Wikipedia:Mediation Cabal/Cases/2011-05-08/Power law

Where is the dispute?
This section should explain where the problem is. Link to the articles where the dispute is taking place.

Power law - Estimating the exponent from empirical data - Two points fitting method

Who is involved?

 * User:Structuralgeol (talk) 12:46, 8 May 2011 (UTC)


 * User:Derek farn

Acceptance of Mediation
Please place your signature here to indicate that you are aware of this mediation process and want to participate in it:
 * Structuralgeol (talk) 12:46, 8 May 2011 (UTC)

What is the dispute?
In date 25 apr 2011 15:26, 30 apr 2011 18:15 and 4 May 2011 11:47, user Derek farn removed (three times!) the cited article adducing vague comments and never entering within the content and its accuracy. main link reference: revision history user talk page:

What would you like to change about this?
I retain that the article in subject may be constructive and I whish it can be republished shortly. I've tried to ask to Derek farn some useful suggestion, nevertheless it aswer was vague and as a response he deleted my article.

How do you think we can help?
I hope someone may inform user Derek farn that constructive contributions in Wikipedia should be encouraged. Furthermore, I retain that terms such as “unknown significance” or “ambiguous quality” are inappropriate when dealing with theoretical topics. A theoretical proof can only be correct or incorrect. So user Derek farn should be invited to deal only with article content.

Mediator notes
I'll be more than happy to take on this case. However, we do have to wait and see if Derek is willing to take part in the Mediation. Once they agree, we can get down to business. Regards, MacMedtalk stalk 00:45, 13 June 2011 (UTC)

Discussion
Derek has opened this up with a statement explaining his changes and reasoning. I'd like to invite Structuralgeol to respond; a note has been left at their talk page. MacMedtalk stalk 21:24, 13 June 2011 (UTC)

Statement by Derek farn
Hello MacMed, I hope we can get to the bottom of what appears to be an attempt by one or more people to promote their academic work via Wikipedia. I think the Edit summary associated with the changes I made to the power law article are not vague and say exactly why I felt the changes were necessary:


 * Reference earlier article which is available online
 * Remove excessive promotion of authors own work
 * rv: What appears to be excessive promotion of his own published paper
 * rv: Lots of text relating to a recently published method whose overall significance is as yet unknown

The article on the Poisson distribution also seems to be a target for this self-promotion and I recently removed a reference to an article not cited in the body of the text. 

The Guerriero et al. articles only seem to be available behind paywalls (the authors web site do not contain freely downloadable copies). Derek farn (talk) 01:36, 13 June 2011 (UTC)


 * MacMed what do we do if Structuralgeol does not reply? The person promoting Guerriero et al articles pops up in various guises, today he is promoting himself in the Poisson distribution article. Derek farn (talk) 11:30, 15 June 2011 (UTC)


 * If he doesn't reply there is not much more we can do. If he/she continues to obviously self-promote using multiple accounts, they may be subject to the rules laid out in this policy. Other than that, there is not much more we can do. Regards, MacMedtalk stalk 13:07, 15 June 2011 (UTC)

Rebuttal from Structuralgeol
Dear all, I apologize for delay in my reply but I read just today this discussion. It seems inopportune to talk about a self-promotion of own work. I’ve written the section in subject thinking to give a little constructive contribution to Wikipedia, which would be appreciated by readers. I think that no errors occur in this section and that this latter should be evaluated only on the basis of its content (see  Dispute resolution) and not on the basis of some bias against its writer. I want to reply to the points afforded by Derek farn.


 * Reference earlier article which is available online

R: Erroneous statement: the method illustrated in Shcherbakov et al. (2006) does not estimate a power law distribution exponent.


 * Remove excessive promotion of authors own work

R: Wikipedia is funded on voluntary work of readers/editors and it’s clear that, if someone deals with statistical fracture analysis for study/research reasons, then likely he write contributions involving these topics. Why self-promotion of own work? The section in subject was (before cuttings carried out by Derek farn) constructive and illustrated thoroughly the treated methods. References were obviously due, in agreement with Wikipedia policies. Furthermore, I want to stress that the cited papers are available for free in university libraries and they can be consulted and/or downloaded (for free by university account) by all professors, researchers and students.


 * rv: Lots of text relating to a recently published method whose overall significance is as yet unknown

R: Terms as “unknown significance” are inappropriate when dealing with theoretical topics. A theoretical proof can be only correct or incorrect. Derek farn should deal with article contents only if he is sufficiently competent in statistics to show some mistake/inaccuracy of the adopted models and methods. Otherwise the provided revision are intended as a result of some groundless biases.


 * Section ‘Confidence interval’ in the Poisson distribution article

R: The section in subject is of course pertinent to the related article and provides very useful formulations. The cited references was again due (in agreement with Wikipedia policies) and I was not able to find these methods/formulations in some papers elsewhere.

Concluding I think that, in making a revision, one should take into account only the content validity, otherwise there is the risk of getting poor the Wikipedia, discouraging editors to contribute. With my best regards, Structuralgeol (talk) 11:24, 17 June 2011 (UTC)

Comment from the Mediator
All of the points from Structuralgeol seem fair to me. I accessed the paper through my university online library, and can verify it's existence and legitimacy. It seems like Structuralgeol is simply trying to add to the article constructively. Derek, do you have any other reasons that the article should not be included in Power law? Regards, MacMedtalk stalk 23:05, 22 June 2011 (UTC)
 * MacMed, I have no serious objections to the "Two point fitting method" text as it stands, which is a substantially cut down version of what previously existed; I would object if Structuralgeol wanted to reinstate excessive amounts of text or add additional references to his own work. There is a huge literature involving power laws (this site is a bibliography of just one kind of power law) and having researchers add a multi-paragraph+picture subsection every time they published a paper involving a power law would quickly dilute the quality of Wikipedia articles.  I have tried (not very hard) to find a source to cite that is publicly available and would encourage Structuralgeol to make a copy of his papers generally available to everybody. Derek farn (talk) 00:05, 23 June 2011 (UTC)

Dear MacMed I invite you to read the full version of the section “Two point fitting method” before cuttings carried out by Derek farn on 25 May 2011 22:37 and then to provide your opinion. This section and the included figures explained thoroughly the method and formulas to the reader (I've spent a lot of time to write it..), referring to bibliography only for details about theoretical derivation of the illustrated equations. Cited references were both due, as: (i) Guerriero et al. 2011 shows the derivation of the first illustrated equation (calculating the confidence interval for exponent of power law distribution) and (ii) Guerriero et al. 2009 illustrates the derivation of the second shown equation (calculating the confidence interval of measured cumulative frequencies). I think this is not an excessive amount of text or literature. The article in its present form (i.e. after abundant cuttings provided by Derek farn) is not at all thorough and it obliges the reader to seek references (one of whose has been removed) from literature, in order to understand something. Substantially the article in the present form says that a useful statistical method for parameter estimation exists, but it does not provide any information about this latter. I retain that this is not in the aims of Wikipedia. I don’t understand why Derek farn talks about “dilute the quality of Wikipedia articles”. I think that researcher’s work should be welcomed by Wikipedia, provided that this is (i) pertinent, (ii) useful and (iii) correct. Why to limit the editor’s work? Finally I want to stress that scientific papers are generally not freely available on web pages (for publisher copyright reasons) but they are accessible by appropriate channels (university libraries, university account etc.). An alternative channel to access to scientific topics is provided right by web sites such as Wikipedia and similar, in which paper content is exemplified to readers.

Why Derek farn delete this article from Wikipedia and after he pretends that someone makes its content available elsewhere on the web? With regards, Structuralgeol (talk) 00:05, 30 June 2011 (UTC)

Suggestion
I think Structuralgeol makes a fair point here. Derek, is it fair to understand that it is not really feasible to provide one's research for free? Also, would you be opposed to some explanation of the method laid out in SG's paper being added to the article?

And to Structuralgeol, can you draft an example and put in on this page (perhaps in a collapsible template) so that Derek and I can take a look and make sure that it covers what will aid the reader's understanding without going overboard? We can take a look at SG's draft and suggest changes that are necessary, before editing the actual article. Does that sound acceptable to the both of you? <b style="color:green;">MacMed</b><sup style="color:red;">talk <sub style="color:black;">stalk 21:19, 3 July 2011 (UTC)
 * MacMed, if Structuralgeol wants to spend time writing on Wikipedia about his research the appropriate action is for him to create new articles discussing the topic of his work. The power law article involves a wide range of topics and inserting a comparatively large amount of material on one topic that is not central to the subject is abuse. Derek farn (talk) 13:28, 5 July 2011 (UTC)
 * Derek, I'd just like to give you a little reminder about ownership here. Let Structuralgeol write a draft with a bit of explanation for the method, and if he decides that he cannot explain the method in such a way that it does not overtake the power law article, then we'll move from there. But give him a chance first. <b style="color:green;">MacMed</b><sup style="color:red;">talk <sub style="color:black;">stalk 16:29, 5 July 2011 (UTC)
 * Dear MacMed, once again Derek farn provides vague (and groundless) arguments in order to hinder publication of the article in subject (I don’t know why…). He started claiming a conspiracy aimed to promote some scientific work, then asserting that references were erroneous and/or redundant etc. After he says that cited references are not available. Once assessed that these arguments were unjustified he stated that the article (in its earlier form) is too long. Now he says that may be appropriate to put the it as a stand-alone article. With regard to this mediation process I’m quite pessimistic, as I think that  Derek farn has already decided that this article has not to be issued on Wikipedia. However I want to submit to your attention a draft of the article illustrated  below. Before I want to reply to  Derek farn. In whatever article or book chapter dealing with a probability distribution, the topic concerning estimation of parameter and its confidence interval are needed. The two point fitting method provides an alternative criterion (showing some advantages and disadvantages) to that of maximum likelihood and Kolmogorov–Smirnov’s. Therefore I retain that this section is (i) pertinent and (ii) useful within the Power law distribution article. Below the draft of section “Two Point Fitting Method”:
 * Structuralgeol, the text you have produced would make a good article in its own right and I suggest that you create an article with this content. However, such a large quantity of material is completely out of keeping with the importance of the subject in an article about power laws.  I am not a lawyer but I believe a conspiracy requires more than one person and as you are the only person promoting these two articles there can be no conspiracy. Derek farn (talk) 00:04, 8 July 2011 (UTC)

Two point fitting method
This criterion can be applied for estimation of power law exponent in the case of scale free distributions and provides a more convergent estimate than maximum likelihood method. The method is described in Guerriero et al. (2011) where it has been applied to study probability distribution of fracture aperture. In some contexts the probability distribution is described, rather than by cumulative distribution, by the cumulative frequency of a property X, defined as the number of elements per meter (or area unit, second etc.) for which it results X > x, where x is a variable real number. As an example, the cumulative distribution of fracture aperture X for a sample of N elements is defined as 'the number of fractures per meter having aperture greater than x'. Use of cumulative frequency has some advantages, e.g. it allows to put on the same diagram data gathered from sample lines of different lengths at different scales (e.g. from outcrop and from microscope). The exponent estimation is based, rather than a least squares fitting line, on the definition of a straight line passing for two points only, i.e. one for each – outcrop and micro scale – data set (see figure). These points are chosen in such a way as to obtain the maximum number of sampled fractures for each data set, these being affected by the minimum truncation/undersampling error (i.e. that occurring at the lower scale, where the image resolution limit may cause underestimation of the number of detected fractures). Monte Carlo simulations have proved that this criterion provides a more efficient estimator than least squares and maximum likelihood methods (Guerriero et al. 2011), which is unbiased under the hypothesis of random spatial distribution of fractures. The two point fitting method provides also a ready criterion to calculate the confidence interval of the power-law exponent. Indeed, the real values of fracture density, for the two points, fall in the respective 95% confidence intervals (see figure below). Furthermore, also the real power law (represented by a line in a bi-logarithmic diagram) falls within one of these confidence intervals, with the same probability. The joint probability that the line falls within the two intervals simultaneously is equal to the product of the probabilities that the line passes through each interval, being the two data set stochastically independent. Therefore, the final probability is equal to $$0.95^2$$, i.e. ca. 0.9. By indicating with $$F_{low1}$$ and $$F_{pp1}$$ the confidence interval limits of the outcrop-scale scan line, $$F_{low2}$$ and $$F_{upp2}$$ those relative to the micro-scan line, s1 and s2 the abscissas of the two points, the following equations for the 90% confidence interval of the exponent m have been obtained:
 * $$m_{upp}=ln(F_{low1}/ F_{upp2})/ln(s_{1}/s_{2})$$


 * $$m_{low}=ln(F_{pp1}/ F_{low2})/ln(s_{1}/s_{2})$$

where $$F_{low}$$ and $$F_{upp}$$ are given, for each data set, by the following expressions (Guerriero et al., 2009), under the hypothesis that each cumulative frequency value follows a Poisson distribution in the sampling domain (i.e. random spatial distribution of fractures):
 * $$ F_{low}=N/L (1-1.96/sqrt(N-1)) $$


 * $$ F_{upp}=N/L (1+1.96/sqrt(N-1)) $$

where N is the number of detected fractures and L the length of the sample line.

Response
Here I am inclined to agree with Derek. This goes a little bit more in depth than is befitting an article about such a general thing as power law. Unfortunately, I also don't think this could be its own article, considering the notability policy requires substantial third-party coverage. Structuralgeol, do you think you could condense this even more, giving just a brief overview of the method, preferably integrated in the text rather than its own section? Regards, <b style="color:green;">MacMed</b><sup style="color:red;">talk <sub style="color:black;">stalk 02:01, 8 July 2011 (UTC)
 * MacMed, the above text is essentially what Structuralgeol originally added to the power law article (under another name) and I cut down to the current wording in that article. Derek farn (talk) 12:26, 8 July 2011 (UTC)
 * Dear all, below there is a condensed (draft) version of the section originally added to the power article (why under another name?..). I think that it would be opportune to place this topic in a proper subsection in order to distinguish it from other methods such as maximum likelihood and Kolmogorov-Smirnov’s, which are illustrated in other sub-sections of sect. 4.3. This condensed version is organized in a very short text associated with a picture explaining the adopted criterion. I retain that use of the figure is suitable as we deal here with a substantially graphical method. I’m confident that you will consider appropriate this section within the Power law article. Regards, Structuralgeol (talk) 23:14, 11 July 2011 (UTC).
 * Structuralgeol, I think the wording in the current article that cites your paper is good wording and better than your suggested change. Are you unhappy with the current wording covering the two-point method? Derek farn (talk) 23:56, 11 July 2011 (UTC)
 * Derek, I think the below draft covers the method to Structuralgeol's satisfaction, and would not take away from the rest of the article. Do you object to having the below placed in the article and wrapping up this mediation? Regards, <b style="color:green;">MacMed</b><sup style="color:red;">talk <sub style="color:black;">stalk 02:28, 12 July 2011 (UTC)
 * MacMed, the article already contains a subsection covering the two point method. Are you suggesting that here be two subsections covering this topic?  As I said, the current wording is better than the suggested new wording. Derek farn (talk) 10:29, 12 July 2011 (UTC)
 * No, the section below would have replaced the old one, but I am inclined to agree with you. Would either of you object to simply inserting the image Structuralgeol proposed next to the existing text? <b style="color:green;">MacMed</b><sup style="color:red;">talk <sub style="color:black;">stalk 12:11, 12 July 2011 (UTC)
 * MacMed, the image does not add anything to the article, it is just noise. Some dots with a straight line through them and some pictures of what I take to be some rock formation has no meaning in this context. Derek farn (talk) 16:43, 12 July 2011 (UTC)
 * Taken with the text below, it adds understanding to the article. I am no expert in mathematics, and that image helped me understand this two point method more than the text did. <b style="color:green;">MacMed</b><sup style="color:red;">talk <sub style="color:black;">stalk 17:52, 12 July 2011 (UTC)
 * Dear all, I read joust now your recent comments. I reply here. The current version of this section within the "Power law" article is not adequate as it consists of a cut down of a more extended article. As a result most of the remaining text is spent to describe inessential topics, such as use of cumulative frequency rather than cumulative distribution. In the new short version proposed below this latter topic is referred to the “Cumulative frequency analysis” article. Here the condensed text (about half of the current version) only reports the aims/applications of the estimation method and refers to appropriate bibliography for more information, while the picture describes quickly the criterion. I think that the section proposed below is notably clearer than current version and it is sufficiently contained as you’ve previously suggested/required. Regards, Structuralgeol (talk) 19:18, 12 July 2011 (UTC).
 * Structuralgeol, the changes you suggest are very jumbled and uninformative. The text seems to have been written with a view to maximizing the number of times "Guerriero et al" is mentioned.  I repeat what I have been saying all along, your aim seems to be primarily to promote your own work. Derek farn (talk) 22:25, 12 July 2011 (UTC)
 * Ok, the number of citations can be reduced as below. Structuralgeol (talk) 23:24, 12 July 2011 (UTC).
 * Structuralgeol, how about making a change that means that only one of your papers is cited? Readers are not going to understand what this topic has a figure containing pictures of rock formations, why not redo the figure to explain the mathematics rather than promoting your research? Derek farn (talk) 11:49, 13 July 2011 (UTC)

Derek farn, the picture simply shows an example of application of this method of multi-scale analysis for power law distribution. I want to explain you this concept by a simple example: if you consult a basic statistics textbook, you may find some examples such as e.g. the distribution of weight or height of students, aimed to explain the Normal distribution. This does not mean that this textbook wants to afford the problem of student diet. The weight distribution represents an example of a character normally distributed. Analogously, the fracture aperture distribution is an example of a character distributed according to a power law. I cannot agree (and I think no one can…) with your comments. You describe the diagram within picture as “…Some dots with a straight line through them…”. Therefore, analogously, you may describe (I make an example involving a famous name…) Bernoulli’s equation as “some letters with an equal sign through them”. You claim that the overview below is “very jumbled and uninformative”. I want to point out that the current wording (proposed by you) is the result of an undifferentiated cutting. You only delete the text exceeding first few lines without a minimum of synthesis. It’s clear that the cuttings to the article, operated by you, are the contribution of someone that does not understand what he is writing. Frankly speaking I think that you are not enough skilled in statistical analysis to comment this chapter. Dear MacMed, I retain that this mediation procedure has failed. I’ve tried over and over to reach a compromise, nevertheless Derek farn seems absolutely reticent. As you can see, he even deny the obviousness in order to hinder my editing as well as this negotiation process. Furthermore, it is sufficient to give a look to Derek farn (talk) talk page (and I invite you to do it) for understand that its conduit is challenging and disrespectful of other people’s work. FIVE YEARS of disputes with many many editors!!! I retain that you may declare as failed this mediation, so I can proceed with formal procedures (e.g. by requesting some action to the Arbitration Committee or to the Wikipedia Community or others). Thanks very much for your kind cooperation. Best regards, Structuralgeol (talk) 10:28, 14 July 2011 (UTC).

Two point fitting method (condensed version)
This criterion can be applied for estimation of power law exponent in the case of scale free distributions and provides a more convergent estimator than maximum likelihood method (as proved by Monte Carlo simulation in Guerriero et al. 2011). The method is described in this latter reference, where it has been applied to study cumulative frequency of fracture aperture in rock. The confidence interval is calculated taking into account that the “true” line is contained within each of the two confidence intervals in fig. (readily calculated by equations provided in Guerriero et al., 2009) with a probability of 0.95 and so, within both, of $$ 0.95^2 $$, i.e. ca. 0.9.

Disrespecting MacMed
Structuralgeol it is very disrespectful of you ask the editor User_talk:AlbertCahalan to intervene in this dispute when MacMed is dealing with it. Derek farn (talk) 19:01, 15 July 2011 (UTC)

To provide a complete record I append you message to AlbertCahalan here:

"Dear Wikipedist editor, I want to submit to your attention an our common problem: disruptive contributions and edit warring operated by user Derek farn (talk). This latter shows systematically a provoking behaviour and lacking of respect for other people’s work, typical of vandalism. I’ve sent this communication to many people having the same problem in order to organize a collective protest/action request directed to e.g. the Arbitration Committee or Requests for comment/User conduct (this latter procedure requires the participation of at least two users) or to the Wikipedia Community. If you agree with this initiative please contact me at this dedicated email address: clipeaster-1971 AT yahoo DOT com. In order to avoid creating of a forum section dedicated to Derek farn I suggest you to delete this communication once you’ve read it and, then, be in contact via email. Any suggestion are welcomed. I look forward to hearing from you. Best regards, Structuralgeol (talk) 18:15, 15 July 2011 (UTC)."
 * Why Disrespecting MacMed? I’ve already given my thanks to him, telling him that I retain this dispute as failed. I’ve asked to User_talk:AlbertCahalan to participate to a Requests for comment on user conduct, as Wikipedia rules require the participation of at least two users (and it’s evident that it will not be hard to find them…). This procedure is different from Mediation Cabal and here is not contested the Power law article but your conduct, which is unfair and in disagreement with the Wikiquette. Structuralgeol (talk) 19:42, 15 July 2011 (UTC).
 * Structuralgeol, it is up to MacMed to decide the status of this mediation. When you say this mediation has failed I take it you mean that your attempt to use Wikipedia to gain excessive publicity for your own academic research has failed.  Looking at Special:Contributions/Structuralgeol I see you are now investing a lot of time fishing around to find somebody who will be sympathetic to your point of view.  Derek farn (talk) 15:07, 16 July 2011 (UTC)

One more comment
Okay guys, I am not offended at all and I do not feel that this is a sign of disrespect to me at all. If the two of you agree that mediation has failed and you would like to move it to the next stage (the Mediation Committee for content or requests for comment about user conduct), then feel free to do so. However, I think that we were approaching a solution above. Would either of you be opposed to leaving the section as it is, but adding in an image. The new image being one created by Structuralgeol, but with no reference to his research, simply theoretical mathematics. If we can agree on that, then we can change the article, call this a success and move on. If you do choose to pursue other avenues of mediation, then that is your decision. Please respond below with your opinion on my suggestion and the ideal way to proceed from your point of view. Regards, <b style="color:green;">MacMed</b><sup style="color:red;">talk <sub style="color:black;">stalk 23:14, 15 July 2011 (UTC)
 * Hi MacMed, I appreciate your patience in managing this dispute. I retain that the article in its present form is not well written nor thorough, nevertheless if this is the only way to close this tiring mediation I agree with your proposal of adding the picture to the article. Furthermore I suggest to replace the current wording with the version below and, if the main problem of Derek farn is the second citation, this latter could be removed. With my best regards, Structuralgeol (talk) 18:38, 23 July 2011 (UTC).
 * MacMed, I think we both agree that the current wording is much better than the proposed changed. Figures are always good when they help explain the subject, but readers are not going to understand what pictures of rock formations have to do with the power laws.  If Structuralgeol wants to add a figure he should create one that explains the mathematics rather than promotes his work. Derek farn (talk) 13:05, 25 July 2011 (UTC)