Complexity, Specification, Design Inference, and Designers
| February 21, 2006 | Posted by Dave S. under Intelligent Design |
I often see misunderstanding of what ID is about. It’s about inferring design by critical analysis of a pattern and the ways that pattern could have come to exist. I find a comparison with a lottery to be the easiest way to understand this.
Suppose there is a state lottery and each month for 12 consecutive months 10 million tickets are sold and one winning ticket is drawn at random. Obviously there must be 12 winners at the end of the year. While each winner beats odds of 10 million to 1 there’s nothing unusual about that as someone must beat the odds each time.
Now suppose that the 12 winners are all siblings in order from oldest to youngest.
This lottery result constitutes a pattern.
First of all we have complexity in the pattern. The odds of any particular sequence of 12 winners are 1 in 10^84 (that’s 10 followed by 84 zeroes). Any single pattern where there are trillions and trillions of patterns possible is complex. But complex things like this happen all the time because the result must be one of those many sequences. A sequence of 10 coin flips, no matter the result, is not complex as there are only 1024 possible results. This is roughly how we define complexity. Complex results happen all the time and in themselves are no indication of design.
Next, the pattern has specification. The pattern conforms to an independently given specification. In this case siblings from the same family is the indendently given specification.
Now we have identified the lottery result as a complex specified pattern (or complex specified information if you will). This is a reliable indicator of design. The more complex the result and the more definitive the pattern the more reliable the design inference.
No matter how convincingly it can be told that the lottery was secure from cheating no reasonable person will be convinced that there was no cheating involved. So we can almost certainly rest assured that the result of the lottery was not random but was the result of design (cheating; rigged).
However, even though we know the result was rigged we have no clue how it was rigged (the mechanism) nor who did the rigging (the designer).
ID is the theory that certain patterns in nature exhibit specified complexity that can only reasonably be attributed to design. ID does not and cannot reveal how the design was accomplished nor what entity or entities did the designing. ID is nothing more or less than design inference based upon high improbability of independently given patterns arising by chance.
Now let’s quickly look at the flagellum. There’s no room for debate about complexity. It’s a precise arrangement of millions of protein molecules from a set of dozens of different proteins, each protein itself a complex pattern. There’s little room for debate that it conforms to an independently given pattern. It’s a propulsion device. Where there is room for debate is in what Bill Dembski calls “probabilistic resources”. These are the resources that “chance” (or unintelligent cause) has to draw upon in forming the pattern. This is why ID seems to be an attack on mutation & selection. Mutation & selection are the leading known probabilistic resource that could form the specified complexity of the flagellum.
Logically one can never prove a negative. ID proponents will never be able to prove that some unknown probabilistic resource wasn’t the source of design in the flagellum. However, this is a problem with nearly every hypothesis in science and it’s why you often hear that all of science is tentative. Some bits are just more tentative than others. This is why most philosophers of science say a hypothesis has to be, at least in principle, falsifiable. If we can’t prove something is true, if we can at least be able to prove it false in principle, then it’s science. The ID hypothesis of the flagellum is falsifiable. In principle a neoDarwinian pathway for its evolution can be plotted on paper and confirmed in a laboratory.
The greater question in my mind regarding falsifiability is whether there’s any method in principle of falsifying a hypothetical neoDarwinian pathway for the flagellum. The only real contender for falsification is a design inference! So you see, if ID didn’t exist, neoDarwinists would have to invent it just so they have a method of falsification in principle for random mutation plus natural selection in creating things like the flagellum.
45 Responses to Complexity, Specification, Design Inference, and Designers
Leave a Reply
You must be logged in to post a comment.
I think that Chris and I are driving at the same point, namely that complexity assessments are often poorly justified. This problem is aggravated by the frequent conflation of Dembski’s definition of complexity with the more common definition, which leads some to attribute self-evident complexity to events with potentially complicated causal histories.
(Note: This is one of 4 main issues that I have with CSI, this one being the least problematic. I can state the other 3 if anyone’s interested.)
Secondclass, my impression is that Dembski sees specification in BOTH simple and complex patterns.
From No Free Lunch:
“What is specified complexity? An object, event, or structure exhibits specified complexity if it is both complex (i.e., one of many live possibilities) and specified (i.e., displays an independently given pattern). A long sequence of randomly strewn scrabble pieces is complex without being specified. A short sequence spelling the word “the” is specified without being complex.”
So “the” is specified without being complex…
And from ISCID.ORG
“The second component in the notion of specified complexity is the criterion of specificity. The idea behind specificity is that not only must an event be unlikely (complex), it must also conform to an independently given, detachable pattern.”
One cannot work with Specified Complexity, absent a full appreciation of Independence.
Irving, you make a good point. I suppose a complex pattern could constitute a specification if it’s independently given, although recognition of such a pattern may, in some cases, be difficult or even intractable. This raises a question of whether CSI requires that the specification be smaller than the event.
This brings up another question I have regarding specifications. Must a specification contain a complete description of an event? In other words, must a spec contain enough information to reconstruct an event exactly? I see problems attached to both a yes and a no answer.
You’ve totally lost the plot here. A family of blood relatives is the specfication in the lottery example. A propulsion device is the specification for a flagellum. These are not intractible or difficult. I suspect you’re just being argumentative. In any case your comments are not constructive. Take a break from this thread. -ds
Dave, I’m sorry if I came across as argumentative. Irving said that specifications could be complex. You seem to be saying that specifications are simple. I’m in your camp, but I’m trying to be agreeable to everyone here.
My question as to whether a specification must contain a complete description of the event is a sincere one.
Specification is some distinguishing characteristic that gives meaning to a pattern. Independence means the meaning isn’t a tautalogy – the pattern must mean something independent of itself. Simple and complex, compete and incomplete description, are irrelevant since meaning can be construed independent of those terms. -ds
Dave, I think if we’d both properly read each other’s post, you would realize that a “Face on Mars” was not my concern. A discussion regarding nature of (wholly valid) pro-design vs. (will you at least consider considering) natural “apparent design” is a worthwhile, a reasonable, position, no?
I added more commentary to yours. I think you misunderstand my position. I don’t discount material mechanisms. However, science is about demonstration and no material mechanism has been demonstrated capable of producing novel cell types, tissue types, organs, organelles, or body plans. Extrapolation of a mechanism demonstrably able to produce small changes is a reasonable position as long as one doesn’t lose sight of the fact that it is extrapolation, might not be the correct answer, and has not even in theory been shown to have a plausible way of creating the observed complexity in living systems in the time and space and environment available. Intelligent design is the only other option on the table that many people see as a viable alternative and people have recognized it as an option for millenia. Furthermore there’s no reason in principle why the intelligent agency can’t be material in origin. In fact as a materialist I believe that the source of intelligence, when and if it is discovered and characterized, will be comprehensible in material terms. It’s difficult for me to grasp why any objective, rational person would exclude design as a live possibility for the origin of complexity in life on earth. I can only conclude that resistance is driven by irrational and/or subjective motives such as fear, ignorance, hubris, financial, and philosphical concerns. -ds
Dave, is independent meaning always a requirement? For instance, does the following sequence have a pattern with independent meaning? DDDDDDDDDDDDDDDDDDDDDDRDDDDDDDDDDDDDDDDDD
If incomplete descriptions are allowed, then is “string of a million bits” a valid specification? How about “10 ton rock”?
Independent meaning is always a requirement for a design inference. The sequence you give has no meaning to me so I can’t begin to make a design inference. However, it may have meaning I don’t know about. Maybe it’s the password someone used for their Swiss bank account. String of a million bits has no independently given meaning and neither does 10 ton rock. -ds
Dave, Dembski attributes CSI to the above sequence. (See here) What independent meaning does he see that you and I don’t see?
DDDDDDDDDDDDDDDDDDDDDDRDDDDDDDDDDDDDDDDDD
“D” represents “Democrat first on the ballot” and R is “Republican first on the ballot”. There is an advantage in being first on the ballot. That advantage is an independently given specification. A string of 41 Democrat advantages and 1 Republican advantage on a ballot is an approximate 1 in 1 trillion possibility and is therefore fairly complex. Thus we have complexity that conforms to an independently given pattern. The only reason I didn’t see it when presented with it before is the definition and significance of “D” and “R” were withheld. What now follows in a design inference is analysis of chance hypotheses (also called probabilistic resources) that could compose this CSI without intelligent agency involved. -ds
“No matter how convincingly it can be told that the lottery was secure from cheating no reasonable person will be convinced that there was no cheating involved. So we can almost certainly rest assured that the result of the lottery was not random but was the result of design (cheating; rigged).”
I don’t see this. All we can reliably infer from your lottery example is that the outcome is almost certainly not due to random chance. But cheating suggests intention, and a mechanism used in the service of that intention.
Perhaps the family works at the ticket factory and managed a not-so-clever swindle? Then our pattern would have been generated by cheating. But perhaps there was a printing error at the ticket factory and 12 identical tickets were all sent to the same store in sequence. If all the family members went to the store together or (more likely) one or two of them bought twelve tickets for the others, then we’d get the outcome you describe, but without cheating.
If you want to infer intentional design in the lottery case, you’re going to have to look at more than outcomes and likelihoods: you’ll have to dig further, finding out whether there were machine errors, and if not, then investigating the family members in question, and probably other lottery employees. We know the outcome is too improbable to result from simple chance, but until we figure out the mechanism generating the distribution we observe, that’s all we can reliably say about the outcome.
“However, even though we know the result was rigged we have no clue how it was rigged (the mechanism) nor who did the rigging (the designer).”
Again, we don’t know that the result was rigged: all we know is that it almost certainly wasn’t due to random chance. Something systematic, non-random is at work here, but we don’t know whether this is a machine error and a lucky shopping day for one family, or a not-so-clever inside job designed to cheat the lottery commission.
“ID is the theory that certain patterns in nature exhibit specified complexity that can only reasonably be attributed to design.”
Then it isn’t a very interesting theory. We study intelligent design in the historical and social sciences all the time, and if we stopped at simply saying “that education policy definitely didn’t arise from chance” then we wouldn’t be taken very seriously by anyone. It isn’t until historians and archeologists, for instance, tell plausible, evidence-based, and independently verifiable stories about designers and mechanisms that they do interesting descriptive and explanatory work.
“ID does not and cannot reveal how the design was accomplished nor what entity or entities did the designing.”
Then it’s not a very interesting scientific appproach. Science isn’t just the business of speculating about patterns (although a lot of good science starts that way … of course, plenty of bad science starts that way as well). Science is about understanding causal mechanisms. If you think important biological systems are designed, then start figuring out ways to identify and study designers and their mechanisms.
You completely overlooked the following:
Tickets are sold monthly for each monthly lottery 10 million of them. One winning ticket is drawn from that number. The scenario you outline, 12 winning tickets sold at one time, was not possible in that circumstance. -ds
Dave, it seems I’m completely confused regarding specificity. I see that the advantage of being first on the ballot is a motive for cheating, but I don’t understand how it constitutes a specification. My understanding of an independently given specification would entail a description of or pattern in the sequence that is recognizable even if the source of the sequence is unknown. Am I way off track?
The source of DDDRDDD need not be known. In fact I don’t know the source. As far as I know it could be a faulty random number generator or a 1 in 1 trillion odd happenstance. The point is that further investigation of chance hypotheses is warranted. -ds
Dave, now I’m more confused than ever.
1) My understanding of the detachability requirement is that the specification should be recognizable independent of any facts behind the occurence of the event, which is why I presented the sequence without mentioning the Caputo story.
You didn’t present the specification. You presented the sequence.
2) My understanding is that specifications are descriptions or patterns, not motivating factors.
Specifications can be anything sets the pattern in question apart.
3) I don’t understand what chance hypotheses have to do with specifications.
Chance hypotheses are a step in inferring design.
Sorry to keep bugging you, but I need some help understanding this.
Well Dave, it looks like you and I have very different definitions of specificity, detachment, etc. My guess is that we won’t be able to find enough common ground for a discussion, so I’ll let it go. Keep up the good work!
Secondclass, there is a difference between a specification, and specified complexity.
“My understanding of the detachability requirement is that the specification should be recognizable independent of any facts behind the occurence of the event…”
Perhaps…”independent of the factors behing the occurence…”
Actually, the odds “of any particular set of 12 people winning the lottery” are not “1 in 10,000,000,000,000,000,000″, but rather 1 in 2.1 times 10^75. That’s roughly 55 orders of magnitude off the mark. I know, I know, it’s pedantic, but typical of certain probabilistic calculations in the context ID.
Actually it’s 1 in 10^84 …….. 1/((10^7)^12). Looks like your math is worse than mine. At least I knew that the correct answer had to be a power of ten. I did it in my head and added 12 zeroes to 10 million when I should have raised it to the 12th power. I’m not sure what your excuse is but I’m dying to hear it. -ds
“Actually it’s 1 in 10^84 …….. 1/((10^7)^12). Looks like your math is worse than mine. At least I knew that the correct answer had to be a power of ten. I did it in my head and added 12 zeroes to 10 million when I should have raised it to the 12th power. I’m not sure what your excuse is but I’m dying to hear it. -ds”
Still off by 9 orders of magnitude, but getting closer. You overlook that there are many ways to draw the same 12 people from 10^7. In fact there are n!/(k!(n-k)!) [where n!=n.(n-1).(n-2)...1] ways to draw k people from a group of size n. Maple 9.5 tells me that’s roughly 2.1 10^75. The reason is as follows: you can order n people in n! ways. There are k! times (n-k)! ways to split n! up in subgroups of sizes k and n-k so you have to divide n! by that to get the result (aka binomial coefficient).
You’re right. I didn’t explicitely say the winners were an ordered set. It’s now explicitely an ordered set. Thanks for pointing out the ambiguity. -ds
I was just wondering has anyone come up with a good layman’s explanation of the Design Inference of Dembski. I have ready “Intelligent Design” and to be honest the chapter on the Design Inference lost me. It seems that Dembski has come up with a way to determine if something specified but I have been unable to grasp his mathematical arguments. Can anyone help me?