You may remember it was back on September 9 2014 that I first asked the illustrious Marks for some calculation justifying the following claim of his:
Dear Prof. Marks:
Here
http://humanevents.com/2014/08/19/biological-information-new-perspectives-from-intelligent-design/
you claimed
"we all agree that a picture of Mount Rushmore with the busts of four US Presidents contains more information than a picture of Mount Fuji".
I asked you there for the details of the calculation that would show this, but you did not reply on that page, so I'm asking again.
Could you please provide me with your calculation to justify this claim?
Regards,
Jeffrey Shallit
I asked again after 3 months, 6 months, 1 year, 2 years, 3 years, 4 years, 5 years, 6 years, and now 7 years.
Still no response.
That is the nature of intelligent design creationism. Lots of wild claims, lots of bluster, but no actual evidence.
Science deserves better than this kind of nonsense. A lot better.
34 comments:
"Science deserves better than this kind of nonsense."
I think, since this reference's Marks' words, the word "deserves" really should be "is". The crap Marks spews out is not science at all, but "deserves" implies that it is.
Good thing then that IDiocy is not science.
Probably hard to do precisely for Mt Rushmore itself, but could be done in a toy version.
Create a simple model of plate tectonics, e.g. two lines of blocks moving towards each other, randomly colored either black or white, and when they collide, randomly causing one block to push up the other blocks that are in the middle. That could be the "natural process" that creates the toy mountain, and if we look at a particular section of blocks on the mountain, it's complexity could be calculated as the number of blocks, since each block is equally likely to be either black or white.
Then the context would be the logo programming language, and the specification would be the necessary commands to draw a section of the mountain. Since we are essentially drawing random noise, the logo program will be much larger than the number of blocks, and the mountain will have negative CSI.
Now, compare this with a situation where the mountain displays a smiley face formed by white blocks with a background of black blocks. The complexity will still be the number of blocks used, given the "natural process" described above, but the logo program will be very concise, especially if the smiley face is very large, thus the mountain will have a large amount of positive CSI.
There you go, challenge met!
Well, there are many, many things wrong with your proposal, Eric, but the most obvious thing is that Marks's claim is about the photos, not the mountains.
Another problem: random images contain MORE information, not less, than structured ones, at least in the definition of "information" that actual mathematicians and scientists typically use. CSI is not a valid measure and is not actually used in mathematics or science anywhere.
take a screenshot of the sim problem solved
csi is a kind of randomness deficiency, a well established concept in mathematocs
First sentence is incompehensible gibberish.
Second sentence is misleading. If CSI is used in mathematics and science, give citations to the peer-reviewed papers where it is applied.
Deficiency in randomness does not imply intelligence. Evident counterexample: salt crystals.
The mathematical form of algoritmic specified complexity (a CSI metric that uses conditional Kolmogorov complexity to measure specificity) is exactly the same form as a conditional randomness deficiency. You yourself use standard randomness deficiency as defined by Kolmogorov in the article you wrote with Elsberry, so you are just being obtuse.
Here is a peer reviewed article that uses CSI:
https://ieeexplore.ieee.org/document/6872591/
I quite enjoy these science/literacy expositions. Reminds me of literate programming where bad code is commented as bad code and not dressed up as anything else.
But randomess deficiency doesn't imply intelligence; I already gave a counterexample.
Where is CSI applied to definitively determine that intelligence created something that was previously unresolved?
There are a decent number of articles applying randomness deficiency to detect intelligent design, for instance network intrusion detection.
But none of them are citing Dembski's work, are they?
Nope, Dembski's work is not cited often outside of ID circles except to criticize. Which I find odd since all of his mathematics is very conventional, as we have discussed regarding CSI and randomness deficiency.
His mathematics is really bad, as the articles I have written show. He made trivial computation errors that led him to overestimate a probability by something like 60 orders of magnitude. His "theory" has proven to have no significant applications, despite the ridiculous and grandiose claims made for it.
Most mathematicians don't claim their work is revolutionary, while simultaneously not being cited by almost anyone who works in the field they claim to be an expert in.
//Then the context would be the logo programming language, and the specification would be the necessary commands to draw a section of the mountain. Since we are essentially drawing random noise, the logo program will be much larger than the number of blocks, and the mountain will have negative CSI.
Now, compare this with a situation where the mountain displays a smiley face formed by white blocks with a background of black blocks. The complexity will still be the number of blocks used, given the "natural process" described above, but the logo program will be very concise, especially if the smiley face is very large, thus the mountain will have a large amount of positive CSI.
There you go, challenge met!//
With this "method" you could prove the Mt. Everest, or indeed any other actual mountain, couldn't be produced by plate tectonics. What are the odds that you just randomly collide two plates and get the exact specific structure of a real mountain found somewhere on Earth? Basically zero. I guess mountains have to be wished into existence now, CSI says so.
Dr. Marks: "Yet we all agree that a picture of Mount Rushmore with the busts of four US Presidents contains more information than a picture of Mount Fuji."
Eric: "Then the context would be the logo programming language, and the specification would be the necessary commands to draw a section of the mountain. Since we are essentially drawing random noise, the logo program will be much larger than the number of blocks, and the mountain will have negative CSI.
Now, compare this with a situation where the mountain displays a smiley face formed by white blocks with a background of black blocks. The complexity will still be the number of blocks used, given the "natural process" described above, but the logo program will be very concise, especially if the smiley face is very large, thus the mountain will have a large amount of positive CSI."
Eric seems to be saying that the face legos have less information, not more information. This would seem to contradict Marks' statement.
As to the complexity of coding random processes, I would think from my experience (using pseudo-random number generators seeded by unpredictable events such as the clock time of key-presses) such programs are simpler than duplicating specific human faces. However, given the fact that the coders (or mountain sculptors) were themselves generated by billions of years of random events, I think a true accounting would assign equal complexity to both images.
The illustrious Robert Marks has, in a way, responded to the challenge. In the podcast “Define information before you talk about it: Egnor interviews Marks” at the site “Mind Matters News”, Michael Egnor says (according to the transcript):
“Dr. Jeffrey Shallot, who is a mathematician in Toronto, claims that Mount Rushmore doesn't have any more information than Mount Fuji. I'd like to ask my guest today Dr. Robert Marks to answer that question”.
Which is, of course, a lie. And, of course, Dr. Marks does not say anything about his (nonexistent) calculations. The discussion starts at 24:40 here: https://mindmatters.ai/podcast/ep158/
Enjoy.
Wow, four mistakes in one sentence.
They misspelled my name.
I'm not in Toronto.
Actually, I didn't make any claims at all about Mount Rushmore versus Mount Fuji at all. I asked for Dr. Marks to demonstrate his claim with a calculation.
And the discussion was not about the mountains, but rather pictures of the two mountains.
So incredibly sloppy, but that's what we expect from "Mind Matters News".
I listened to it.
Marks could not justify his claim at all. All he did was reassert it by saying, "It's obvious."
How would you do the calculation Dr. Shallit?
If you're talking about Kolmogorov information, it's not computable in general. But one could estimate it by taking the particular photographs in question and trying to compress them. Larger result = more information.
Marks's claims about "meaningful information" are gobbledygook, since there is no rigorous definition of what "meaningful information", nor any agreement about how to compute it. If there were, he could do the calculation he asserted.
So why do you think a picture of Mt. Rushmore is less compressible than Mt. Fuji?
Also you are wrong. Less compressible does not mean more Kolmogorov information, since compression is just an upper bound on Kolmogorov information. One compression algorithm may say Mt. Rushmore is less compressible than Mt. Fuji, and another algorithm may say the reverse, and a third may say they are equal. So, compression doesn't tell you at all which has more Kolmogorov information.
"So why do you think a picture of Mt. Rushmore is less compressible than Mt. Fuji?"
Why did you stop beating your wife?
If your question is based on a lie, don't expect an answer.
"Also you are wrong. Less compressible does not mean more Kolmogorov information, since compression is just an upper bound on Kolmogorov information."
Yes, thank you for telling me a basic result of Kolmogorov complexity that I teach every year in my course, which is that compression algorithms can only provide an upper bound.
"compression doesn't tell you at all which has more Kolmogorov information"
Every compression algorithm gives you a computable approximation, which is provably the best we can do. It is unreasonable to expect more, and the compression algorithms we have have proved very useful -- much more useful than Dembski's nonsense. See, for example, Ming Li's work on chain letters.
So, what is your complaint? We can use an approximation of Kolmogorov complexity to calculate an information measure for the photos. As we discuss previously in this thread, we can also use Kolmogorov's randomness deficiency metric, which is the same thing as ASC. There's no problem with Dr. Mark's claim. It's well supported by Kolmogorov's well known work.
I see lots of babble, but still no calculation. Why are ID advocates so afraid to produce some numbers? William Dembski did one calculation in his book, and it was off by 60 orders of magnitude.
Provide the calculation showing that, and I quote, "a picture of Mount Rushmore with the busts of four US Presidents contains more information than a picture of Mount Fuji".
Oh, no; you failed to commemorate the 8th anniversary!
It seems you've missed the "The Robert Marks Evasion: 8-year anniversary"
Seems you've missed this year's aniverssary
Seems like you've missed the 8th anniversary.
@Jeffrey Shallit you are an intelligent individual, I'm sure you can do the calculation using a CSI approach that does justice to Dembski's formulation and comes out correctly. You're just waiting for something to nitpick and declare ID defeated yet again :)
@Mikkel Rumraket Rasmussen your comment seems to have disappeared, but I have it in email. You're basically missing the specification portion of what I described. Plates randomly colliding as I describe will most often not be concisely describable, unlike the large smiley face.
Why do creationists always want to reverse the burden of proof?
If the calculation is so easy, as ID creationists keep saying, why can none of them actually do it?
I already gave an example how to do the calculation for a toy example. What's the problem?
You didn't actually provide any numbers!
And you didn't provide the calculation for the particular claim.
Two strikes already.
Post a Comment