The conversation began on this post about how economic corrections are healthy but quickly diverged from the realm of finance into the more general realm of information theory. I'd recommend reading the comments there, but I want to post here my latest thoughts on the matter because I think they highlight a major reason that I'm skeptical about biological evolution.
Information is conserved, in a manner similar to energy and mass, except with losses over time due to entropy. Information can be thought of as the property that distinguishes useful energy from heat. To believe that biological life is due to evolution is to believe that natural selection and random mutations are transferring information from the environment into biological matter, with some loss due to entropy. (Due to entropy, there is an ever-decreasing amount of information in the universe.) This perspective reveals two difficulties facing the theory of biological evolution:
1. There doesn't appear to be a lot of information in the universe. There are a very limited number of physical laws that pretty much explain everything. We probably don't know all the laws yet, but I suspect if we did they could be contained in a single book. Cosmic background radiation exhibits almost no information whatsoever, for example. In order for information to be transferred from the universe to life through evolution, there must be enough information in the universe to start with.
Most naive experiments with evolutionary algorithms begin with a few simple rules and hope to generate complex results from them, but cannot do so. You don't get more information out of a system than what you put it, and you generally get less.
2. If there is a lot of information in the universe, where did it come from? This is a different question from "where did all the energy/matter come from?" because the presence of energy and matter does not necessitate the existence of information. As I said earlier, entropy transforms useful energy (with information) into waste heat (without information). Like our constant supply of energy and matter, our ever-decreasing supply of information came from somewhere, but where?
Random mutations don't net add information. They may increase information locally by moving it from somewhere else, but every transaction is degraded by entropy so in the net there's a loss of information. As with Brownian motion, the more interactions in the system the faster the whole tends towards the average. Contrary to the arguments of believers in biological evolution, longer time spans and larger populations actually make biological evolution harder to believe, not easier.
In my experience, information is never created without intelligence. I've never seen an example of a non-intelligent system that creates information, and I think the burden of proof lies with anyone who argues that it's possible. I'd even argue that, as a useful definition, an "intelligent system" is a system that creates information rather than just moving it from one place to another. I believe that God and humans fit this criteria, but I'm not convinced that any other systems or entities do.
Update:
There seems to be some confusion in the comments about what information actually is. Let's use an illustration. Wrote jez, suggesting a way that information might be created without intelligence:
7) a short computer program generates a continuing list of prime numbers
However, all the information required to represent every prime number can be written down in a very simple recursive form. Actually doing the work to process that function and generate a list of prime numbers doesn't create any new information. In fact, unless your list is infinitely long it contains less information than the function itself. My argument is that mathematics and the laws of physics encode all the information that makes our universe work. If biological evolution really happens, then the information must be derived from those sources. See again my problems with this theory, above.
Oh now be honest, MW. You're not merely "skeptical" of biological evolution and natural selection.. you wouldn't give it any legitimacy even if your burden of proof on the matter were met.
Just because information is like energy in some ways, this by itself does not mean that no new information can be created. You can't say "There are some similarities between A and B, therefore if blah is true of A, then blah is true of B".
I don't know enough about the properties of information to know (or even educatedly guess) whether it can be increased overall by non-intelligent processes, but intuitively I don't see what the problem is. The environment in which an organism lives is full of challenges and opportunities regarding the preservation of live. Of the infinite possible mutations that organism could generate in its offspring, some would make it more or less capable of coping with those challenges, some would make it exploit those opportunities more or less, and so some mutations tend to get preserved. Again, I don't know very much about information, but it seems to me that you're not so much creating new information as you are reflecting information about the environment on the things that try to live in it.
I'll go read this, this and this and I'll get back to you. That second link is a critique of Dawkin's critique of the information-theory argument against evolution. The third link mentions a book where a whole chapter apparently is dedicated to showing how evolution creates new information. We'll see.
Mark: Not really.
BM: I'm using the metaphors to try to explain information theory, not as a basis for argument.
What you're describing in your second paragraph is a local increase in information, which is possible but not likely sustainable over time. Plus, where does the information come from?
Where does it come from? From the environment.
Some organisms will live to have children, others will not. That's one bit of "input" (approving or disapproving) from the environment for every single living thing that has ever lived. Or you can choose the input to be which organisms have grandchildren (i.e. have offspring which eventually have offspring) and which do not. In either case, it's one bit per organism.
Genomes may contain more bits than this one-bit-per-generation model accounts for (since each nucleotide pair is 2 bits, having 4 possibilities). That's because mutations can add or change many bits at once. So the genome does not have as much information content as all those nucleotide pairs would make it seem like. It's the phenotype, the gene expression, the observable characteristics, that get selected for. Which characteristics are good for life to sustain itself in this given environment? That information reveals itself as more and more random mutations (and gene combinations, when you have sexual reproduction) have a go at it. Once you have life and an environment around it, the information about what characteristics are good become implicit in that system. Like how you can plot a curve or a surface from a simple equation, but that curve has all kinds of interesting properties (such as how a parabola focuses lines bouncing off it) implicit in the equation. If you plot a parabola, where does the information "come from" about where the focus point is? That information was implicit in the equation, it was there to begin with, as soon as you asked a question about the focus point the answer existed. Same thing about life and an environment: The information was already there, since the environment was already there. Asking what characteristics will be selected for is like asking where the focus point of the parabola is. You're not really "creating" information. I know it's an imperfect analogy but I hope you can see what I'm saying.
You say that the universe did not have much information to begin with. I say that is incorrect. The universe has LOTS of information, since there are many questions to be asked about many systems (i.e. all the properties of all the subatomic particles in the universe, and of all groups of subatomic particles, groups of all sizes, from neutrons to bacteria to asteroids to galaxies), and they all have answers.
That third link I mentioned leads to a book that spends a whole chapter on how evolution "generates" information (or, rather, moves it from the environment into the genome). I managed to find and download the chapter online. It contains a mathematical proof that, while asexually-reproducing species can indeed only get one bit per generation, sexually-reproducing species can get sqrt(G) bits per generation, where G is how many bits their genomes have. I'm not going to pretend that I understand the proof, but this is an Information Theory textbook so it's probably pretty reliable.
If you want to check it out, go to
http://www.amazon.com/gp/reader/0521642981/ref=sib_dp_pt/104-6133955-2075154#
search for "evolution" and click "Go". The chapter is on pages 269 to 280. If the reader stops letting you click ahead (which it will after you read 3 pages or so), just search again and click on the link to the next page you have not read yet.
Information's a bit abstract. I'd like to re-ask an earlier question: what stops information-rich mutations from happening? Is the answer "nothing"? Do you believe that natural selection is unable to distinguish and preserve information-rich mutations? (they must also increase fitness, obviously).
I'm trying to understand what your saying about information, and compare it with what i remember from my undergrad days. I find examples and counter-examples are always useful. Consider the following events:
1) a ribosome builds a protein from instructions in an RNA sequence
2) a chemical reaction forms an ordered crystal from dissolved reactants
3) a star builds heavier elements out of hydrogen
4) a tornado forms
5) a more persistent weather phenomenon such as the Gulf stream emerges
6) an embryo develops from a fertilised ovum
7) a short computer program generates a continuing list of prime numbers
In each of those events, is information created (locally); if so, where does it come from?
I've made a longish list of scenarios in the hope that you'll find a selection of them useful -- I don't particularly intend for you to deal with them all! Or, if you have examples which better illustrate the concept, obviously use them instead. But I'd be grateful for some concrete examples -- the danger with keeping it abstract is that we could easily be talking about different things while using the same words.
For what it's worth, your objections sound like a veiled appeal to 2nd law, which is a long-refuted argument against evolution. The earth is not a closed system. Entropy can be decreased locally if energy is released to the surroundings (eg. a fridge).
BM: the states you mention of every particle in the universe is not really information. If we know the macrostate of the whole universe, the additional number of bits required to describe its microstate is called the reduced Gibbs entropy. It's kind of the opposite of information. When MW talks about "information loss due to entropy", I think that's precisely what he's referring to.
BM: "That third link I mentioned leads to a book that spends a whole chapter on how evolution "generates" information (or, rather, moves it from the environment into the genome)."
Yes, that's my whole point. That's all biological evolution can do. I wouldn't call that "generation", just transferring information from one place to another. Where did it come from originally?
jez: We're not exactly talking about energy, that was intended to be an analogy. Let's look at a couple of your examples:
1. No information is created. The information for the protein existed in the RNA already.
2. No information is created. The amount of ordered matter increased locally, but the information already existed in the laws of chemistry/physics.
3. No information is created. The information is entirely encoded in the laws of physics.
6. No information is created. The whole process is encoded in DNA and the various laws of physics, etc.
7. No information is created. The information is wholly contained in the program, and just written into longer form by actually recording the prime numbers.
I shouldn't make as many assumptions as I did, so I'll try to be more clear. Take number 7 as a concrete example. All the information required to represent every prime number can be written down in a very simple recursive form. Actually doing the work to process that function and generate a list of prime numbers doesn't create any new information. In fact, unless your list is infinitely long it contains less information than the function itself.
"mathematics and the laws of physics encode all the information that makes our universe work. If biological evolution really happens, then the information must be derived from those sources."
That sounds right to me. Again, I have not thought much about this, and I still have some reading to do, but it does seem to make sense that evolution moves implicit information from the universe (like the implicit information in a computer program or in an algebraic function) into a more explicit form in the genome and its expressions.
Where does the information "come from"? It's implicit in the universe. It's a property of particles and groups of particles given the laws of physics (and thus chemistry and biology). That's like asking "Where does gravity come from" or "Where do all these atoms come from".
I can see three possible answers: "It's always been around. The universe goes through cycles that include Big Bangs", or "There are an infinity of possible universes and they all exist, spontaneously arising from the quantum foam, each like a bubble in a vast ocean of a sparkling liquid", or "God did it" (that last one of course being no explanation at all since the next level of questions - where God came from and why - are swept under the rug as unanswerable, so we're left in the dark about why things turned out as they did, about where anything comes from).
I find it interesting that energy 'wants' to spread out and distribute itself evenly within a given area. Matter is quite the opposite. It 'wants' to clump together. But in doing so, it converts to energy and spreads out... eventually.
Matter is far more complex than energy. Hmmm...
One would do well to read the Wikki page on Information Entropy.
I have always thought that information acts more like energy than matter. It tends to want to dissipate... the more you cram in one place, the shorter the life span of the vessel. Just look at the storage medium of today. Sure, we can pack a whole lot of bits in one place but mean time between failure is a blink of the eye compared to stone tablets. Stone has terrible data density.
I was judging at a science fair one time and a kid duplicated the data correction scheme inherent in DNA and applied the algorithm to file storage. You could literally randomize whole sections of the file and it could recover. Zero loss. DNA has a lot of redundancy and fail safes built in to it. It has too, because there is a lot of information packed in there and it 'wants' to not be packed to tightly.
It is incredible that copies of creatures are made SO perfectly as often as they are. It's a shock that genetics don't suffer from information entropy any more than it does, and here we have people claiming that we not only have a lossless system, but we actually have a system with a net GAIN in information density over time.
Well, which is it? Information acts like matter, or energy? neither? Explain your answer please.
I find it interesting that energy 'wants' to spread out and distribute itself evenly within a given area. Matter is quite the opposite. It 'wants' to clump together. But in doing so, it converts to energy and spreads out... eventually.
Matter is far more complex than energy. Hmmm...
One would do well to read the Wikki page on Information Entropy.
I have always thought that information acts more like energy than matter. It tends to want to dissipate... the more you cram in one place, the shorter the life span of the vessel. Just look at the storage medium of today. Sure, we can pack a whole lot of bits in one place but mean time between failure is a blink of the eye compared to stone tablets. Stone has terrible data density.
I was judging at a science fair one time and a kid duplicated the data correction scheme inherent in DNA and applied the algorithm to file storage. You could literally randomize whole sections of the file and it could recover. Zero loss. DNA has a lot of redundancy and fail safes built in to it. It has too, because there is a lot of information packed in there and it 'wants' to not be packed to tightly.
It is incredible that copies of creatures are made SO perfectly as often as they are. It's a shock that genetics don't suffer from information entropy any more than it does, and here we have people claiming that we not only have a lossless system, but we actually have a system with a net GAIN in information density over time.
Well, which is it? Information acts like matter, or energy? neither? Explain your answer please.
What is it about any living thing that contains information that does not exist already in chemistry or physics? Respiration exists already (given chemistry) just as much as "23 is prime" exists (given the definition of a prime). What is it about a evolving a pancreas or a feather that is different from calculating the next prime?
My list of events all seem to be counterexamples. So, I guess I'm asking for a positive example of "information creation".
MW: just noticed your update. It misrepresents my intent somewhat. I didn't supply that list expecting any of them to be creating information, except for maybe the weather examples -- phenomena emerging from chaotic systems is maybe a case where the bare laws of physics are not the whole story. Although no magic happens and every element of the system obeys well defined laws, it is not physically possible to measure initial conditions such that the system can be predicted arbitrarily far into the future. In other words, knowledge of the speed of light and the nature of the Higgs Boson would never allow you to predict the coastline of Africa, and that could maybe constitute emerging information.
But the examples are not intended as challenges, just as talking points.
If energy is only an analogy, then what is the basis for your assertions about information loss? I assumed that you derived them from information's relationship with entropy via Boltzman's constant. Are you justifying them (your assertions) some other way?
Bernardo: Consider higher-order examples of information. Where did the information in "Moby Dick" come from, for example? Does the book derive naturally from mathematics and physics? I'd contend that human intelligence (/ free will) are capable of adding information to the universe in a way that non-intelligent processes are not.
A "strong" artificial intelligence is commonly required to be creative (among other things), and creativity implies the creation of information. At this point, only humans (and God) are creative.
Unless you'd argue that writing "Moby Dick" didn't create any new information in the universe, but I don't think that's defensible. "Moby Dick" might exist randomly within Borges's library of Babel, but in the real world it seems clear that every book (even the not-so-great) contain information that didn't exist in the universe before.
jez: Do you mean my assertion that information is lost to entropy, see Information Entropy, particularly "Relationship to thermodynamic entropy".
I feel like this conversation is disjointed because I'm only able to get on here once a day or so, so sorry about that :)
Michael, I have to admit that you could be right, as far as I know. My intuition says you're wrong (Yes, it does say something to the effect of Moby Dick being inherent in the universe, since it is an arguably-inevitable expression of an intelligence, which is a feature that evolved naturally in response to a natually-occurred environment... if you're a naturalist like me, who believes that the universe is one of an infinity of books in a Borges-style library), but I do need to learn more about information theory before I am able to discuss this any further, before I can answer you with more than imprecise analogies and "What I think feels right". So I'll stop wasting your time on this one, for now. And I do mean that in a friendly and sincere way. But when I have learned enough to put together a scientifically-based argument for where information comes from in a naturalistic universe, I'll ask you what you think of it, and I look forward to exploring this issue further.
The Embryo is an interesting example. I have been thinking about the creation of a new person: the process of DNA creation may be fairly deterministic, but the creation of a soul and a brand new consciousness is where the laws of the universe fail. It is new information, but where does it come from? I have an idea, but I wonder if anyone has any that attribute it to something within the universe.
Sorry, I didn't see the "Moby Dick" thread; I think my comment is along the same lines..
When a human writes a novel he is not creating net information because all the while he is dissipating energy. Because Boltzmann's constant is so small, the entropy he contributes to the surroundings amply offsets the information content of his novel, however long it is.
As for "creative", that's not a precise enough term to be of great use in this kind of discussion imo.
The same is true of a DNA strand, all the molecular processes dissipate easily enough energy to offset the local increase in ordering as eg. binding sites emerge on a strand (or the loss of information as binding sites are abandoned).
Take a look at this experiment.
MW: do you have any comments about the weather systems? would you say that micro-states (noise) contribute to highly visible macro-states, a la the butterfly effect in a way which is fundamentally not predictable? Is this local information gain?
B: I'm sure I need to do more reading too....
jez: I didn't comment about the weather systems because I'm not sure how they fit in. Intuitively they seem like localized order that arises from a chaotic process.
"the creation of a soul and a brand new consciousness is where the laws of the universe fail"
Says who?
I don't know; I guess I am only assuming others experience this process in somewhat the same way as I do.. That no matter what science reveals about processes and matter and whatnot, my sense of self, the existence of self-recognition remains distinct from that. I would like to add that it didn't just poof and appear one day (if it did, when?); so, by induction, it has existed since I was conceived of (though my capacity to remember the early days is limited by biology).
Mpayne: if your induction argument were valid, why stop at conception? Could you not have existed previously, and your corporeal body be a mere vessel for your intellect?
I identify myself and all those things you mention with my brain. If I cut off a leg or a finger, or remove one of my lungs, I remain the same person in myself. However, patients who have injuries to the brain, or tumors removed, often go through personality changes, suffer from reduced cognition etc. (I'm thinking of one fellow I knew who, among other things, became politically a little more left wing following the removal of a tumor.) That leads me to think that the brain is involved, either as the source or as an aerial for receiving my soul/spirit/personality/whatever. (It is also the interface at which science's insights into matter and processes and whatnot meet with our sense of self.)
At conception, I did not have a brain...
The human brain is still a pretty big mystery. There's a lot we don't know about it. I think it would be quite mistaken to rule out or dismiss the brain as having anything biologically or chemically to do with our "soul", etc.
MW: your description of the information content of a list of prime numbers is, I think, the algorithmic or Kolmogorov complexity of the string of primes. Although both can be measured in bits, this is not the same thing as the Shannon information in that list, so be careful about applying your ideas about information loss to them as if they were the same quantities. No receiver knows every prime, even if he's running the computer program, so for any receiver there exists a prime which contains some Shannon information. When you talk about "information", are you consistently referring to algorithmic complexity?
Emergence is a familiar property, take what we've been talking about: thermodynamics. We can follow your scheme of considering the laws of physics to be the total complexity of the universe. However, in if we want to ignore the mostly irrelevant behaviour of every atom, we consider statistical quantities such as temperature, pressure etc. We develop algorithms based on those quantities, and the emergent property of entropy arises. Is this new complexity? It isn't a new fundamental law in the same class as gravity or QM, but it is a theory important and true nevertheless. It's even the basis for your objection to evolution.
The algorithmic complexity of the macro theory of thermodynamics is pretty obviously less complex than the complexity of an equivalent atomic theory, since you need to include the reams of initial condition data in your atomic theory algorithm. However, the illusion is emerging complexity because no-one cares about individual atoms, and entropy feels immediately like an algorithmic thing rather than a data thing.
So too does evolution arise from the impossibly detailed atomic model. Because we prefer to think about organisms and behavior than atoms, it looks like things are gaining complexity.
The question remains: what is it that prevents information-rich mutations from happening? It seems obvious to me that informative mutations have may occur unimpeded and that natural selection (success or failure to produce offspring) is a huge information input compared to the whole life of the organism.
jez: Yes, I think I've been using "information" somewhat interchangeably with Kolmogorov complexity. Definitely different than Shannon's concept of information, true, but I'm not sure if/how that affects the relevance of his observations to the topic. I'll have to ponder it some more.
As for your question: nothing prevents "information rich" mutations, but one would expect them to be exceptions. Longer time periods and larger populations would expect to see regression to the mean.
re complexity loss over time: I'm not certain either. Intuitively, I don't think it does, ie the sum of the complexity of the mutation with the complexity of the original is an upper bound on the complexity of the new sequence (just add a special case to the algorithm).
OTOH the complexity of the universe is physics + conditions (and the nondeterministic effects of quantum mechanincs might even be significant too -- the exact moment a nucleus decays is not unimportant, especially since that's the kind of thing that causes a genetic mutation), so there doesn't need to be a net gain.
The rich mutations are rare compared to meaningless or deleterious mutations, but of course I'd argue that selection pushes things towards their niches, latching onto the few informative contributions and discarding the counter-productive ones.
"The mean" would be a purely mineral universe? Is that any less complex at the atomic level than one with biology?
jez: I think the "mean" is, eventually, just evenly distributed energy, or a giant black hole, or something. I've read somewhere that a black hole has only two properties: location and mass. Or maybe they spin too (?). Anyway, they're very simple.