Home » Intelligent Design » ID and Indirect Measurements

ID and Indirect Measurements

One criticism of ID that tends to come from those who might normally share our worldview (such as Thomists) is that ID attempts to measure meaning, while meaning is unquantifiable.

I argue that this is partially correct – we currently don’t know how to quantify meaning or meaningfulness. Unlike others, I am not about to give up the search for a way to do this, but nonetheless I do agree that at present it is unquantifiable. However, ID doesn’t measure meaning.

Instead, ID measures an indirect indicator of meaning – CSI, active information, etc. Indirect measurements are nothing new in science. In fact, thermometers are a great example of an indirect measurement. We can’t measure temperature directly. So, instead, we take mercury, which reliably expands when its temperature increases, and put it in a tube. We use the expansion of mercury to tell us that the temperature has increased. Likewise, ID doesn’t measure meaning, but instead takes an indirect measurement – dramatic increase of active information, for instance, tells us that someone has added information about the search space.

However, I don’t want to necessarily imply that ID measurements are equivalent to measurements of meaning. It has not been shown that meaning and CSI, for instance, directly vary together. More probably, CSI is a measurement of one aspect of meaning which probably does directly vary with it. However, the current claims of ID is simply that, past a threshold, these measurements accurately indicate the presence of meaning.

Eric Holloway has argued that ID-like reasoning is required for rationally believing in other minds. His argument is basically this:

  • Searle’s Chinese Room argument shows that there is a separation between processing and understanding – one may in fact process without understanding
  • However, this leads to a problem – if our words can be processed without being understood, how do we know that other minds are actually present and understanding us, rather than just processing what we say?
  • CSI allows us to recognize intentionality. Even though it doesn’t give us access to the meaning or even the semantics, it does tell us that the target required intelligence to achieve.
  • Therefore, we have an indirect measurement that allows us to infer that the message originated using intentionality.

In short, indirect measurements have often been used to signal the presence of things which are not directly measurable. CSI and active information, while they are not direct measurements of meaning, do seem to serve as reliable reporters of meaning.

  • Delicious
  • Facebook
  • Reddit
  • StumbleUpon
  • Twitter
  • RSS Feed

18 Responses to ID and Indirect Measurements

  1. We still don’t know the meaning/ purpose of Stonehenge…

  2. the intelligent agent need not assign a meaning to the pattern….

    Neither CSI nor semantic information presuposes the other. This in my view is a tremendous asset of CSI, for it allows one to detect design wiouth necessarily determining the function, purpose, or meaning of a thing that is designed
    ..
    Mereology and statistics, not syntactics or semantics, are the rock-bottom foundational aspects of information. Indeed, a scientifically fecund study of information can proceed soley on the basis of mereology and statistics….

    The sufficiency of mereology and statistics as a foundation for information and the dependence of CSI solely on these foundational aspects means that CSI bypasses many an impasse that semantic information has had to confront.

    Bill Dembski

    Available online for free:
    No Free Lunch Highlights

  3. Very interesting issue, and I’m not sure where I come out on it yet. You are talking about “meaning,” which is one way to look at it. I think it is sometimes difficult to parse meaning, however. After all, what does the Mona Lisa “mean”?

    Just a thought that has been percolating in my head more recently: is the key to design “intention” rather than “meaning”? In other words, is there any design without intention? Intentionality seems to apply both to things like the Mona Lisa or other work of art (which may or may not have a particular meaning), as well as to functional machines or writings that convey a particular message. Isn’t it the fact of intention which causes the intelligent agent to act and produce the design?

    Or maybe we can get even more basic and say that the instantiation of information in a physical system is the key. (Could I design by accident, meaning without intention?)

  4. Joe,

    Therefore, Stonehenge was not designed. Jk.

  5. Collin,

    PLUS, well, we know that mother nature can make rocks, all kinds of rocks. Stonehenge is just rocks, different kinds of rocks. So parsimony would have mother nature building it and humans coming in later. :roll:

    Now we don’t need know the meaning cuz there ain’t one.

  6. Sal:

    Personally, I cannot accept that CSI does not need meaning. This is a problem I have always found in Dembski’s approach. I don’t believe anyone can really define specification without the concepts of meaning or function (which are just two different aspects of conscious cognitions).

    I understand (but I could be wrong) that Dembski aims at defining specification in objective terms. I don’t believe that such a thing can be done.

  7. gpuccio,

    Specifications are subjective. It would seem that given their subjective nature, designs could not be identified. The requirement is that the subjective specifications are not post-dictive (after the fact).

    Perhaps to illustrate, when the French soldiers came across the Rosetta stone, I would almost bet none of them could understand the ancient language it was written it, much less the meaning of what was written. Nevertheless they were able to infer the engravings on the stone were designed, not random scratches on a rock.

    That sort of accords with my ideas about detecting design. To understand a design requires understanding its meaning. One can detect a design before understanding it. That is the formulation for a lot of Bill’s work.

    Sal

  8. Personally, I cannot accept that CSI does not need meaning.

    I believe the point is that we don’t have to know what that meaning is before we determine design.

  9. johnnyb:

    Just my view:

    a) You say:

    “Searle’s Chinese Room argument shows that there is a separation between processing and understanding – one may in fact process without understanding”.

    I agree, but only in part. A conscious agent can process only if he understands the rules of processing. He may not understand the final meaning, but he has to understand those intermediate meanings that are sufficient to his processing task.

    On the contrary, machines can process material systems, even those bearing objective information, witout any conscious understanding of anything, because they operate by law, according to the way they have been programmed.

    b) You say:

    “However, this leads to a problem – if our words can be processed without being understood, how do we know that other minds are actually present and understanding us, rather than just processing what we say?”

    No creative processing of words (or any other thing) can happen without conscious representations. CSI is only generated by conscious represebtations, and nothing else. Spoken or written discourse, unless it is only a goofy repetition or remix of pre-existing discourses, incarnates some unique conscious experience. We know we are conscious, and we intuit and infer that others are conscious, and one of the most powerful clues to consciousness in others (although not the only one) is their ability to generate CSI at will, just as we do.

    c) You say:

    “CSI allows us to recognize intentionality. Even though it doesn’t give us access to the meaning or even the semantics, it does tell us that the target required intelligence to achieve.”

    I am not sure what you mean. Intentionality means recognizing a purpose or function, that are conscious representations and are semiotoc. We may not understand the full meaning of what we see, but if we recgnize intentionality we are indeed understanding some meaning. Otherwise, we could never define specification and measure the specification linked complexity.

    d) You say:

    “Therefore, we have an indirect measurement that allows us to infer that the message originated using intentionality.”

    THat is not very clear to me. My point is very simple:

    1) We need to define a definite specification for the object. That is the first step. Without that, nothing can be done.

    Specification can only be recognized by a conscious being. Either it is a meaning (like the content of Hamlet) or a function (like the function of a protein), it is a semiotic entity, the form impinted by a conscious representation and that only a conscious perceiver can recognize when the same, or a similar, representation is evoked in his consciousness by that form.

    We don’t measure specification, be it a meaning or a function. We just perceive it.

    2) So, what do we measure? That is simple, too. We measure that complexity linked to that specification. IOWs, we measure the minimum complexity that can evoke that specification (convey the meaning, or effect the function). IOWs, we measure the improbability of the specification, how unlikely it is that it may arise in a random system, without any conscious intervention (without design).

    Why do we measure that? It is not because design needs to be complex. Some designs are very simple, and yet they are fully intentional and designed.

    The problem is, as we interact only with the object, and not with the designer or the act of design, we are inferring design indirectly (as you very correctly state). Our first step, recognizing specification, is simple, but still we could be wrong: an object can always evoke in us a meaning or a function, but still they could be a pseudo meanign or a pseudo function, in the sense that we see them in the object, but they were never designed, they were never consciously represented by anyone and then omprinted into the object. They are in the object by chance, as one of the many possile forms that object can assume in the system of which it is part.

    That’s why specified complexity is necessary: a meaning or function that needs, let’s say. more than 500 bits to be expressed through a material system never arises by cahnce in any system. We know that empirically, and it is also perfeclty reasonable if we understand the laws of probability and the limitations of pur physical universe.

    So, we just measure the complexity, but not any complexity: just the complexity needed to convey that meaniong or function we have recognized and defined. And we do that to be really sure that our recognition of meaning and function is not a mere error, an illusion: we want to be sure that that meaning or that function was in the beginning conceived in a consciousness, and only after that it was imprinted into the material system.

    Specified complexity gives us complete assurance of that.

  10. Neither CSI nor semantic information presuposes the other. This in my view is a tremendous asset of CSI, for it allows one to detect design without necessarily determining the function, purpose, or meaning of a thing that is designed

    Likewise “Shannon Information.” It doesn’t tell us what the meaning of a message is, and the measurement is independent of any meaning in the message itself. But it’s pure idiocy to argue (as some have) that Shannon demonstrated that information can be meaningless.

    Given the association between CSI and Shannon information, I question that Bill would argue that CSI does not need meaning.

  11. Sal:

    I don’t agree.

    Specifications are subjective, but they can be made objective. For example, I can subjectively understand what the function of some protein is, and then I can give an objective definition of that function, and a way to ascertain or measure it. so that anyone can verify if a protein has that function or not. That is what I call “subjective objectivity”: something that all subjective perceivers can verify.

    For instance, if the function is to accelerate some biochemical reaction, I can define as functional any protein that can, in defined lab settings, accelerate that reaction at least at predefined some level. Then I can, in principle, measure how many proteins in a certain search space have that function.

    You say:

    “The requirement is that the subjective specifications are not post-dictive (after the fact)”

    Again, I don’t agree. If I give a good objective definition of a function, it does not matter that it is post-dictive. For instance, I define the function of a protein as above. Why do I do that? Because I have seen a protein that does exactly that. So, my definition is post-dictive. And so?

    The fact remains that i can test a random library of proteins of that same length, and meausre if the function is in them or not. The function is measured objectively. That definition can be used in any context.

    I will make a very siple example to show that a subjective definition of function can be used to objectively measure CSI and infer design.

    Let’s say that the object I observe is my tablet. It is a material system. I can define many functions for it. One could be:

    a) any object that can perform a specific list of calculations (word processing, spreadsheet functions, digital pictures visualization and processing, and so on). I can also give very detailed definition of the functions that must be available, and offer objective ways to measure them in any given object. With that definition, my tablet will have the required functions, and the complexity linked to that set of functions will be very very high. We can securely infer design for an object that does all that, including my tablet.

    But I can also define a simple function, such as paperweight function. I can again define the function of a minimal paperweight for my desktop, such as any smooth physical object that weighs at least scuhc and such, and can keep my papers still on my desktop. Again, my tablet could well comply with that definition. But, in this case, the complexity of that function is very low (any solid object of a certain weight will do), so I cannot infer design for my tablet using that function and its linked specified complexity.

    Let’s go to the Rosetta stone example. Please, read, my post to johhnyb, the part about Searle’s Chinese room. OK, in some cases we don’t need to understand the full meaning of what we are seeing to define specification, but still we need to understand a partial meaning. For instance, we can recognize that we are seeing characters, even if we don’t know what charcters they are, or teh language they are expressing. There is, however, some understanding of meaning or function, although they may not be the complete meaniong and function. Still, what we undesrtand can be objectively defined, and measured in any given object accordin to our definition, and the complexity tied to that definition can still be measured, at least in principle. We could ask, for instace, how likely it is to have any material state arise in a random system which has exactly that kind of objective properties that have evoked in us the idea of design in the Rosetta stone.

    So, I do believe that one can never detect a design before understanding at òleast part of its designed properties.

  12. Sal:

    I don’t agree.

    Well, I’m glad we can disagree on a few things.

    I’m open to other means of detecting design. I think studying mereology and statistics will arive at some methods of detecting a small subset of designs. I don’t mean to imply those are the only methods.

    Sal

  13. One criticism of ID that tends to come from those who might normally share our worldview (such as Thomists) is that ID attempts to measure meaning, while meaning is unquantifiable.

    I agree with the Thomists on that. And I see that as the Achilles heel of ID.

    I argue that this is partially correct – we currently don’t know how to quantify meaning or meaningfulness. Unlike others, I am not about to give up the search for a way to do this, but nonetheless I do agree that at present it is unquantifiable. However, ID doesn’t measure meaning.

    That seems to be a strange view.

    On the one hand, any intelligent designer would need to be able to quantify meaning, for otherwise it would be impossible to design creatures that experience meaning. The inability to quantify meaning is what is currently holding back AI from designing artificial persons.

    On the other hand, if you are successful in quantifying meaning, you will have provided the final step for completing a materialist account of everything.

  14. NR:

    I agree with the Thomists on that. And I see that as the Achilles heel of ID.

    So you think ID attempts to measure meaning?

    Did Shannon measure meaning?

  15. Neil:

    I have repeatedly answered about that. ID in no way tries to measure meaning. Meaning cannot be measured. It can only be understood.

    On the one hand, any intelligent designer would need to be able to quantify meaning, for otherwise it would be impossible to design creatures that experience meaning.

    To quote you, that seems to be a strange view. Whoever said that the designer “designs creatures that experience meaning”? The point in ID is that the designer designs biological information. It is not biological information that experiences meaning. It is consciousness that experiences, not our proteins. This is the usual equivocation of reductionism, and I am in a sense amazed to hear that from you.

    Just to be clear, a complex eye (biologically designed) is certainly necessary to perceive the outer world through light, but it is not the eye that perceives. The same can be said for the optical nerve, the brain, or any other material system.

    Biological complexity is designed. Consciousness is not designed. It just exists. It is not complex. The perceiving self is simple.

    The inability to quantify meaning is what is currently holding back AI from designing artificial persons.

    Or rather, let’s say that “the impossibility to quantify meaning is what will always hold back AI from designing artificial persons”.

    On the other hand, if you are successful in quantifying meaning, you will have provided the final step for completing a materialist account of everything.

    Which is, in fact, an impossible task.

    I had already answered you on these lines here, at posts # 31 and 32:

    http://www.uncommondescent.com.....nt-design/

    but I think you have not commented on them.

    I hope you will comment on these points here, especially on the fundamental point that in no way ID tries to measure meaning.

    Just to be clear, I want to clarify that what you can certainly measure are the effects of that special kind of meaning that is a function. As I have said many times, what we can and must do in ID is define a function (so that any conscious observer can verify objectively if it is there or not), providing a quantitative way to evaluate its effects. That is the functional specification. We measure the effects of the function only to verify if “the function” is there or not. But we never measure “the function”, because a function is a semiotic entity, like meaning.

    What we definitely measure in ID is the complexity that is necessary yn the material system to ensure that the function, as defined semiotically by us, be present in the system and may give the defined effects in the defined quantity.

    Again, I believe this is a very specific answer to your generic statement that “the arguments about CSI… are based on the conflating of mechanical causation with information processing”. This is simply not true, and I would like that you tried to support your statement in some way..

  16. Sal:

    I’m glad, too, that we can disagree on a few things. :)

    But, as you say that you are “open to other means of detecting design”, I would like to ask you if you really think that we need mereology and statistics to define the function of a protein, a property that is usually clearly stated for most known proteins in any protein database.

    Do you really believe that biologists who daily assess protein functions do that by mereology and statistics? Or just by looking at the results of simple experiments in the lab, and understanding what they mean?

    And, after all, definign the functions of protein and measuring their functional complexity is all we need for the main argument in ID, as clearly stated in this recent post:

    http://www.uncommondescent.com.....ein-folds/

  17. Neither CSI nor semantic information presupposes the other.

    Sure it does.

    There is no such thing as meaningless information. Meaningless information is an oxymoron.

  18. Mung:

    “There is no such thing as meaningless information.”

    Then what do you call “news about the Kardashians”?

Leave a Reply