Uncommon Descent Serving The Intelligent Design Community

Does information theory support design in nature?

Categories
Computing
Design inference
Intelligent Design
Share
Facebook
Twitter/X
LinkedIn
Flipboard
Print
Email

Eric Holloway argues at Mind Matters that design theorist William Dembski makes a convincing case, using accepted information theory principles relevant to computer science:

When I first began to look into intelligent design (ID) theory while I was considering becoming an atheist, I was struck by Bill Dembski’s claim that ID could be demonstrated mathematically through information theory. A number of authors who were experts in computer science and information theory disagreed with Dembski’s argument. They offered two criticisms: that he did not provide enough details to make the argument coherent and that he was making claims that were at odds with established information theory.

In online discussions, I pressed a number of them, including Jeffrey Shallit, Tom English, Joe Felsenstein, and Joshua Swamidass. I also read a number of their articles. But I have not been able to discover a precise reason why they think Dembski is wrong. Ironically, they actually tend to agree with Dembski when the topic lies within their respective realms of expertise. For example, in his rebuttal Shallit considered an idea which is very similar to the ID concept of “algorithmic specified complexity”. The critics tended to pounce when addressing Dembski’s claims outside their realms of expertise.

To better understand intelligent design’s relationship to information theory and thus get to the root of the controversy, I spent two and a half years studying information theory and associated topics during PhD studies with one of Dembski’s co-authors, Robert Marks. I expected to get some clarity on the theorems that would contradict Dembski’s argument. Instead, I found the opposite.

Intelligent design theory is sometimes said to lack any practical application. One straightforward application is that, because intelligence can create information and computation cannot, human interaction will improve computational performance.
More.

Also: at Mind Matters:

Would Google be happier if America were run more like China? This might be a good time to ask. A leaked internal discussion document, the “Cultural Context Report” (March 2018), admits a “shift toward censorship.” It characterizes free speech as a “utopian narrative,” pointing out that “As the tech companies have grown more dominant on the global stage, their intrinsically American values have come into conflict with some of the values and norms of other countries.”

Facebook’s old motto was “Move fast and break things.” With the current advertising scandal, it might be breaking itself A tech consultant sums up the problem, “Sadly Facebook didn’t realize is that moving fast can break things…”

AI computer chips made simple Jonathan Bartlett: The artificial intelligence chips that run your computer are not especially difficult to understand. Increasingly, companies are integrating“AI chips” into their hardware products. What are these things, what do they do that is so special, and how are they being used?

The $60 billion-dollar medical data market is coming under scrutiny As a patient, you do not own the data and are not as anonymous as you think. Data management companies can come to know a great deal about you; they just don’t know your name—unless, of course, there is a breach of some kind. Time Magazine reported in 2017 that “Researchers have already re-identified people from anonymized profiles from hospital exit records, lists of Netflix customers, AOL online searchers, even GPS data of New York City taxi rides.” One would expect detailed medical data to be even more revelatory.

George Gilder explains what’s wrong with “Google Marxism”
In discussion with Mark Levin, host of Life, Liberty & Levin, on Fox TV: Marx’s great error, his real mistake, was to imagine that the industrial revolution of the 19th century, all those railways and “dark, satanic mills” and factories and turbine and the beginning of electricity represented the final human achievement in productivity so in the future what would matter is not the creation of wealth but the redistribution of wealth.

Do we just imagine design in nature? Or is seeing design fundamental to discovering and using nature’s secrets? Michael Egnor reflects on the way in which the 2018 Nobel Prize in Chemistry has so often gone to those who intuit or impose desire or seek the purpose of things

Comments
EricMH:
@Mung, did you determine if entropy = disorder?
No. :) It would depend on what you mean by entropy and what you mean by disorder. If you are talking Shannon entropy and the "order" or "disorder" of a probability distribution I am still looking at that. Can we tell how ordered or how disordered a probability distribution is? Patrick decided to get back into it over at PS. He said I could learn some facts from an article about entropy. One of those facts seems to be that entropy is "missing information." So entropy = information = missing information = disorder.Mung
October 30, 2018
October
10
Oct
30
30
2018
07:38 AM
7
07
38
AM
PDT
Upright BiPed:
There is so much wrong on this thread. It’s depressing.
Put on some blues!Mung
October 30, 2018
October
10
Oct
30
30
2018
07:30 AM
7
07
30
AM
PDT
@daveS, yes, I think your experiment is a way to detect design. The rub is in your second question. If it can be detected by AI, then it might be optimized by AI, potentially removing the human from the loop. You'll have to guarantee the picture generator has no knowledge of the AI picture grader, so the grader is an independent specification. If you can do this, then your AI grader will detect intelligent agency.EricMH
October 30, 2018
October
10
Oct
30
30
2018
07:26 AM
7
07
26
AM
PDT
PS to #116: A related question: Could a computer be programmed to detect design using the above process? In other words, is this a task that AI can perform?daveS
October 30, 2018
October
10
Oct
30
30
2018
06:54 AM
6
06
54
AM
PDT
EricMH, What do you think of this (unoriginal) idea for testing whether CSI is useful for detecting design: 1) Begin with an 800 x 600 array of pixels, all white. (Or another array size, if preferred). 2) An experimenter sets some probability distribution on the power set of the set of pixels. The experimenter would also have to lay out how she is going to calculate Kolmogorov complexity of sets of pixels. 3) Have someone else (the "subject") choose a subset S of the pixels, either deliberately designed or not. These pixels are then turned black, creating a picture of sorts, which is submitted to the experimenter. 4) The experimenter then calculates (or estimates) CSI(S) and uses this to decide whether S was deliberately designed. Repeat steps 3 and 4 until a sufficient number of trials have been performed. Do you think that is a valid way to, at least in principle, determine the efficacy of CSI?daveS
October 30, 2018
October
10
Oct
30
30
2018
06:21 AM
6
06
21
AM
PDT
@KF very interesting point on how mathematics comes from the law of identity and non contradiction. I will have to ponder this. For the thread in general, it is clear there are many different takes on how information theory applies to detecting design. The OP is Dembski's law of information conservation is corroborated by mainstream information theory's law of information non-growth, which proves natural processes cannot create CSI. So, despite the skeptical criticism, Dembski's work is mathematically sound. It's also common sense because math exerts no causal influence on the world, so the fact the world can be explained so well mathematically has no natural explanation, i.e. "The Unreasonable Effectiveness of Mathematics in the Natural Sciences". https://www.dartmouth.edu/~matc/MathDrama/reading/Wigner.htmlEricMH
October 30, 2018
October
10
Oct
30
30
2018
05:19 AM
5
05
19
AM
PDT
PS, that is also connected to the point that if a world now is (manifestly!) then something always was. For, true nothing, non-being has no causal powers. At the heart of this is that necessary beings are eternal, without beginning or prior cause, they are framework to any world. Such thinking is unfamiliar to the scientific mindset and yet it is its underpinning, metaphysics sets the table of possible beings and how they ever could be ordered in a world. Science then operates on that table. PPS, we properly distinguish order which may have low information content [e.g. asasas . . .] from functionally coherent and complex integrated organisation which is inherently of high information content and has a lot of meaningful information in particular. (Ponder the previous sentence as a string example.) One may also distinguish randomness, such as gtiq3789jskvutftu[o . . . .kairosfocus
October 30, 2018
October
10
Oct
30
30
2018
04:58 AM
4
04
58
AM
PDT
Folks, it may help some to understand that necessary entities are framework for any world to exist. For instance ponder that absent distinct identity, a world does not exist.By direct force of logic, we thus have A and ~A, therefore twoness. Thence immediately the panoply of numbers and the logic of structure and quantity that accompanies that, i.e. mathematics is part of the fabric of reality, at least in that core part that we ponder. This is then the deep foundation for the observations on how powerful mathematical reasoning is. And much more. KFkairosfocus
October 30, 2018
October
10
Oct
30
30
2018
04:48 AM
4
04
48
AM
PDT
Upright BiPed: "It’s depressing." Yes, it is.gpuccio
October 30, 2018
October
10
Oct
30
30
2018
01:42 AM
1
01
42
AM
PDT
UB, please, point to what is wrong, or at least to a sample. readers might learn a few things. thanks.PaoloV
October 30, 2018
October
10
Oct
30
30
2018
01:00 AM
1
01
00
AM
PDT
There is so much wrong on this thread. It's depressing.Upright BiPed
October 29, 2018
October
10
Oct
29
29
2018
07:58 PM
7
07
58
PM
PDT
EugeneS, Sorry, I don't understand at all this insistence on "function". If we search for extraterrestrial life, archeological artefacts, geologic events, organic traces, etc., we infer presence based on specific nonrandom patterns, NOT ON FUNCTION. This is real life science happening every day in different fields. I see gpuccio @91 tries hard to make "function" relevant, including forcing art into "function", but he's totally unconvincing. Meanwhile, I propose a Design detection process that should be crystal-clear and uncontroversial TO ALL except where I equate 'necessity' with 'design'. What's wrong with that proposal?Nonlin.org
October 29, 2018
October
10
Oct
29
29
2018
03:52 PM
3
03
52
PM
PDT
SA, "NS doesn't do this or that" is much weaker argument than "there is no NS". Why go half measure? I am with the amats on this. But let's see what actually happens in nature. I conclude (with proofs) that they're all illogical: NS, micro/macro-evolution, "random mutations", "divergence of character", "DNA is recipe", etc.Nonlin.org
October 29, 2018
October
10
Oct
29
29
2018
03:38 PM
3
03
38
PM
PDT
Nonlin, "If ‘necessity’ is not produced by the same Designer, then what’s the source of that ‘necessity’?" Everything that had a beginning was designed. Full stop. However, there is the scientific method with its limitations of scope and its power of inference, intended for mundane pragmatic purposes. If you do not distinguish between order and function, you will not achieve much in this sense (hence my criticism of sloppy reasoning). I agree with you that everything, apart from God, is designed (if I understand you correctly), but in terms of making scientifically non-trivial claims, you need that distinction.EugeneS
October 29, 2018
October
10
Oct
29
29
2018
03:12 PM
3
03
12
PM
PDT
@Nonlin.org The information theory aspect of ID is not highlighted as much for a variety of reasons. I think the main reason is math is not something a lot of people find easy to deal with. It's easier to grasp specific tangible things, and it is hard to believe an abstract subject like mathematics could conclusively demonstrate anything in the physical world. There are also a lot of criticisms from professors like Shallit and Perakh, and without a solid grasp of Dembski's math, it's hard to know if their criticisms are correct or not. Regarding necessity and design, on a philosophical level the AMATs don't really have much to stand on. As you point out, nothing can be granted, because nothing comes from nothing. So, I agree, necessity and order cannot come from anything other than an intelligent designer, because they are something.EricMH
October 29, 2018
October
10
Oct
29
29
2018
02:56 PM
2
02
56
PM
PDT
nonlin
…pretty much same as mine including “how can you accept ‘natural selection’ and ‘micro-evolution’, but not macro-evolution”?
Theoretically that would be consistent. If selection preserves small changes, then over time it should mean macro-evolution. But in this case, the empirical does not match the theory. Mutations do not have the innovative power or frequency and selection does not have the time needed to extrapolate from micro to macro levels. Science cannot explain why organisms withstand micro-evolutionary change but resist change of species. Adaptations fluctuate around a mean and organisms have something inbuilt in them that resists macro-changes and which preserves (heals, restores) them as a species. For Darwinism, species are always on their way to becoming something else and with mixed environments and non-directional development there's no reason why they are separated by distinct features.Silver Asiatic
October 29, 2018
October
10
Oct
29
29
2018
01:29 PM
1
01
29
PM
PDT
ErichMH, Big mistake granting this and that. You're also granting "natural selection", "micro-evolution" and "laws of nature". Of course "random" is unknowable from the outcome and has been greatly abused by amats. And what is the origin of Turing machines if not design? If you don't mind, I must insist with one question: "If ‘necessity’ is not produced by the same Designer, then what’s the source of that ‘necessity’?" ...and one more new one: How come Evolution News returns nothing on "specified complexity"? Don't they agree with Dembski?Nonlin.org
October 29, 2018
October
10
Oct
29
29
2018
11:13 AM
11
11
13
AM
PDT
gpuccio @91, If you don't wan't to discuss, don't :) But I will reply to your comments and will comment on your views anyway. There's a clear definition of order: "the arrangement or disposition of people or things in relation to each other according to a particular sequence, pattern, or method". No chance to get confused. I know "sensitivity and specificity". Don't worry. You're just repeating your argument but not clarifying anything with "descriptive/prescriptive information". Look at how little traction Dembski got, and look at the criticism against him: http://wasdarwinwrong.com/kortho44.htm ...pretty much same as mine including "how can you accept 'natural selection' and 'micro-evolution', but not macro-evolution"? 'Function, ‘contingent’, ‘specified’, ‘complex’ and their arbitrary thresholds are unnecessary complications - read again comment 65 to see how I improve on those... Sorry, I see your disagreements, but no solid counterarguments.Nonlin.org
October 29, 2018
October
10
Oct
29
29
2018
11:01 AM
11
11
01
AM
PDT
@Nonlin.org, sorry for the misunderstanding. We can get regularity from a Turing machine, which is a generalization of all finite mechanical processes. An extremely long string of 1s has a very short program, so is very likely. No designer required, if we grant the AMATs a Turing machine. Of course, why grant AMATs anything, including randomness? Nothing comes from nothing, and randomness is something. However, we are working within the context of modern science, which already grants randomness and Turing machines. So, we have to have some third causal source besides randomness and Turing machines if we want to say ID has something to offer modern science.EricMH
October 29, 2018
October
10
Oct
29
29
2018
09:48 AM
9
09
48
AM
PDT
EricMH @90 I didn't say "all amats". Just "many". If 'necessity' is not produced by the same Designer, then what's the source of that 'necessity'? I am definitely not a fan of Spinoza. You're misreading. Not sure what "fitting"/"overfitting" means. Also not sure who believes "infinite chaos". Ask any amat and they would say "natural selection" is not random.Nonlin.org
October 29, 2018
October
10
Oct
29
29
2018
09:27 AM
9
09
27
AM
PDT
@Bob O'H, CSI hasn't changed dramatically, I just added another term for the specification. The rest is the same. The dependency issue is taken care of by establishing a penalty for dependency, which is the size of the specification. I had to add the dependency penalty because currently there is no well established approach to quantifying how independent a specification is. In the objective case we have to consider all possible specifications, because they are no longer limited by what humans can dream up. In the subjective case, independence was a heuristic. We just assumed descriptive language of some unrelated phenomena, such as human designed artifacts, were independent of biological phenomena that we'd never discovered before. Seems reasonable enough. But it isn't a quantitative heuristic. I just made specification dependence quantitative so we can talk about the objective CSI that is independent of observers. Addressing the math. Since we're minimizing over all S, we can remove the S from oCSI, giving just oCSI(X,P). What we want is to guarantee E[oCSI(X,P)] <= 0, so that give enough measurements we can definitively eliminate P as a valid source of X, and that criterion is met by oCSI.EricMH
October 29, 2018
October
10
Oct
29
29
2018
09:20 AM
9
09
20
AM
PDT
Eric @ 95 - so -CSI is now the sum of Shannon and Kolmonogorov? And there is no log N? And how does this reduce the dependence from taking the maximum? I'm now utterly lost trying to understand what's going on. I suspect I'm not the only one.Bob O'H
October 29, 2018
October
10
Oct
29
29
2018
09:06 AM
9
09
06
AM
PDT
@Mung, did you determine if entropy = disorder?EricMH
October 29, 2018
October
10
Oct
29
29
2018
09:04 AM
9
09
04
AM
PDT
EricMH:
I agree we are talking about a measure called SMI. What does it measure?
That is the question! Is it measuring information and in what sense does it quantify "an amount of information." But first, whatever it is measuring, it has to do with the probability distribution. To lose sight of that makes it easier to go astray. Does it measure "information" or "information content"? If it does, it is measuring the information or information content associated with a probability distribution. If you have two distributions that are the same you will get the same quantity regardless of "message content." As to what it measures, there are a variety of interpretations about that. Are some interpretations better than others? Are some interpretations of "entropy" better than others? I believe so, but that can wait for a future post. Choice, uncertainly, surprisal, information ... entropy = all the above? LoL.Mung
October 29, 2018
October
10
Oct
29
29
2018
08:53 AM
8
08
53
AM
PDT
gpuccio:
I think that, if we keep our concepts well defined and separated, everything is rather clear.
That is exactly what I am trying to encourage! :)Mung
October 29, 2018
October
10
Oct
29
29
2018
08:37 AM
8
08
37
AM
PDT
@Bob O'H, here's another take on your objection. There are an infinite number of possible specifications, so the log2 N approach always gives everything an infinite specification, making all events have negative CSI. So, we need a way of indexing specifications that doesn't require an infinitely long number. One way is to index each specification by its optimal compression, K(S), so every specification has a finite index. In this way, the CSI for an event is calculated as -log2 P(X) - K(X|S) - K(S), which is the same as -log2 P(X) - K(X,S). Let's call this objective CSI (oCSI): oCSI(X,S,P) = -log2 P(X) - K(X,S). Thus, the objective CSI is found by minimizing oCSI(X,S,P) over all possible S, where the dependence penalty is K(S).EricMH
October 29, 2018
October
10
Oct
29
29
2018
05:58 AM
5
05
58
AM
PDT
@Bob O'H, those are good points. Picking from the subset doesn't make the empirical measurement completely objective, but gets closer to the objective value. Using log2 N introduces a penalty for the reduction of independence you noted. There are two questions getting combined in your responses. One question is whether there is such a thing as objective CSI. Another question is how our subjectively derived specifications (from Bill and Bob) approximate the objective value. My answers are dealing with the latter scenario you proposed.EricMH
October 29, 2018
October
10
Oct
29
29
2018
02:57 AM
2
02
57
AM
PDT
Oh, I hadn't seen that part of the specification. But I don't get the point - you will (presumably) add log2 8 bits to every specification, so you still end up with the same ranking. The absolute value will change, so "the" value of CSI is subjective in that it now depends on the number of specifications. All of this still deosn't get around the problem that if you take the maximum of the specifications, the specification isn't independent of the data.Bob O'H
October 29, 2018
October
10
Oct
29
29
2018
01:42 AM
1
01
42
AM
PDT
@Bob O'H, in the case of selecting from multiple specifications, we have to make choosing the specification part of the description length. For instance, if we have 8 possible specifications then we add log2 8 = 3 bits to the description, because we'll need 3 bits to pick one of the specifications.EricMH
October 28, 2018
October
10
Oct
28
28
2018
05:18 PM
5
05
18
PM
PDT
Nonlin.org: I still don't want to discuss with you again. Just a list of suggestions: 1) Try to understand the difference between descriptive information and prescriptive information, as clearly discussed by Abel, and how both are designed (if complex). 2) Note that only intelligent agents do science. 3) You are very confused about order. I cannot even understand what you mean with that word. Maybe you could try to offer explicit definitions, instead of vague and conflicting "ideas". 4) I am not bitter. Just tired of having to do with you. 5) Design has always a purpose. That purpose can be to do something by the designed object. In that case we speak of prescriptive information, and of function, That's what we observe in biological objects. Of course, design can also generate descriptive information: a book, a poem, a painting. The purpose here is to express conscious experiences and to evoke similar experiences in others. Or just to convey meaning. We use function definition to detect prescriptive information. The problem is not that we miss a lot of designs. As I have explained many times, the purpose of detecting design in biological objects is not to have high sensitivity, but absolute specificity. In case you are interested in real science, the tradeoff between sensitivity and specificity is one of the main points in understanding empirical science. Have a look at that. However, we can define "functions" for descriptive information too. Not in terms of "doing something" with the object, but rather in terms of "conveying some experiences or meanings" by the object. That approach can be used to measure functional information, too. For example, a piece of text can convey specific meanings, or not. I have used that kind of definition in my OPs about English language. But that approach is not specially useful with biological objects, that carry mainly prescriptive information. OK, that's already too much discussion for now.gpuccio
October 28, 2018
October
10
Oct
28
28
2018
03:11 PM
3
03
11
PM
PDT
1 15 16 17 18 19 20

Leave a Reply