather, and in that manner I suppose that when it comes to biotech, computational number-crunching might actually provide tools for the ‘gestalt thinking’ I have spoken about.
I stand by my assertions though. I believe that in most complex systems, the only computation available to understand the system completely, is the computation in the system itself. I don’t want you think I’m making a semantic argument here. I think the complexity of such systems is so great that they are not understandable or computable in any meaningful terms.
It may not be that artists, as we understand that term, will be the ones doing the gestalt thinking, but, in the same way that art is always led by technology, I believe your Biological Simulation Experts will get to a point where they are using ‘artistic’ thought processes to fully grok their systems. I really do think this stuff will get to be outside the utility of computers. Computers will be aids, but the real processes will be intuitive.
As I said, I can’t prove it. It’s a hunch.
As far as evolutionary control of gene expression goes, yes, you’ve made some good points, but nature has one thing on her side that we don’t – evolutionary failure. Nature can and does make imperfect life-forms. All the gene tinkering by nature results often in failures. Myopia, albinism, dwarfism, merged foetuses – you name it, nature’s mucked around with it. Nature uses the ultimate ‘suck-it-and-see’ method. We simply don’t know how the evolution/gene interface works, but we can be absolutely certain that it isn’t a simple process.
So genetic variation cannot be judged by our view of the successes; if humans are to start tinkering with genes we will not be allowed failures. We are prevented from using trial and error, as nature does. We will have to invent better methods, and I think those methods will involve gestalt thinking approaches. I stand by my assertion – there will not be one-on-one correspondences with genes to effects. I don’t believe nature does it, and I don’t believe we’ll do it either.
I don’t want to sound like I think that we’ll discover no way at all to control the effects of genes – we have already, and we’ll get better at it. But I think there will come a time that no matter how much we understand, and how much we can model, where we will be making informed intuitive guesses as to how to proceed.
But as I say, I’ve stuck my speculations down in print for all eternity. You can write to me in 10 years time and say “Nyah nyah nyah – see you were wrong, you arty wanker!”
Thanks again for your thoughts.
“My suspicion is that we will never be able to make computations of that complexity (what you’re asking is that we get to the point where we plug in all the data and get a simulation that shows a hurricane of a certain force hitting New Orleans at a particular time – I’m highly skeptical that we’ll ever get that kind of utility out of computers, no matter how many calculations they can do.â€
The “butterfly affect†limits how well certain nonlinear dynamic systems can be modeled. If small differences in initial conditions cause big differences in system outputs then it will be very hard to predict long-term outcomes. Such systems are badly behaved. In other cases, the nonlinear system dampens the initial differences and yields stable states.
The human brain is an example of a nonlinear dynamic system that continually moves between bounded, stable states. The behavior of the system can’t be completely predicted. However, supplying similar inputs tends to produce similar outputs. The human brain is an exceedingly complex, nonlinear dynamic system and yet humans are fairly good at predicting how other people will react. (Communication depends on this fact.) The human brain is a well-behaved system.
Are biological species examples of badly behaved nonlinear systems? Do small variations at the genetic level cause big differences at the phenotype level? Certainly some mutations have drastic affects, including death. But many other mutations have limited or minor phenotype affects. Is a biological species mainly chaotic or mainly stable?
“I believe that in most complex systems, the only computation available to understand the system completely, is the computation in the system itself.â€
When you put the word “completely†in your statement I can agree. For the foreseeable future there will remain a significant degree of uncertainty in computer modeling. However, the natural methods of producing plants and animals result in considerable uncertainty. I expect computer modeling, combined with genetic engineering to yield results much more predictable than natural breeding methods.
The “system itself†will be used to tune the models. By studying yeast, protein metabolic pathways are modeled and gene regulatory networks are modeled. Scientists have begun to explore double gene knockout yeast cultures. There is an ongoing project to examine all single gene knockouts in mice. The computer models will be based on experiments on the “system itselfâ€.
(Note that seeing the outcome of a system is not the same as understanding the system. We can see a baby and not understand how a baby comes into being. When a computer model accurately predicts reality then scientists can claim to understand the process.)
“if humans are to start tinkering with genes we will not be allowed failures.â€
Scientists are already tinkering with genes and they certainly can fail. Bacteria, yeast, plants, and goat milk glands are presently being used to make valuable protein products. Perhaps, you meant we can’t tinker with humans? But the natural process already creates failures. Early genetic engineering will used to correct nature’s mistakes. As the technology improves it will be used more aggressively. There may come a day when the natural method will be viewed as wrong. Good parents may owe their children a good genetic start.
“I stand by my assertion – there will not be one-on-one correspondences with genes to effects.â€
That statement seems different from your earlier statement, “we will not be able to make direct correlations between particular genes and what those genes ‘do’.â€
“One-to-one correspondence†sounds too much like the layman fallacy that a single gene controls each trait independently. The correlation between genes and high-level traits won’t be one-to-one and it will be complex.
“You can write to me in 10 years time and say “Nyah nyah nyah – see you were wrong, you arty wanker!â€â€
Hey, I’m sure we’ll both be wrong. That is part of the fun of making conjectures. Your post and comments are interesting and show that you have been thinking about the difficulty of modeling complex genetic interactions. I appreciate an opportunity to think aloud about this interesting topic.
The human brain is an exceedingly complex, nonlinear dynamic system and yet humans are fairly good at predicting how other people will react.
Oh, yes indeed, but here we definitely use some kind of intuitive gestalt thinking. In fact, I would go so far as to say that all human interaction is based on being able to ‘artistically’ (I use the word advisedly) manouvre through complex inter-human communication space. I submit that it’s by no means a perfect art – some people are really good at it, and some people are not. If we were all good at it, we’d have a lot more stable world, I’d venture.
Are biological species examples of badly behaved nonlinear systems?
I’m not suggesting any particular biological species is a BBNS, but I do suggest that the entire scope of biology and evolution is. I think it is a constant struggle on the ‘edge of chaos’ as Waldrop would say. The biosphere constantly self regulates to maintain the fine balance of working/not working. If we examine living biological entities then they are by definition, successful. But as I said, nature has the luxury of failure, and exploits it constantly.
I expect computer modeling, combined with genetic engineering to yield results much more predictable than natural breeding methods.
Agreed.
Perhaps, you meant we can’t tinker with humans?
No, you’ve misconstrued my meaning, or perhaps I put it badly. What I mean is that nature has had a huge amount of time, and a big sandpit in which to play. I don’t believe that humans, whether in human genetics, or elsewhere, have as much leeway. It is true that we will be substituting guidance where nature has had to use trial and error, but once again, I assert that that guidance will have a creative aspect, as much as a scientific one, if it is to have any value.
Even your example Good parents may owe their children a good genetic start immediately suggests aesthetic judgements, as much as that’s not an emphasized point in biotech, because it opens the FrankenGene door. Blue eyes or brown? Long-legged or large-breasted? Muscular or wiry?
Sure, some things are clear cut – we don’t want our kids to have diabetes from the get go. But that’s only part of the whole shebang, wouldn’t you agree?
“One-to-one correspondence†sounds too much like the layman fallacy that a single gene controls each trait independently. The correlation between genes and high-level traits won’t be one-to-one and it will be complex.
Yes, I was oversimplifying. We agree on the main point.
I appreciate an opportunity to think aloud about this interesting topic.
Also! And thanks.
I’m a computer network engineer / programmer.
Please believe me when I say the Internet is nothing more than a bunch of unreliable machines that can not thing, only follow simple directions.
The chances of the Internet becoming intelligent is pure garbage. Ask an automobile mechanic of the odds of a car becoming self aware and driving itself.
The comments made about the internet waking up one day are made by people with little understanding of computers.
There is NO evidence that the internet has gained any sense of intelligence. There is nothing “unknown” going on inside of computers, EVERYTHING they do is known and understood.
Joshua: Thank you for your comments.
I don’t really think you understand the basis of my arguments, nor have much undersanding at all of non-linear systems. If you did, you would not have evoked the hoary ‘car becoming self-aware’ argument. The internet, and indeed even much simpler computer systems, are most profoundly not as simple as cars. Cars have at the most a few hundred, maybe a thousand functioning parts. Computers can have millions, even billions of simple interactions.
However, when computer systems reach many billions of billions of connections and functions that are someday inevitable, it will simply be impossible to predict what they might do. It does not matter that we understand what all the individual ‘bits’ do, this is profoundly not the point.
There is evidence already in the interactions of many kinds of complex software that this card is in play. Two apps running on a computer conspire together to cause some kind of bizarre behaviour that none of the people involved in writing the software or making the hardware predicted from the outset. You yourself must have encountered these kinds of incidents. That’s emergent phenomena. It does not mean that anything useful results of course; more likely it screws up the utility of the machine. But at the moment the systems are relatively simple. As the systems become more and more complex and the interactions between them become more and more numerous, the magnitude of the complexity explodes beyond calculating.
In the short term this probably means lots of unpredictable behaviour – server failures, software clashes, network blocks. However, my argument is that given a sufficient amount of complexity which will be reached in our lifetimes we will see emergent cogent behaviour in the internet (whatever that ‘internet’ is by that time). We will see things happen that we didn’t predict and didn’t program.
You may think that we will always be able to figure out the interactions that lead to whatever behaviour we witness. But we won’t. I don’t even have to defend this assertion – it is a mathematical certainty; in a system of complex interactions a point is reached beyond which it is impossible to backtrack. I didn’t make this up – you can read about it in numerous published mathematical papers.
Please note that I did not at any time say that the internet would ‘become intelligent’ or ‘wake up’. These are highly emotive and heavily coloured terms that are generally used by uninformed journalists to make such complex ideas seem risible. I said that it would exhibit behaviour that we could not predict and might look to us to be spooky. In other words, that we will not understand entirely what it is doing, and it will do some things we didn’t tell it to.
If your mind is not already closed to these ideas, I suggest you read some work on complexity theory and anything by Ray Kurzweil. And re-read my post (and what I’m actually saying) and the comments on it.
Emergence is a tricky subject. As I said, the results from complex mathematical interactions are counter-intuitive and sometimes bizarre. Understanding the components is not the same as understanding the system. No matter how clever you are.
Wow, this post provoked a lot of thought… probably not nearly as educated thought as some posters’, but still…
Anaglyph:
I really much like your idea of artist-like people being in control of the genetic engineering, mainly because I think most artists (most of the good ones, at least), would also be intelligent enough not to use their power to create weapons (“supersoldiers” and the likes).
Also, I am in kind of a religious crisis right now, trying to find out what I believe and what I don’t… I had been pretty sure for a long time that there was some kind of force controling what is happening in this world; Your approach of putting Natural Selection in that place makes a lot of sense to me.
This whole conversation also invoked thoughts of various Sci-Fi-scenarios:
Your paragraphs about the emergence of the internet reminded me of the “ghosts in the machine” that are mentioned in “I, Robot” (and I’m sure also to some degree in Issac Asimov’s work itself, I only read a few of his books, though, when I was 13 or so, and only for fun).
I think current computers (and even their inevitable improved versions) will not be prone to this phenomenon, but the upcoming quantum computers (the first working one was demonstrated just a few weeks ago, I think) will … the concept of quBits just calls for unforeseen reactions, IMO.
Fly’s statement “There may come a day when the natural method will be viewed as wrong. Good parents may owe their children a good genetic start.” and your response made me think of “Gattaca” (where “natural” children are regarded lesser people – the “hero” goes to great lengths to take the place of an “engineered” person, because he wouldn’t have been able to even get a job otherwise).
I really hope that a situation like that will not arise when the days of human genetic engineering come.
I apologize for only using movies as references (and popular entertainment-movies at that), I know there is few to back their points, but then again, predicting the future is a vague business anyway!
I LOVE this image!
Just stumbled upon this wonderful article.This is extremely exciting.I’ve bookmarked it and will come back routinely.Hope to see more such articles.file_links\xrumer_resource\km_sig.txt,1,L]
Yeah, thanks ‘Karen’. The JREF appreciates your support.