Talk:The Search for a Search - Measuring the Information Cost of Higher Level Search

From RationalWiki
Jump to: navigation, search
Icon creationism.svg

This Creationism related article has been awarded BRONZE status for quality. It's getting there, but could be better with improvement. See RationalWiki:Article rating for more information.

Copperbrain.png

I know that there is not much of an interest for this kind of work here :-) But this wiki is so well-suited for such a critique, much better than my blog....

larronsicut fur in nocte 12:28, 21 October 2009 (UTC)

No, by all means, while it may be a bit arcane, it's spot-on with our mission. Have you linked to your essays on his papers from William Dembski? If not, you should. ħumanUser talk:Human 20:37, 26 October 2009 (UTC)
I've added it there. It's certainly well on mission to challenge this sort of thing. Scarlet A.pngpostate 20:40, 26 October 2009 (UTC)
Thanks! And I do apologize for the arcane language... larronsicut fur in nocte 22:00, 26 October 2009 (UTC)
Well, by arcane I just meant that some readers might not be able to follow it, due to the content/subject matter. Also, wouldn't it make sense to move them both to the mainspace sometime? ħumanUser talk:Human 22:43, 26 October 2009 (UTC)
Main space? Sounds great! larronsicut fur in nocte 23:22, 26 October 2009 (UTC)
Dang, and you didn't think your work was mainspace-worthy? Sheesh dude. ħumanUser talk:Human 05:29, 27 October 2009 (UTC)
Of course it's worthy, but will everyone think so? Or will it get deleted? Well, I'll just move it... larronsicut fur in nocte 13:13, 27 October 2009 (UTC)

Done for now[edit]

I laid down the main points of my argument - but of course there is much more to do... larronsicut fur in nocte 23:32, 8 November 2009 (UTC)

Should I crow?[edit]

The article A Search for a Search isn't linked any longer from R. Marks's page of publications. And the the Evolutionary Informatics Lab doesn't list under its publications any longer...

I'd written to R. Marks (and W. Dembski) at first on Oct 17th, 2009, and got an out-of-office reply an instant later. I wrote again on Oct 21st, 2009, and hinted him to the problems I had with their article, again. On Nov 2nd, 2009, R. Marks wrote back, thanking for the note and promising to look into it.

On Nov 8th, 2009, I restated and elaborated my concerns. This time, the out-of-office reply took a couple of hours...

Today (Nov 9th, 2009), I was made aware that the article A Search for a Search isn't linked any longer from R. Marks's page of publications and that the Evolutionary Informatics Lab doesn't list it under its publications any longer...

Is something happening?

larronsicut fur in nocte 12:39, 9 November 2009 (UTC)

Wow, the Evolutionary Informatics Lab seems to be quite disingenuous: it still has the announcement for the article, with all the nice buzz words (Horizontal No Free Lunch Theorem, Vertical No Free Lunch Theorem), but it doesn't link to the article A Search for a Search as it did earlier - instead the link goes to the article Conservation of Information in Search, where one fails to find these terms... larronsicut fur in nocte 13:20, 9 November 2009 (UTC)

I don't think that I'll add the remaining two sections: without the horizontal free lunch theorem, they don't make much sense... larronsicut fur in nocte 23:42, 9 November 2009 (UTC)

Dembski unhappy about this page. Threatens to sue.[edit]

See here. I don't think Dembski understands fair use. The comments thread has unfortunately been closed. JoshuaZ 04:28, 17 November 2009 (UTC)

Aleady discussed at length here. The verdict is, come and ave a go if ya think yer ard enough. I think that was it. -- =w= 04:31, 17 November 2009 (UTC)
I'm wondering if he spotted this because above we talked about linking to it from his RW entry and then added it. A short "self Google" by himself would have found it pretty quick as we're on the first page (albeit at the bottom). Scarlet A.pngpostate 17:00, 17 November 2009 (UTC)

No, I'm afraid not - I mailed R. Marks to present my critique of their article and linked to this page. Also, I tried to contact W. Dembski himself - no as in vain as I thought ... larronsicut fur in nocte 19:04, 17 November 2009 (UTC)

Ah I see. I'm not sure if that makes it worse that you basically said "hi guys, want to respond to this criticism" and Demski replies "SUE SUE SUE!!!!!" rather than him just stumbling across it. Scarlet A.pngpostate 19:22, 17 November 2009 (UTC)

I am not a lawyer, and I do not know if this is protected under "Fair Use", but I think it out not to be protected. As MCC points out, the decent thing to do is push for wide dissemination and commentary. However, if commentary were enough legal justification to "excerpt" a work in full, it would be almost impossible to enforce copyright. Coarb 07:36, 5 December 2009 (UTC)

Introduction[edit]

For the less intellectually able amongst us and for anybody who may land here if this becomes controversial, do you think that we could have a layman's introduction telling us what this is about and why it is important?--BobNot Jim 19:08, 17 November 2009 (UTC)

I'm inclined to agree. A lay summary (at least for a non-maths expert) and a few other wiki-links for context would be useful. Scarlet A.pngpostate 19:23, 17 November 2009 (UTC)
I just added another layer of complications The Search for a Search - Measuring the Information Cost of Higher Level Search#Section IV The Search for a Good Search, but I'll try to think about a down-to-earth introduction.
Summary of this article: D&M wrote an article filled with nice new words (active information, horizontal NFL), much impressive jargon, and failed, as they made a couple of sophomoric blunders. This article was intended to be a nail to the coffin of Darwinian evolution - though even if the article had any value, that would be quite a stretch.
Dembski announced that this article will appear in a peer-reviewed journal, therefore adding to his credentials. Most probably, it will not.
larronsicut fur in nocte 23:08, 17 November 2009 (UTC)

Denise O'Leary[edit]

A must read: Idea's aren't made for stealing

Dear Denise: These ideas I wouldn't even take as a gift, they are to flawed...

larronsicut fur in nocte 21:13, 18 November 2009 (UTC)

spoofed, wasn't I? Well done! larronsicut fur in nocte 21:15, 18 November 2009 (UTC)

Spoofed? I don't understand. -- =w= 21:18, 18 November 2009 (UTC)
it's a clever O'Leary - imposter (imposteress?), doing well for a lady of 92 years.
Of course it's easy for them to talk a good fight, but I doubt they'll be sounding so confident when the case goes to law, because one thing the Intelligent Design movement is renowned for is its record of performance in the courtroom.
That's just great! larronsicut fur in nocte 21:23, 18 November 2009 (UTC)
P.S.: One minor flaw: comments are allowed
I got it just before you posted. Now I feel pretty dumb. It's scarily realistic though - esp considering Dembski's recent objections. Damn you Poe's Law! -- =w= 21:25, 18 November 2009 (UTC)
A blog in the scienceblogs group has picked this up. I've stumbled across Good Math, Bad Math before: he's a big critic of Demski's information "theory." Sterile Xbox 00:41, 19 November 2009 (UTC) (My bad; that's in the Oleary post. cant reed)
Another post on this article, FYI. Sterile Xbox 22:05, 7 December 2009 (UTC)

W. Dembski is right - in Section IV[edit]

I wrote:

So, we're talking about Bochner integrals or even Pettis-Gelfand integrals. Very interesting stuff! Only one problem: Though Marks and Dembski undertook some pain to get the space right over which they are integrating, they have a problem with the function in this integral: The Bochner (and Pettis-Gelfand integral) \int f d\mu is defined for a function f: X \rightarrow B, where X\, is a measure-space and B\, is a Banach space, i.e., a complete normed vector space.

What's our f\,? It's the identity on \mathbf{M}_k, ergo

f: \mathbf{M}_k \rightarrow \mathbf{M}_k.

But though \mathbf{M}_k is a metric space, it's - in general - not a Banach space and so, the integral doesn't make sense, presenting quite a problem for Theorem IV.1 Conservation of Uniformity.

Well, he seems to be right for \mathbf{M}_k if k \ge 1 - as the space of the measures on a compact metric space is indeed a Banach space with the Wasserstein/Vasershtein metric. I got distracted by \mathbf{M}_0. And so, his Conservation of Uniformity is correct in his first formulation (though not his more general second one, here, he has to exclude k=0.)

My apologies to Dr. Dembski on this one - and everyone I was able to confuse here :-)

larronsicut fur in nocte 04:44, 21 November 2009 (UTC)

Paper finally published![edit]

On 29 August 2010 William Dembski informs us at Uncommon Descent:

Search for Search Paper Finally Out

William A. Dembski and Robert J. Marks II, “The Search for a Search: Measuring the Information Cost of Higher Level Search,” Journal of Advanced Computational Intelligence and Intelligent Informatics, Vol.14, No.5, 2010, pp. 475-486.

Link

My first reaction can be found here. I'll try to update the article over the next days... larronsicut fur in nocte 08:38, 30 August 2010 (UTC)

A short follow up:

In my previous post I criticized the Horizontal No Free Lunch theorem for searches consisting of at least two queries. But even in the case of a single assisted query, I can't get my head around it.

Dembski and Marks define an assisted query as any choice from Ω1 that provides more information about the search environment or candidate solutions than a blind search. As an example, they give for instance, we might imagine an Easter egg hunt where one is told “warmer” if one is moving away from an egg and “colder” if one is moving toward it.

The ultimate assisted search is no search at all, the unveiled Easter egg. So, imagine a blind (A), a drunken (B), and a seeing, sober (C) chess-player, and ask them to locate the White Dame (♕) on a chess board. (A) will be find it with a probability of 1/64, (C) with a probability of 1, and lets assume for (B) a probability of 1/2.

Then (A) represents the blind search, (B) an assisted search and (C) what the authors call a perfect search on our search space Ω = {a1,a2,....,h8} (|Ω|=64).

Imagine that ♕ is positioned on a1, so T1 = {a1}. Then we have three probability measures, αa1, βa1, and γa1 for each one of our players, which allow us to calculate the probability for each one to find T1:

  • αa1(T1)= 1/64
  • βa1(T1) = 1/2
  • γa1(T1) = 1

And we can calculate - according to (3.1) (10) the active information of these measures for various sets and combinations, e.g.

I_+ (\beta_{a1}|\alpha_{a1})\{a1\} = \log_2(1/2) - \log_2(1/64) = 5\;.

So, the drunken player has more active information on the whereabouts of ♕ than the blind one.

I_+ (\gamma_{a1}|\alpha_{a1})\{b1\} = \log_2(0) - \log_2(1/64) = -\infty\;.

A more surprising result: if ♕ is on b1, but (C) sees it on a1, he is much less informed than (A).

But of course, γa1 isn't the right measure if ♕ is on b1, we would expect γb1: the idea of an assisted search seems to be that it assists to the right direction - for our Easter egg hunt: if your helper directs you always to the same place, whether there is an egg or not, he is no helper indeed, he's just a nuisance:

Trivially, a search helped in this way can't be differed from an unassisted search.

But I'm afraid that exactly what W. Dembski and R. Marks do in their Horizontal No Free Lunch Theorem: let's take the partition \tilde{T} = \{\{a1\},\{a2\},...,\{h8\}\} of our Ω1 above. What does

\sum_{T \in \tilde{T}} \alpha_{a1}\{T\} \cdot I_+(\gamma_{a1}|\alpha_{a1})\{T\}

mean? Nothing else than that our seeing, sober man errs 63 times out of 64 when looking for ♕. That just doesn't make sense.

The meaningful expression would have been:

\sum_{T \in \tilde{T}} \alpha_{T}\{T\} \cdot I_+(\gamma_{T}|\alpha_{T})\{T\}

And while the first sum adds up to -∞ the latter one is not negative at all: it's just 6.

nota bene: I think this is the idea behind Rob's (R0b?) comment to my previous post. So, thanks Rob!

larronsicut fur in nocte 19:48, 31 August 2010 (UTC)

Jebus Christ Larron, your grasp of all things mathy-like is overwhelming. Nice to see someone who knows what they're talking about pwning the cretinists. CrundyTalk nerdy to me 18:40, 27 September 2010 (UTC)

The "lead" is awful[edit]

The stuff above the TOC is an incomprehensible mess. Can someone digest the topic into a readable paragraph or something? ħumanUser talk:Human 02:22, 7 March 2014 (UTC)

Agreed. I'll try to fix it up in a little bit. Wehpudicabok [話] [変] [留] 02:23, 7 March 2014 (UTC)
I've taken it out, because I didn't see the point of a detailed account of publishing delays coming before any mention of what the article was about before an explanation of what the damn thing is about or why we should care about it. And that is still totally unclear as the article stands now. TeenageWasteland (talk) 02:37, 7 March 2014 (UTC)
I put a little in summarizing the claims and context of the paper. Thanks to WP for citations. It might be worth putting the stuff about delays back in, but much further down; there was absolutely no need to have it right at the start like that. Wehpudicabok [話] [変] [留] 06:50, 7 March 2014 (UTC)
Thanks for trying to make this article more readable. It was never meant to appeal to a great public (see first lines on this talk-page): I don't think that there are more than a dozen people out there who actually bother to read Dembski's papers in detail.
Nowadays, this page is just of historical interest for RationalWiki, as Dembski threatened to sue us over it (#Dembski unhappy about this page. Threatens to sue.) Frankly, that's the main reason to keep it around :-)
When I wrote the article, I intended the TOC of the page to mirror the TOC of Dembski's paper - that's why so much stuff was pout into the introduction without giving it additional structure - and why Dembski's original abstract appears before the TOC.
--larron (talk) 07:22, 7 March 2014 (UTC)