|
Author
|
Topic: Convergence or Divergence?
|
Jay
Member
Member # 268
|
posted 27. June 2002 19:28
The biological world is full of proteins that share sequence and structural similarities. Any protein you find is likely to share significant sequence and structural similarity to dozens of others, and even to other protein families with completely different functions. In today's scientific literature, it is en vogue to infer that raw similarity (simple side by side comparison of two systems) means homology, or relation by descent and subsequent divergence. This automatic labeling has become such a rubber stamp that the very term 'homology' has become misused, as people talk of proteins being x% homologous. Whenever two proteins are found to share significant sequence or structural similarity, one can be reasonably sure that relation by descent will be automatically inferred.
But how is this? Why can we automatically say that the sharing of some similar information means physical descent relationship? There is no written guarantee that similarity means descent relationship. This is inferred.
Suppose that we compare two large proteins. They are very different in structure and function, but, they each share the 3 amino acid motif 'abc' somewhere within them, but are otherwise completely different.
Now, in this case, what are we to call this? Is this divergence from a common ancestor? Or is it convergence? Likely, we'd run this through mentally and say that it's probably convergence since a three amino acid similarity is not a good argument for relationship here.
Suppose now that we find two proteins that share 95% amino acid identity and identical function. What do we infer then? We again run a quick judement analysis and assume homology since it is very likely (in our minds) that such high sequence similarity indicates relationship.
But now what if we find two large (say, 500aa) proteins that are very specific, complex, and precise in their work (each does a very different job), and find that they are very well conserved within their types for, say, the billion years that they have been around. We find, however, that they share several short, identical amino acid motifs (say 3 motifs of 10 aas). But otherwise, they have very different, well conserved sequences and functions. Basically, they are robustly different except for a few shared features.
For perhaps an even better example, one of two proteins that share no sequence similarity but are almost identical in form, please see a previous post of mine here:
http://www.arn.org/cgi-bin/ubb/ultimatebb.cgi?ubb=get_topic;f=1;t=001728
What do we make of this now? Is this convergence or divergence? When comparing distant species, biologists often confront systems which are very different, and by all that we can see, always have been, but nontheless have some very significant similarities. And as pointed out, the usual response to these is to take their comparisons at face value and claim homology of they simply align in a statistically significant way.
However, automatic labeling aside, it seems that good, thought-through claims to homology must rely on things other than simple sequence or structure comparison. I propose that good evaluation and assignment of homology would need to compare the likelihood of two evolutionary scenarios:
1.) The chances that chance+law could independently produce two convergent proteins with such smilarities.
2.) The chances that the two proteins could have diverged into their respective types from an ancestor.
It would need to compare these scenarios in order to rule one out and favor the other, as both are theoretically possible.
Such a comparison, however, assumes to know certain things:
1.) For the convergence scenario, it assumes to know (in some fashion) the natural forces that might shape and mould each protein and make them similar to each other (and roughly the probablity of said forces doing this), and/or the statistical chances that the convergence simply happens by accident. In other words, to make this judgement between homology and convergence, we assume to know enough about the natural forces that might cause convergence, so that we can rule it out as less likely.
2.) For homology, we assume to know the probability boundaries that separate these two protein types from each other (evolutionarily speaking). These include the barriers of function, regulation, folding conservations, etc... that might prevent these two systems from evolving from a common ancestor. We assume to know enough to say that the pathway of divergence is more likely than the pathway of convergence (when we assign homology).
If we are to honestly rule out the possibility of convergence and call something homologous, we must mentally tally up the probabilities associated with convergence and divergence of two systems, and compare them. We then make the judgement that one is more likely, and so call it the tentative explanation.
Now, it should be quite evident that in almost all cases of homology assignment (at least between proteins with different structures and functions) we basically know almost nothing about *either* scenario. For instance, we:
- Have not ruled out all possible competing scenarios, such as convergence
- don't know all possible combinations of chance+law that might favor convergence
- don't fully understand each protein in question
- often don't have the slightest clue what probability barriers might prevent ever having a common ancestor, and so do not even know the probability of our preferred explanation!
So, in summary, what folks in the literature have done time and again is rule out one evolutionary scenario for another - namely, the ability of RM&NS to independently produce proteins that are *specified* by the fact that they are similar. All of this, and they don't know hardly any of the probabilities involved in each scenario.
Now, this is not simply an attack on Darwinism. In speaking to ID sympathizers, I propose that we can actually learn something by drawing a parallel between those who assign homology (in a rigorous way, rather than simple sequence comparison) and those who infer ID.
For the ID theorist, the following scenarios are compared:
1.) That system X was produced via unguided forces of chance+law, and is consistent with the pattern that chance+law produces.
2.) That system X was produced via non-natural forces beyond the reach of chance+law, and so is consistent with a pattern that is not within the domain of chance+law.
Now, Dembski, Behe, et al have been roundly criticized over the past few years for drawing any sort of ID inference because they did not:
1.) Eliminate all possible natural scenarios for formation.
2.) Allow that some new law or explanation may arise
3.) Understand the inner workings of their model systems better (with the idea that the ID inference rests here on simply not understanding the system well enough)
See the parallels with the previous example (assignment of homology)?
Again, I propose that the ID inference shares many key similarities with the assigment of homology. In each case, the knowledge is limited. And in each case, there are good examples (flagellum for ID, and proteins that we have watched diverge in the lab for evolution), and less clear examples (much smaller IC systems for ID, and distantly similar protein systems for evolution). So in both cases, the exact boundaries of the inference are hard to come by. And just as the ID inference rules out competing scenarios without necessarily knowing the probability that an ID event would occur, so the homology inference often rules out competing scenarios (like convergence or even independent design) without really knowing the probability that the two systems in question could even really be divergent (i.e. without knowing the probability of its explanation).
So perhaps this comparison will be a call for two things:
1.) A fairer, more even handed amount of skepticism of the ID inference when it doesn't know all and test all
2.) A greater degree of questioning of the homology inference (and elimination of the automatic homology assignment!) with the hope that this will lead evolutionary theorists to join IDsts in delineating the theoretical limits to the power of RM&NS to produce novel systems and to evolve one system into another. It is time that the evolutionary theorists joined up in this effort!
In fact, I further propose that a clearer understanding of the probabilities involved in the evolution of systems will benefit *both* those who wish to determine real homology, and those who wish to determine intelligent design. And so the work of those such as Behe and Dembski is really a boon to both sides.
And finally, this ends hopefully opening a new avenue of ID research. Suppose that we learn to delineate the theoretical chance of formation of some biological systems, and find that *both* convergence and divergence are quite unlikely for two systems that nontheless share lots of similarity. What does this mean? The ID theorist could then probe into whether we might be looking at an example of common *design*, where the two systems share bits of information due to the fact that they were created from common designer(s). That may well explain the sharing of information in spite of the high probability barriers associated with it. The link I referred to above addresses this as well. We already know that lots of systems out there share key, small inserts of common information such the zinc-finger domains, beta barrels, etc... and it may be that we are sitting on a mound of commonly designed objects, but that we simply can't see it yet, since we really have no clue in most cases whether two systems are convergent, divergent, or commonly designed.
Thanks,
jay
[ 28 June 2002, 01:00: Message edited by: Jay ]
IP: Logged
|
|
Paul A. Nelson
Member
Member # 26
|
posted 28. June 2002 11:09
Jay, I think you're onto something here, although I expect that many readers will find your points (or arguments) so counterintuitive that they may dismiss what is a very promising set of questions. I am terribly pressed for time right now, but plan to have more to say in this thread within the next few weeks. I'll also try to suggest some useful literature dealing with the questions you raise here.
The most widely-held assumptions are often those most in need of being hoisted into the air for a total inspection.
[ 28 June 2002, 12:16: Message edited by: Paul A. Nelson ]
IP: Logged
|
|
Jay
Member
Member # 268
|
posted 28. June 2002 11:34
Thanks Paul, I would really appreciate it. I also am pressed for time, but have a few more topics related to this that I plan to bring up.
It almost startled me a few weeks ago when I was reading some stuff just how little we really knew about the limits of evolution, and how much we really understand things like homology. It was so far less than I expected that I also almost dismissed it.
** A practical application for understanding design vs. homology:
I have some ties to some guys who study folding patterns by studying mutational covariance. They infer covariance using statistics gleaned from 'homologous' sets of proteins, i.e. they compare what seems to be mutation rates among homologous sets and see if there is covariance. But here's the problem - we don't know if these data sets really are sets of homologous proteins or not! They just assume that at face value by sequence similarity, as this is all that we are currently capable of doing. And the problem is that we often get very confusing results when doing these models. This has always been a problem.
I have this hunch that we are going to find out one day that the reason that lots of this homology modeling stuff gives some confusing results is that we were comparing sets of proteins that aren't really homologous at all, but just shared some high similarities as well as new *designed* features. In other words, they are the products of intelligent common design! So trying to read and interpret this data as though it is a set of homologous, divergent proteins that mutated away from each other is naturally going to give bad results, if that is not what happened. However, to make it more confusing, some good results may still come out as *some* of the sequences in the system may truly be homologous! So the truth may stay buried for awhile!
In truth, we don't know what proteins really are homologous for the most part, and so we don't know right now what sequences to include in statistical calculations. Finding the limits to evolution between types (doing work similar to Dembski's) would be *very*, very practical for lots of these guys. I cannot stress that enough. Understanding evolutionary limits and detecting possible common design/convergence vs. homology would have real, practical applications in this field. Much more than I could list right now. It would, for instance, lead to a much, much greater understanding of protein folding and more intelligent drug effect predictions, as our homology modeling could be based on a real understanding of homology vs. common design. This is not just a philisophical exercise! The field of bioinformatics would practically benefit very much from this!!! We need to pursue the detection of design if for no other reason than to benefit this field.
Detecting design is practical indeed! It may even do such practical things as accelerate our discovery of new drugs!
Thanks,
jay
[ 28 June 2002, 11:49: Message edited by: Jay ]
IP: Logged
|
|
Frances
Member
Member # 169
|
posted 30. June 2002 20:20
Jay,
very interesting posting. However I would like to point out to you that I do not believe that there is a similarity between ID's (Behe, and Dembski) approach and homology inferences.
First of all ID proposes a reliable filter for ID, that is, no false positives. Dembski is clear about this, he argues that his ID inference does not suffer from false positives. This is only possible if one can show that one has indeed looked at all possible scenarios and rejected them. So ID is eliminative in nature, it does not propose a mechanism of ID merely a method to (reliably) infer ID.
So unless ID drops the reliability argument, one has to deal with the fact that our knowledge may be limited. Also, unless ID can propose a positive way to identify ID, it will always have to rely on eliminative arguments with all the problems of such. Your two scenarios of ID versus evolution show the eliminative nature. ID is not a positive argument but a negative one, "non-natural forces beyond the reach of chance and law".
Now protein evolution. Does protein evolution rely on a positive or negative argument? Does protein evolution provides for a (hypothetical) mechanism?
But despite this, you do make a valid point that protein homology itself needs additional evidence. As far as I can tell, protein evolution hypotheses are based on positive predictions to explain the observed data.
I will give some references that seem relevant. The following paper
quote:
PROTEIN EVOLUTION AND PROTEIN FOLDING: NON-FUNCTIONALCONSERVED RESIDUES AND THEIR PROBABLE ROLE by O.B. PTITSYN
discusses protein
sequence divergence and a mechanism to explain.
Another paper describes protein sequence comparison and protein evolution
As I understand it protein evolution can provide us with phylogenetic data to compare with other phylogeny methods. As I understand homology modeling it is based on the premise of a common ancestor. Over time they acquire mutations but their function and structure remain preserved. A better understanding and predicting ability of protein folding will surely help us resolve many issues.
Understanding protein evolution may certainly help us reject certain evolutionary pathways and thus could be leading to an ID inference. However ID is eliminative in nature which seems to be quite different from protein research as far as I can determine.
[ 30 June 2002, 20:58: Message edited by: Frances ]
IP: Logged
|
|
Jay
Member
Member # 268
|
posted 01. July 2002 18:28
Hi Frances,
Thankyou for your reply! I appreciate the honest criticism.
Frances: "First of all ID proposes a reliable filter for ID, that is, no false positives. Dembski is clear about this, he argues that his ID inference does not suffer from false positives. This is only possible if one can show that one has indeed looked at all possible scenarios and rejected them."
Well, I'll let Dembski answer things about him for himself if he wants to. From where I stand, however, it seems possible that the ID inference could make some false positives for the reasons I outlined. But, the chances of doing this can go down as our understanding of the complexity and ICness of the system go up. The real question here, however, is at what level the critic will accept the presumed positive. When I hear things like 'ID has no evidence' I presume that they find *all* of these presumed positive identifications of design a la Dembski to be completely absurd and not convincing in the least. Not even one percent so, or else we would have to admit that the ID inference does have at least some shred of scientific interpretation going for it. And to be quite frank, I find that rather hard to believe that the inference is that bad!
What makes it even harder for me to believe that critics really find the ID inference so untenable for scientific reasons, is what I presented in the first of this thread - if you are really that critical of the ID inference, and really do find all of these positives so absurd and completely crazy (and not even remotely scientific!), then how are we justified in doing something else that is done routinely in science, namely, calling two different systems homologous?
I wanted to highlight the way that we handle these two sets of data and their interpretations and show that on one hand, some data is treated with the utmost skepticism and dismissal (ID inference), while on the other hand, interpretation of other data that is often less rigorously interpreted (homology) is casually treated as fact, even in technical literature. Clearly, there is a problem of uneven weights and measures here.
Frances: "So ID is eliminative in nature, it does not propose a mechanism of ID merely a method to (reliably) infer ID."
Well, I don't really agree, and do plan to post some stuff on this soon. The ID inference begins with our ability to spot design in things that we create as designers, and then extrapolates those principles to systems in nature that look comparable in some ways. This is just to begin looking. It then seeks to confirm the suspicion by finding data patterns that are consistent with premeditated assembly, and seeks to rule out interpretations of the data that suggest unguided assembly. This has close parallels to much of evolutionary interpretations of homology.
Here's the way that I see most homology asessments: The author sees sequence similarity, and from what he knows of having observed limited instances of similar sequences come from similar sequences (say, in reproduction), *extrapolates* this to a much more diverse set of data to assume homology, while presumably also ruling out competing scenarios. This process is most apparent in papers that deal with distant homology. Much of what he is relying on here is the extrapolation that since similar sequences always come from similar sequences in the limited span of time that we can observe, that this must be a good way of interpreting *any* sequences that share similarity... Both design and homology inferences make heavy use of extrapolation.
It should be stated again that we know hardly anything about our homology scenario, even though we assume it. We do not know what the actual barriers are that would prevent evolution between our 'homologous' proteins. We honestly have no idea if the chances of crossing from one type to another is 1 in 2 or one in 10^-100000. Our extrapolation, then, is rather weak since it *assumes* to know that the very small amount of divergence that we actually observe is a good indicator of the *probability* that any similar sequences must be related. But we can then ask, why do we presume to know that this extrapolation is good?
See the problem? We are taking a very, very small data set and subconsciously using it as an analogy to calculate the probability of divergence of all natural systems (as a way of eliminating competing ideas like convergence), even though we don't know if the extrapolation is at all even close to valid.
So we often interpret homology with an unverified extrapolation, eliminating competing ideas even though we don't know the probabilities of our favored scenario, or of the competing ones!!
Now, I'll interpret from my own standpoint and point out that many of these natural machines that we see are very analogous to machines that we make with intelligence, in fact, many work much better then ours and would look *more* intelligent in a side by side comparison - say, if they were displayed side by side at a tech show. I will then use this rather strong analogy (at least in my opinion) to *suspect* ID in the first place. My suspicion will be further strengthened by finding that, in fact, such a system does indeed defy natural processes in its formation. So by analogy (to start suspecting it), finding that the system lines up with more formal ideas of what designed data patterns are, and subsequent eliminative data to follow up, I find it a *more* reasonable way of looking at things. The parallels between this and homology asessment seem rather clear to me, except that ID seems to be making a more concerted effort to question its most fundamental assumptions. Homology asessment seems to simply assume them.
And once again, I find it rather odd that those who label things homologous would be so unbelievably skeptical of the ID inference, given that their own inferences are often in such a sorry state (yet they still run with them).
...By the way, thankyou for those links. They are quite relevant to myself and some of the guys that I work with. They will greatly appreciate it, if they don't already have it!
Frances: "As I understand it protein evolution can provide us with phylogenetic data to compare with other phylogeny methods. As I understand homology modeling it is based on the premise of a common ancestor. Over time they acquire mutations but their function and structure remain preserved. A better understanding and predicting ability of protein folding will surely help us resolve many issues."
Now here is the really interesting stuff, where I think we'll begin to see more of the practical usefulness of the design inference. We are moving more and more into reliance on homology modeling to understand how proteins really work, and predict things like active sites or folding contacts. But often, the results rom these homology models are confusing, and misleading. The likely reason in most cases is that we are drawing the wrong data for our modeling. In other words, we may be selecting sequences for our alignments that are not, in fact, due to common descent and subsequent divergence. Should this be the case, then our homology model will be wrong, as it works off the assumption that everything in the data set is divergent and related by RM&NS. But suppose that these 'divergent' sequences were really the products of separate design, or convergence?
However, for some data sets, homology modeling does work. For instance, we can often infer the function of a gene if we find something that really looks like it. That was one of the main points in the intro of that last paper you cited. But, while this is very useful, simple gross comparison and guessing of function is a far, far less sensitive process than many of the newer homology modeling ideas. While this straightforward comparison of two very close genes to infer function relies on getting 'close enough' to make a good guess as to function, some of these other homology modelings rely on the premise that the *difference* between divergent proteins in a data set is due to RM&NS. These homology models need the difference to calculate things, and so they tend to take as wide of variation as they can and still work. So it's no longer simply trying to find really, really close looking stuff, but rather probing into the how's and why's of the differences between varied data sets. This is where we start to have problems with defining homlogy. How do you know where to cut off your sequence alignment (i.e. what sequences in the set are really homologous?)
Now here's where the design inference becomes very useful in a real, practical sense. A refined, more powerful design inference might be able to detect which protein sets are likely to be truly independent from each other (meaning that they were not likely able to evolve from a common ancestor), vs those that could be produced from an ancestor by natural processes.
Homology modeling, ironically, needs something that can detect the limits of the ability of natural processes, to accurately define which proteins in a set are indeed related by natural processes, thereby cleaning up our data set! And cleaning up a data set would lead to more accurate statements about protein structure and function, and could really mean the discovery of new active sites or folding principles!
So, the design inference may turn out to be very, very practical and may end up being used in quite a few labs! Unfortunately, it is still in its infancy, but as it refines itself, it may well find itself, for instance, cleaning up homology modeling databases... It seems that in the end, it is perhaps even more useful to talk about *connecting* homology comparisons and design detection, rather than just contrasting them, as it shows the potential practical usefulness of the design inference.
Thanks,
jay
[ 01 July 2002, 18:30: Message edited by: Jay ]
IP: Logged
|
|
Frances
Member
Member # 169
|
posted 01. July 2002 21:36
Hi Jay,
A quick response to a very interesting posting of yours. I do not want to trivialize your reply so I will revisit it in more depth later but I would like to address some of your claims.
You seem to disagree with my claim that ID is eliminative. You respond by saying that this is incorrect since ID relies on 'our ability to spot design in things we create as designers and extrapolate'. But that's Paley's argument and the argument has been shown to suffer from some real problems as any extrapolation based on similarity has. After all when Darwin (arguably) showed a naturalistic pathway to explain the evolution of life, the argument of design which was based on a similarity argument not positive evidence, failed. So unless one can eliminate any causes that would generate 'apparant design' ID remains powerless. This is why Dembski et al proposed a method which would 'reliably' infer design. Hence the design inference or in Behe's case ICness.
Without any positive evidence to support ID there is little hope for such approaches to be productive in science in my opinion.
You further continue to explain how ID confirms its hypothesis through elimination but that's not how homology works or for that matter most science. Science proposes a hypothesis which it tries to falsify it does not propose a hypothesis by attempting to falsify all 'other' hypotheses.
An interesting quote from you which I would like to repeat verbatim
quote:
Now here's where the design inference becomes very useful in a real, practical sense. A refined, more powerful design inference might be able to detect which protein sets are likely to be truly independent from each other (meaning that they were not likely able to evolve from a common ancestor), vs those that could be produced from an ancestor by natural processes.
I fail to see that this is the case. Actually science is really already doing work for ID, namely by showing that some apparantly homologous sequences are not truely so. But what would that show? Intelligent design? Why? Without understanding the intelligent designer or the mechanisms of ID, how can we really make any predictions?
Homology may make certain assumptions, so does radiometric dating for instance. Both methods have internal or external ways to verify the validity of the findings. To suggest that ID makes a more fundamental attemtp to question is stating the obvious, ID can only be shown imho by questioning natural pathways and chance. The questions are 1) is such an approach useful 2) does regular science not do this?
So far I fail to see why ID is better than the present scientific approach or how ID can provide us with additional scientific knowledge. Unless ID can provide us with mechanisms for design. Without that ID remains unconstrained.
[ 01 July 2002, 21:42: Message edited by: Frances ]
IP: Logged
|
|
Mike Gene
Member
Member # 149
|
posted 01. July 2002 23:35
Frances: But that's Paley's argument and the argument has been shown to suffer from some real problems as any extrapolation based on similarity has. After all when Darwin (arguably) showed a naturalistic pathway to explain the evolution of life, the argument of design which was based on a similarity argument not positive evidence, failed.
Paley's argument fails as a deductive proof for the existence of God. Nevertheless, analogical reasoning is a valid way of investigating the world. A good analogy does constitute positive evidence in such a context. We just need to remember that positive evidence is not the same thing as proof, nor does it mean everyone is rationally obliged to embrace the positive evidence and follow its lead (i.e., the positive evidence might not be sufficient to overcome the ambiguity of a situation, allowing for alternative explanations). As for Darwin, all life is built around machinery and programming. Darwin fails to explain this, as the evolution of life is built upon the matrix of this machinery and programming.
As far as "mechanisms of design" go, we need to remember two things. First, the Darwinian mechanism, more often than not, usually boils down to an appeal to a just so story, complete with appeals to lucky coincidence (the heart of the cooption stories). Secondly, asking for the "mechanism of design" is asking a teleological explanation to express itself as a non-teleological explanation. For the simple sake of argument, imagine the first cells deposited on this planet were bioengineered. How would one really determine the mechanism of design, when such a determination would depend on identifying blueprints, protocols, recipes, etc.? We can attempt to reverse engineer a system to uncover its hidden logic (a fingerprint of some mind), but does anyone really expect to find lab notes. complete with the step-by-step procedure behind the implementation of the design?
IP: Logged
|
|
Frances
Member
Member # 169
|
posted 02. July 2002 01:29
Frances: But that's Paley's argument and the argument has been shown to suffer from some real problems as any extrapolation based on similarity has. After all when Darwin (arguably) showed a naturalistic pathway to explain the evolution of life, the argument of design which was based on a similarity argument not positive evidence, failed.
Mike: quote:
Paley's argument fails as a deductive proof for the existence of God. Nevertheless, analogical reasoning is a valid way of investigating the world. A good analogy does constitute positive evidence in such a context.
I agree that analogical reasoning can be a good start to explore the world but I also argue that analogical reasoning fails as 'positive evidence'.
I ran across a good article on this
The Skeptics dictionary states it well
quote:
One quality of a good analogical argument is that the characteristics cited as shared characteristics must be truly shared characteristics. If there is doubt that one of the items being compared (the universe) possesses the most significant shared characteristic (of being orderly and purposive), then the analogical argument is not a sound one.
Mike continues quote:
As for Darwin, all life is built around machinery and programming. Darwin fails to explain this, as the evolution of life is built upon the matrix of this machinery and programming.
So Darwin does not explain the origins of life (the machinery and programming) but that does not mean that his theory does not attempt to provide a positive theory to explain the observations.
I notice that Mike repeats the assertions that Darwinian stories "more often than not" are an appeal to a just so story. Even assuming that Mike is correct and I disagree with Mike, Darwinian inquiry does not necessarily end with just a 'story'. First of all a mechanism is proposed open to falsification, predictions are made that are positive in nature.
If intelligent design cannot propose a positive story, a mechanism, an identifiable designer then its stories will remain nothing more than "it appears to me that natural forces and chance could not achieve this" and the problems with analogous reasoning will remain. So unless there is unambiguous evidence of a designer, such as blueprints or recipes or evidence of the creator, design will be able to do little more than attempt to eliminate known (and if it wants to claim that ID is free from false positives, unknown) mechanisms. Intelligent design in the manner used by Mike here is already applied in archeology where we have independent positive evidence to help us in our decisions. But I do not believe that the designer in ID will be that visible to us.
Dembski promised me that once I returned from my one month absence from ISCID I could raise the filter proposed by Wilkins and Elsberry in "The advantages of theft over toil: the design ingerence and arguing from ignorance" Biology and Philosophy 16. November 2001, 711-724.
Here is a preview:
They argue that quote:
"One cannot make a substantive case for rarefied design except through analogy, and Hume showed this fails since no ordinary designer works to produce the sorts of artifacts that livings things would be if they were artifacts. So if analogy is not used, no explanatory power inheres in rarefied design inferences. Of course this does not preclude ordinary design inferences of the origin of life, for example (cf. Pennock 1998 on the Raelians, a group who believe life was seeded by aliens), but the evidence that would make this a viable explanation is lacking, especially when compared to regularity explanations in terms of known chemistry, physics and geology."
See also Wesley's powerpoint presentation
[ 02 July 2002, 01:31: Message edited by: Frances ]
IP: Logged
|
|
Jay
Member
Member # 268
|
posted 02. July 2002 02:10
Hi Frances,
Frances: "A quick response to a very interesting posting of yours. I do not want to trivialize your reply so I will revisit it in more depth later but I would like to address some of your claims."
Thankyou. And I, in turn, appreciate your comments and hope not to trivialize them either!! Btw, I wrote this just as you replied again to Mike, so if I don't address something in your reply, I will later.
I'll chime in and agree with Mike's reply. Notice also that I wrote that analogy is our way of *suspecting* design in the first place. It is not a proof. I pointed this out to compare with evolutionary theory's similar tactic of suspecting by analogy, namely, looking at a very small, limited set of naturalistic behavior and arguing by extrapolation that any sequence similarity (or any other 'homolgous' pattern) means that the systems were once physically the same.
Frances: "After all when Darwin (arguably) showed a naturalistic pathway to explain the evolution of life,"
This is sure to be a rabbit trail, but I want to point out something here. Let's compare three hypotheses:
1.) That all of life is ultimately the result of unguided chance+law, and that all organisms are physically related, and have come about through a process of divergence also powered by chance+law.
2.) That all of life is ultimately from chance+law, but that organisms have arisen more than once, and that many features we see are the result of 'convergent pressures' that select for certain traits, structures, sequences, etc... So life is both divergent and convergent.
3.) That life had its beginnings in purposeful engineering, and that certain original 'types' were the result of purposeful construction. After that, chance+law was able to run these machines and do with them as it wanted.
This is hardly comprehensive, but I believe it highlights something important. We often speak of Darwin giving a clean, scientific mechanism, with ID giving only a vague speculation. But if we look at these critically, we see that explanation #1 is indeed quite general. It is every bit as general as #2 or #3. It also leaves lots of unknowns, as does the ID explanation. For example:
- life *should* show nested hierarchies because that's the pattern from our hypothesis... however, when those patterns prove to be nonexistent for certain classes (as in the three domains and many phyla) then we say that it still fits our hypothesis, but now due to LGT. So if life shows nested hierarchies, good. If not, fine.
- life's biological machines are from chance+law. Many times, we can chart those pathways. But if we cannot - even if there are whole vast areas where the data is abysmally unclear and confusing (or even makes it look unlikely) - we can simply say that it must be some *form* of chance+law, perhaps one we don't know about.
In the above, both pattern and process are quite general and vague. The pattern is... well, I don't really know what pattern there must be for #1 (i.e. one, if broken, that would falsify the hypothesis). The process is "chance+law," but this is also very general. We know this from the attitude towards IC systems. No specifics are really proposed most of the time, instead, there is simply an appeal to 'natural processes' in general, even to unknown ones (like self-organizing princinples). Chance+law is not well defined at all. And yet interpretation #1 still holds in the minds of many, in spite of its being very vague. My point is, Darwin's *overall* theory (not talking about the small, observable evolution) is every bit as vague as saying that life has intelligent beginnings. Sometimes it just takes a double-take to see that. To prove this point (that it is vague) from a purely *naturalistic* standpoint, try showing it to be clearer than explanation #2. That was the original point of this thread.
So, in summary, Darwin didn't really show a naturalistic pathway, per se, he merely provided a very vague, general overview of one way to interpret the world. ID does the same thing.
Frances: "So unless one can eliminate any causes that would generate 'apparant design' ID remains powerless."
Ok, back to square one! Unless you can eliminate any causes that might cause apparent homology, the inference of homology remains powerless!
Frances: "This is why Dembski et al proposed a method which would 'reliably' infer design. Hence the design inference or in Behe's case ICness."
Yes, and I aplpaud them for that. Analogy alone is not enough to strongly infer design, at least for me. Now, I wish they'd do this for homology and define its limits!! I have some data sets that need a good cleaning!
Frances: "Science proposes a hypothesis which it tries to falsify it does not propose a hypothesis by attempting to falsify all 'other' hypotheses."
But I never suggested it work that way! When I go about suspecting design, I start by finding data that fits my initial suspicions of design, i.e. systematic breaking of nested hierarchies, irreducibly complex systems, the sudden appearance of large sets of conserved information, etc... I then do a follow up by, in a sense, trying to falsify this by looking for likely natural causes. That's what I see the ID theorists doing, but ID critics seem to interpret this as though we simply go pick up any random object and try to disprove all possible natural interpretations. But it really doesn't work that way, at least for me.
Frances: "Actually science is really already doing work for ID, namely by showing that some apparantly homologous sequences are not truely so. But what would that show? Intelligent design? Why? Without understanding the intelligent designer or the mechanisms of ID, how can we really make any predictions?"
Well first, let's first be a little skeptical back, just for fun. Most homology asessments *do not* understand the mechanisms and pathways that would be needed to really transform one protein type into another - this is especially true for the more 'divergent' sequences. So may I assume that your skepticism carries over with me to homology on this one? Are you prepared to argue against these homology asessments on the same grounds that you criticize ID?
As for this homologous/convergence detection, all I've seen is more of the same face value interpretations that I've been sharing here. For instance, they might take two proteins that are completely different, but maybe share 3 aa's (see my examples and link in the first post). This will likely be interpreted as being convergent, but the underlying reason is not that they have ruled out divergence, but that 3 aa's aren't statistically all that hard to converge on. There is no work (to my knowledge) that actually critically attacks evolutionary theory in relation to comparing similar proteins to see if divergence is less likely than convergence. None. As I said several times, it's almost always this superficial approach where they run the quick mental calculation that 3 amino acids and a pinch of structural similarity must not be all that hard to converge on.
But show me a paper where they take two proteins with a good deal of sequence similarity, and show why divergence is less likely than convergence (with an emphasis on comparing the likelihood of the two pathways). It would indeed be a boon for ID theorists!
I believe that the underlying reason that such work is avoided is that it's even worse, from a naturalistic standpoint, trying to explain why these sequences could be so similar yet unrelated. It simply makes sense, from a naturalistic viewpoint, that all of this stark similarity must somehow be related. My colleagues would simply laugh at the idea of even considering convergence for high sequence similarity as they intuitively 'know' that such a thing is very, very unlikely (again, from a naturalistic standpoint). Look at how they treat Christian Schwabe.
But, here's where a design perspective may provide a fresh *impetus* to pry under these rocks. Design would have no problem with the idea of similar, independent constructions. So it would have no problem considering - even suspecting - that two systems that shared a lot of sequence similarity could still be unrelated. Here's the key - it's not that evolutionary theorists *can't* look for constraints on homology and compare it to convergence, it's that, from their standpoint, it doesn't seem to be the right question to even ask! That is a key thing to keep in mind! I'm not saying that evolutionary theorists cannot do this work, or even interpret the results from a naturalistic stanpoint (I argued the hypothetical convergence (not design!) vs. homology debate, after all!). It's that common design would expect this pattern, and purposely go looking for it. Naturalists, on the other hand, don't consider convergence all that good for sequences (neither do I, quite frankly), and so have no other choice but to assume that somehow, sequence similarity is homology. It's all a matter of perspective. And in this case, I think that design provides a very needed new perspective.
Frances: "Homology may make certain assumptions, so does radiometric dating for instance. Both methods have internal or external ways to verify the validity of the findings."
Yes, there can be external checks, like observing it, or finding that the pattern of supposed homology that aligns with other patterns that should be similar i.e. if a protein phylogeny aligns with an organismal one. Even that method can still be questioned, but nonthless, it is still a fairly good reason to *suspect* homology in my mind. But the problem is that most 'distant' cases of homology don't have that indepedent verification. And it's not due to lack of trying. We're finding that the data, in many cases, is simply not going to confirm homology ever, as it sometimes appears to even contradict it (i.e. protein phylogenies go against organismal ones, or paralogy rooting gives crazy values).
Frances: "So far I fail to see why ID is better than the present scientific approach or how ID can provide us with additional scientific knowledge."
Perhaps we're just shouting past each other here, but as I pointed out several times, work like Dembski's would be of great practical value to me and to many others who study the inner workings of proteins and do sequence alignments. Design is quite useful, IMHO. It just needs to be more refined (at least for help in homology interpretation) and sensitive to things like single genes rather than just large IC systems. Give it time. But even unrefined, it is quite fascinating, and opens new way of looking at things, and a new *impetus* to ask new questions like whether proteins with high sequence similarity are due to divergence, convergence, or common design. That, for me, is enough to keep me interested. I expect more great things from ID theory in the future.
Thanks,
jay
IP: Logged
|
|
Mike Gene
Member
Member # 149
|
posted 02. July 2002 12:21
My apologies to Jay, as my sidequest seems to focus on more general issues than the specific issues/hypotheses he raises. Jay, please e-mail me if you want me to continue or pull back.
Frances writes: I agree that analogical reasoning can be a good start to explore the world but I also argue that analogical reasoning fails as 'positive evidence'.
Analogical reasoning is a good place to start because it constitutes positive evidence that points in a particular direction. Why else would one think it a good place to start?
The links that Frances provided are not relevant, as they are focused on the "Argument From Design," one is looking for a proof that delivers certainty. But I'm thinking of analogy in a scientific sense, not a philosophical sense.
There are many instances in the scientific literature where analogy is used as positive evidence in favor of a hypothesis. For quick-n-dirty examples, consider:
quote:
Estimation of the time of origin of the major HCV genotypes (types 1-6) is problematic, but our data and analogy with other viruses suggest that divergence occurred at least 500-2000 years ago. -
J Gen Virol 1997 Feb;78 ( Pt 2):321-8
The origin of hepatitis C virus genotypes.
Smith DB, Pathirana S, Davidson F, Lawlor E, Power J, Yap PL, Simmonds P.
In analogy with sucrose and its higher homologs, raffinose and stachyose, which may act as protective agents during maturation drying in plants, these trehalose homologs may also have a protective role in mycobacteria, perhaps during latency. -
Eur J Biochem 2002 Jul;269(13):3142-3149
Trehalose-based oligosaccharides isolated from the cytoplasm of Mycobacterium smegmatis.
Ohta M, Pan YT, Laine RA, D Elbein A.
Of course, in science, we don't stop with the analogy. We attempt to test it and flesh it out. And it seems that this is what Jay is seeking to do. But the main point is that analogy can indeed exist as positive evidence. We just need to stop think of it as some deductive argument attempting to deliver certain proof in order to see this.
As for my point about just-so stories, I just think it important to recognize the essence of the "mechanism" delivered by Darwinian thinking. Consider that we both agree that birds evolved from some reptilian ancestor. But what was the mechanism? Dawkins proposes the mainstream accounts in his book, "Climbing Mount Improbable." I invite everyone to read that chapter to determine for themselves if just-so stories are being raised. Of course Darwinian inquiry doesn't necessarily doesn't end here, but where has it gone further?
Now, it is also important to consider that non-telic explanations do have a methodological advantage in that the 'causes' are regularities that can be manipulated and studied in the here-and-now. Design, in contrast, entails the intervention of some mind that cannot be directly studied, unless we can currently interact with the designers. Yet the methodological advantage of the former doesn't translate as it being more valid. It's just easier to work with. It's just easier to arrive at consensus . Yet even with these advantages, the mechanisms proposed in Darwinian inquiry are either quite vague or just so stories. Why expect more from ID when it is methodologically hindered, in comparison? Is this really an objective way of going about this problems?
Frances: f intelligent design cannot propose a positive story, a mechanism, an identifiable designer then its stories will remain nothing more than "it appears to me that natural forces and chance could not achieve this" and the problems with analogous reasoning will remain.
Thus far in my investigation (and ridiculously brief):
1. A positive story - evolution was front-loaded through the seeding of a heterogeneous population of sophisticated cells.
2. The mechanism - advanced bioengineering is behind the origin of the cells.
3. The designer - some form of being with human-like intelligence.
Want more details? In time. But for now, simply recall the "level of detail" that comes with accepted Darwinian proposals for the origin of some biotic feature. It doesn't really offer much more detail that my investigative explanation.
So unless there is unambiguous evidence of a designer, such as blueprints or recipes or evidence of the creator, design will be able to do little more than attempt to eliminate known (and if it wants to claim that ID is free from false positives, unknown) mechanisms. Intelligent design in the manner used by Mike here is already applied in archeology where we have independent positive evidence to help us in our decisions. But I do not believe that the designer in ID will be that visible to us.
Here is where we part company. Frances seems to be saying that unless we come up with " unambiguous evidence of a designer," we should throw our hands up and abandon all effort to use ID as an investigative guide. But what if life was designed? Does this entail that we should be able to find "unambiguous evidence of a designer." Nope. In fact, I cannot think of one good reason why one would argue, "Since life was designed, it follows we should be able to get our hands on the blueprints or recipes." None. Thus, Frances is elevating epistemology above ontology. That is, even if ID is behind the origin of life, we are to invent other accounts for lack of evidence that doesn't even follow from the design of life. This approach simply makes no sense to me.
A better approach is to think about ways to circumvent the problem of having no unambiguous evidence of a designer. Such approaches may be inherently risky and even fuzzy, but at least it is an attempt to work with Nature rather than shape Nature in accord with our limitations. Of course, this doesn't mean Frances is obligated to follow along. If he/she needs independent evidence of a designer to take design seriously, we can hardly expect him/her to think about ID seriously.
The point is that Frances has no basis to insist that we first find the designer in order to explore Nature from an ID perspective. My interest has never been to get non-teleologists to agree with me. My interest has been to satisfy my own genuine curiosity about these issues. If my reasoning and inferences resonate with others, great, as a lone-wolf investigation is hindered by the shear amount of data that stands before it. If it doesn't resonate, oh well, as this doesn't eliminate my genuine questions. But to get me to stop, someone is going to have to do much better than argue we "need the designer." Intellectual challenges call for innovation, not abandonment.
Finally, Frances quotes Wilkins and Elsberry:
quote:
"One cannot make a substantive case for rarefied design except through analogy, and Hume showed this fails since no ordinary designer works to produce the sorts of artifacts that livings things would be if they were artifacts.
Hume is irrelevant, as his focus was on using analogy to a) prove b) the existence of God. More importantly, the fact that life is built around codes and machinery illustrates this assertion is flawed. Furthermore, any difference between artifacts and living things need not reflect design vs. nondesign. On the contrary, the difference may reflect inferior design vs. superior design.
quote:
So if analogy is not used, no explanatory power inheres in rarefied design inferences. Of course this does not preclude ordinary design inferences of the origin of life, for example (cf. Pennock 1998 on the Raelians, a group who believe life was seeded by aliens), but the evidence that would make this a viable explanation is lacking, especially when compared to regularity explanations in terms of known chemistry, physics and geology."
When it comes to the origin of machine and codes (and all evidence indicates life is machine-dependent and code-dependent), I think a teleological cause is superior to regularity explanations in terms of known chemistry, physics and geology. W&E's assertion would resonate if we lived in World B:
[Some Hotel Lobby...]
Bob: Hey Brian, where were you?
Brian: Sorry I'm late, but I found myself in some conference room where many scientists were clearly excited about their work.
Bob: What were they talking about?
Brian: I couldn't tell, with all their technical talk and all. But they were talking about some 'code' and how this 'sophisticated machinery' was needed to tap into it. Something else about some checkpoint circuitry that implemented decisions using feedback loops.
Now, which of the following responses from Bob seems to more naturally follow?
World A. Bob: Hmmm, sounds like they're studying some new technology.
World B. Bob: Hmmm, sounds like they're studying the effects of geology.
In the past, I have asked many critics of ID to outline just what type of data would cause them to suspect a teleological cause for something like the flagellum. Needless to say, they have had tremendous difficulty answering such a question. The reason is that we've developed our minds and approach to be dependent on independent knowledge of the teleological cause, which in practical terms, means that we are only able to determine teleological causes if the cause was a human being. But imagine you did uncover something about the biotic world that caused you to suspect a teleological cause. How would you go about testing this suspicion? It's one thing to pooh-pooh Dembski and Behe, but in doing so, one is merely arguing that ID has not been proven or a non-teleological cause has not be disproven. What such skepticism lacks is a demonstration that any ID inference is indeed mistaken. In science, there are no criteria used to distinguish between teleological and non-teleological causes, such that we can say that X has been tested and a non-teleological cause, to the exclusion of a teleological cause, has been demonstrated. Instead, there are only a myriad of arguments that say we should continue to ignore ID.
What all this means is that anyone nagged by a growing suspicion of ID is going to have to "start all over." Yet the problems plaguing anyone in this position are many and large. For example, there is all the baggage that comes with ID. For a very long time now, teleological explanations have been tied to religious apologetics. This situation is only made worse with many in the ID community who focus on the political agenda. And embedded in all this is the use of teleological thinking as a reaction to the reigning non-teleological paradigm. Things are made worse yet in that the science of biology has coopted many teleological concepts, thus driving teleological thinking into the realm of the purely reactionary. In politics, President Clinton successfully coopted many Republican issues, making it harder and harder for Republicans to find a distinctive voice. This strategy is currently being used by President Bush (a turning of the tables). Well, in biology, we need to rely on many strong, teleological concepts to fuel our experimental approach. Heck, we're dealing with a topic were we must rely on concepts such a "codes," "machines," "checkpoints," "feedback," "circuits," etc., yet this is all housed on an overall non-teleological approach. Well, sheesh, for the teleologist, the only toy left to play with is coming up with some "extraordinary" demonstration/proof of design. Doesn't sound like a good recipe for forming and testing hypotheses to me.
At this point in time, teleologists need to retrain their minds to think about biotic reality in teleological terms, not just in anti-non-teleological terms. Thus, this is a period of time when teleologists need to think big and reinterpret the gobs and gobs of data that not only exist, but have been interpreted in non-teleological terms. The new experiments will inevitably follow.
IP: Logged
|
|
Mike Gene
Member
Member # 149
|
posted 03. July 2002 02:26
There is one more point I forgot to stress. It's difficult to read someone's argument from a Power Point demonstration, but the gist seems to be that Elsberry belongs to the "designer-dependent" school of thought. That is, he (and many others) argue that the only way to detect design is to make use of the independent information we have regarding the designers. Of course, this is a very effective method if one wants to naysay ID (and its clear that the objective of many is to find ways to naysay ID), but it is a useless approach for someone trying to flesh out teleological suspicions about biotic reality. Take any proponent of the designer-dependent school of thinking. Ask them a simple question, "How could I go about gathering evidence to support the hypothesis that the flagellum (or something like it) was intelligently designed?" Their answer would be, "Er,.....um,......find the designer!" In other words, their answer is useless and helpless. The investigator is advised to become Indiana Jones, searching the Universe for the hidden designers and their protocols. Is it a coincidence the folks who exert great energy naysaying ID would offer such useless and helpless advice?
In contrast, I find the ideas of Behe and Dembski to possess utility.
Lastly, Frances earlier writes: First of all ID proposes a reliable filter for ID, that is, no false positives.
Since when does reliability translate as infallibility?
Perhaps I should start walking to work, as my car could break down (it's happened), making it an unreliable method of transportation. But wait, I could twist my ankle. Hmmm. There is no reliable way to get to work. No choice but to sit at home, I guess. Nothing like a little philosophical justification to become a couch potato. lol
[ 02 July 2002, 14:25: Message edited by: Mike Gene ]
IP: Logged
|
|
Micah Sparacio
Member
Member # 6
|
posted 02. July 2002 15:24
I'm curious, Mike, how did you happen to post on July 3rd at 2 in the morning (it is only July 2nd at 3:21pm Eastern US while I write this post). Curious.
IP: Logged
|
|
Mike Gene
Member
Member # 149
|
posted 02. July 2002 20:37
Impressed, eh? Many ID critics seem to want time travel, as this is the only way I can think of when trying to capture/study the designer. So...I've been working on it.
[ 02 July 2002, 20:39: Message edited by: Mike Gene ]
IP: Logged
|
|
Jay
Member
Member # 268
|
posted 02. July 2002 23:45
Hi Mike,
I like the point about reclaiming concepts. How many times have we heard the phrase "evolution designed object so and so" or "evolution created", "evolution perfected" etc... this 'evolution' starts to sound like some giant, myterious machine-creator in some of our lab meetings.
To clarify this even more, for those of you working in the lab or taking classes, when they say things like 'evolution created', or 'nature designed', etc... substitute mentally "object x happened to appear through a series of accidents that happened to persist". It will sound ridiculous in almost all cases.
When working, we almost never talk about the systems that we study as though they were accidents! We always approach them as though they had purpose and were built well. We have indeed co-opted teleological terms and called them naturalistic.
Also, I once again see the need on the part of some critics to have ID essentially proven before some will take it seriously. I want to again remind the interested reader that such proof is not needed by the same critics to infer something as important as homology, which, again, often suffers drastically from lack of evidence either for the scenario (divergence from an ancestor, or homology) or for competing scenarios (convergence). Without even knowing the probability of the favored scenario or the competing ones, the inference is still made. Yet when it comes to ID, we now suddenly need solid proof. There is a clear disparity here in judgement.
This says to me that the ID theorist should take some comfort and realize that ID is actually doing quite fine as far as these origins inferences go. It also says that the critic who does not need ID and is just fine without it is not likely to ever be convinced short of extraordinary proof. So it tells me that I should concentrate on working out this inference of mine, and while trying to understand honest criticism, should not get discouraged or bogged down by those that claim that the ID inference has 'no evidence'.
Thanks,
jay
[ 02 July 2002, 23:46: Message edited by: Jay ]
IP: Logged
|
|
Frances
Member
Member # 169
|
posted 03. July 2002 00:14
Mike: Lastly, Frances earlier writes: quote:
First of all ID proposes a reliable filter for ID, that is, no false positives.
quote:
Since when does reliability translate as infallibility?
quote:
Perhaps I should start walking to work, as my car could break down (it's happened), making it an unreliable method of transportation. But wait, I could twist my ankle. Hmmm. There is no reliable way to get to work. No choice but to sit at home, I guess. Nothing like a little philosophical justification to become a couch potato. lol
I agree that it is quite funny but Dembski indeed suggests infallibility when he argues no false positives.
quote:
And this brings us to the problem of false positives. Even though the Explanatory Filter is not a reliable criterion for eliminating design, it is, I argue, a reliable criterion for detecting design. The Explanatory Filter is a net. Things that are designed will occasionally slip past the net. We would prefer that the net catch more than it does, omitting nothing due to design. But given the ability of design to mimic unintelligent causes and the possibility of our own ignorance passing over things that are designed, this problem cannot be fixed. Nevertheless, we want to be very sure that whatever the net does catch includes only what we intend it to catch, to wit, things that are designed.
I argue that the explantory filter is a reliable criterion for detecting design. Alternatively, I argue that the Explanatory Filter successfully avoids false positives. Thus whenever the Explanatory Filter attributes design, it does so correctly.
or more recently
quote:
By contrast, I contend that specified complexity be a reliable criterion for detecting design in the sense that it does not give rise to false positives (i.e., attributions of design that end up later having to be overturned). The justification for the criterion's reliability is an inductive generalization: In every instance where the complexity-specification criterion attributes design and where the underlying causal story is known (i.e., where we are not just dealing with circumstantial evidence, but where, as it were, the video camera is running and any putative designer would be caught red-handed), it turns out design actually is present; therefore, design actually is present whenever the complexity-specification criterion attributes design.
Of course the alternative is to accept that there will be false positives but then the ID filter becomes far less convincing since it now has to deal with the possibility of identifying design where there is none. Since ID ala Dembski is based on eliminative evidence and not on positive evidence this would mean that unless we can find some of this positive evidence, the design inference has lost much of its appeal. The same problem applies to IC. If ICness were a reliable indicator of design, that is, no natural pathways could possibly exist, then identifying IC systems would be enough to infer design. This is what Behe attempted to do. Of course since indirect pathways were at least allowed as possible, although considered unlikely by Behe, ICness also is not a reliable indicator of design with no false positives. The moment false positives are allowed, ICness or ID inference by relying on eliminative arguments fail. Note that this does not mean that there is no design but now positive evidence of design is needed.
I found this interesting comment on christianitytoday.com
quote:
And of course he is right: our clever code that produces a "false positive" is not detachable. But there is an obvious problem: the DNA code is not detachable either! Certainly Dembski cannot believe that a mathematician might have figured out that life on earth would be based on a code constructed of four protein bases arranged in the shape of a double helix. Wasn't God free to create life based on six protein bases, or from some other arrangement altogether?
Dembski is caught in a dilemma. If he gives up the detachability criterion, there will be no end of "false positives." If he doesn't give up the detachability criterion, he has reduced biology to a purely mathematical, a priori discipline, and he has denied a fundamental doctrine of Christian theology—namely, that God was perfectly free to create in any way he saw fit.
Jay writes
quote:
Also, I once again see the need on the part of some critics to have ID essentially proven before some will take it seriously.
Nope, what some of the critics, or in this case, this critics wants is that IDers realize and accept that the ID inference does suffer from 'false positives' and thus that the ID inferences by virtue of being eliminative has a lot of problems. Thus rather than relying on negative, eliminative evidence, this ID critic wants positive evidence before IDers generate claims without much further ado that natural forces and chance could not have generated a particular structure. After all, since ID is eliminative, would this not presume that the relevant probabilities have been generated?
On the other hand, one may argue that evolutionary science suffers from the same problem but as I have argued, it does not rely on elimination of alternatives, rather it proposes a hypothesis which can be tested and falsified.
So I repeat my question: What would ID contribute to science here? Jay suggests that ID could contribute, so I wonder how? And more importantly how would ID be a better explanation than its alternatives? What evidence do we have that indicates design in nature? Unless we can generate some positive hypotheses of ID, what does it contribute? What constrains an intelligent designer?
I have seen a few suggestions. Mike suggest that life was "front loaded" through bio-engineering by some form of being with human like intelligence. While I would argue that this is hardly the designer of interest to most ID supporters, I want to point out that science can perfectly well deal with such regularities. But what evidence is there to support such an intelligent being(s), what supports the idea that human like intelligence could front load life to allow it to evolve into the myriad of species we now recognize. What exactly was front loaded? Jay proposes a similar scenario. In both cases it is not evolution and its mechanisms that are doubted but rather the origin of life which is suggested to be intelligently designed.
So now we have two competing hypotheses, intelligent design and abiogenesis. What positive hypotheses does ID propose? How can we detect design if not through elimination of natural pathways.
Abiogenesis and OOL research has addressed many aspects of how life may have arisen, and while they hardly have found THE answer, I would argue that their scenarios seem to be much stronger than "some intelligent designer did it through bio-engineering", how does this explain anything or more relevantly how can it be shown to NOT explain everything. That is, is such an argument constrained? In what sense?
I want to again remind the interested reader that such proof is not needed by the same critics to infer something as important as homology, which, again, often suffers drastically from lack of evidence either for the scenario (divergence from an ancestor, or homology) or for competing scenarios (convergence). Without even knowing the probability of the favored scenario or the competing ones, the inference is still made. Yet when it comes to ID, we now suddenly need solid proof. There is a clear disparity here in judgement.
This says to me that the ID theorist should take some comfort and realize that ID is actually doing quite fine as far as these origins inferences go. It also says that the critic who does not need ID and is just fine without it is not likely to ever be convinced short of extraordinary proof. So it tells me that I should concentrate on working out this inference of mine, and while trying to understand honest criticism, should not get discouraged or bogged down by those that claim that the ID inference has 'no evidence'.
Mike argues that Elsberry's and Wilkins argument about Raelian claims is flawed when it comes to code and machines and that a teleological explanation is superior. But that is begging the question. Why would a teleological explanation be superior in instances when it can be shown from science that perfectly natural pathways may exist? So why would teleology be required or why would it be superion?
Mike attempts to strengthen his argument by refering to terms that suggest 'design' but one has to realize that the usage of such terms hardly should be used as evidence of design. That would be similar to Paley's argument with similar flaws. Mike may argue that analogy is positive evidence, but unless it can be supported by independent and positive evidence, there is little or no hope that analogy will be useful.
Mike also objects to me suggesting that we need to identify the designer or that we need to identify the mechanisms. Since ID ala Dembski relies on probability scenarios, understanding motive and mechanisms are very important. For instance Dembski argued that the similarity of two figures in separate papers was evidence of 'design' but design in what sense? Was it deliberate? Was it accidental ? Unless we can establish the causal history of the event, what hope do we have to determine what kind of 'design' really was involved here?
Mike also objects to Wilkins approach to a better filter suggesting that their insistence on independent evidence is an effective method to naysay design. But alternatively one has to realize that any method which gives design a privilged treatment, such as Dembski's filter does by allowing it to be the default explanation once chance and regularity have been eliminated, would certainly be an effective method to suggest design where there need not be such.
So why would one insist on pathways for chance and regularity and not for design? Wilkins and Elsberry have very well pointed out that such a treatment, however tempting it may be, is one full of flaws.
[ 03 July 2002, 00:37: Message edited by: Frances ]
IP: Logged
|
|
|
Printer-friendly view of this topic