User talk:Stevenj/Archive 4

Laplace transform
Hi Steven, I could use some pearls of wisdom on how to introduce the Laplace transform. See Talk:Laplace transform, where there is some question over the validity of thinking of the transform in terms of moments. The literature is fairly silent on a decent way to interpret the transform that can be summarized in plain words suitable for the lead of the article. Your input would be most appreciated. Thanks, Sławomir Biały  (talk) 12:58, 23 July 2010 (UTC)


 * I have to admit that I don't really use Laplace transforms except in teaching where the syllabus requires it; I have never personally found them useful for any real application. (The historical application of Laplace tables to solving certain ODEs has been obsoleted by computer-algebra systems.)  Because they don't have a useful inverse transform (Bromwich and Post are not useful), it is hard to interpret Laplace transforms in a Fourier-like way as a change of basis (you need an inverse transform to write down the basis functions explicitly).  And because inverting them numerically is hard, although some algorithms have been proposed, they aren't terribly useful as a computational technique either.


 * The only area in which I've heard mutterings that Laplace remains truly useful is in control theory, so you might want to talk to someone in that area. — Steven G. Johnson (talk) 23:30, 23 July 2010 (UTC)

Any chance they'd be useful in financial mathematics? And if so, would you know about it? I think I've got something somewhere around here written by Gian-Carlo Rota that says there's some reason why Laplace transforms should be paid attention to. I'll see if I can find it.

If I recall correctly, the theorem that says that the two-sided Laplace transform is one-to-one is a fairly substantial result whose proof is not trivial. Here's corollary in theoretical statistics: Suppose
 * $$ X_1, \dots, X_n \, $$

are independent random variables and each has a normal distribution with mean &mu; and variance &sigma;2. Consider functions
 * $$ f(X_1, \dots, X_n) \, $$

that enjoy the following properties: the value of &fnof; does not depend on &mu; or &sigma;2 (if you like, the value of &fnof; depends on (X1, ..., Xn, &mu;, &sigma;2) only through (X1, ..., Xn)); and regardless of the values of &mu; and &sigma;2 we have
 * $$ E(f(X_1,\dots,X_n)) = \mu. \, $$

Show that among all such functions, the one with the smallest variance is the sample mean
 * $$ (X_1 + \cdots + X_n)/n. \, $$

("All such functions" means we're not considering just weighted averages of the Xs or the like, e.g. the average of the max and the min of the Xs is one of many that have the desired properties; the median of the Xs is another.)

I'll see if I can find that thing by Rota. Michael Hardy (talk) 00:12, 24 July 2010 (UTC)


 * I'm not saying that there isn't some beautiful and non-trivial mathematics associated with Laplace transforms. I just see them as something with extremely limited modern practical utility in my own experience.  Since I make no claim at expertise in all fields, I cannot say for certain whether there is some compelling application in other fields that I'm not familiar with. I've tried to look for such applications in the past, however, and most of the claimed applications seemed tenuous to me; YMMV.


 * Rota taught MIT's ODE course for a long time, which to this day includes Laplace transforms. When I've asked around the department to see why they think Laplace transforms are important to include in the syllabus, the reply I get is twofold.  First, it is seen as a pedagogical step leading conceptually towards Fourier transforms but without complex numbers or distributions.   Second, because the engineering department wants us to teach it, and in particular the control-theory people demand it.


 * (You can also use Laplace for things like impedance in circuits, but Fourier works for that as well, and all of the EE people I know think of complex impedance in terms of Fourier rather than Laplace. When I took EE courses as an undergraduate, I recall a cursory pass over Laplace as a prelude to everything subsequently being Fourier.  In practice, every real signal is a tempered distribution to which Fourier is applicable.  And for merely mechanically solving initial-value problems there are now much more powerful automated tools than resorting to tables of Laplace transforms.)


 * (I am not one to ask about finance applications. A quick literature search finds a few papers using Laplace methods to numerically solve Black-Scholes PDEs.  However, for that matter you can also find a few papers employing Laplace methods to solve other PDE initial value problems, and yet to cite this as a major application would be highly misleading: because Laplace transforms are so hard to invert numerically, they are basically a fringe method for PDEs that are occasionally attempted but haven't really caught on, and other time-stepping methods are far more widespread.  Out of the thousands and thousands of papers on numerically integrating PDEs, I could only find a handful that used numerical Laplace transforms.  Since financial modelling isn't my field and the terminology is somewhat different, I don't know whether the same is true there.)  — Steven G. Johnson (talk) 02:15, 24 July 2010 (UTC)

I was really just guessing about financial applications. What I had in mind is this: interest rates are in units of frequency, i.e. 1/time. So what is the present value of a revenue stream as a function of the interest rate (assuming continuous compounding)? In effect, that's how much you need in the bank in order to make withdrawals (possibly continuously) according to a prescribed schedule forever. Admittedly, when the question is put that way, it seems unrealistic since interests rates fluctuate. But that function of the interest rate is the Laplace transform of the rate of withdrawal as a function of time. OK, just a toy problem. Michael Hardy (talk) 04:04, 24 July 2010 (UTC)


 * To be fair to Laplace, I suppose I should give it some credit for cases where the Fourier transform is analytically continued in the complex-frequency plane, which is common when one is talking about causality, stability, and so on. And in physics, it is common to talk about Wick rotation to purely imaginary frequencies (Matsubara frequencies in thermodynamics), and I suppose that could be thought of as Laplace transforms.  Most people in physics think of this as Fourier transforms that have been continued, however.   The nice thing about these applications is that one never needs to invert the "Laplace" transformation. — Steven G. Johnson (talk) 12:26, 24 July 2010 (UTC)

Thanks for the replies Steven and Michael. (Is this the MIT mathematics department or something?) My concern has been that many secondary sources do treat the Laplace transform as either an analytically continued Fourier transform or as a purely operational gadget. I don't find this really satisfactory for explaining what it is in the lead of an article, however. On the other hand, those sources that do devote some time addressing what a Laplace transform "is" generally seem to be too sophisticated to distill into something easily communicated. Sławomir Biały (talk) 10:28, 26 July 2010 (UTC)

Generating Trig Tables
Hello Steven

I think this article should at least reference Hipparchus, Ptolemy and later Regiomontanus and Copernicus. All of whom generated quite comprehensive trig tables using (as far as I can make out) Ptolemy's theorem (that's a misnomer if Hipparchus used it in the first place). I have indicated some of the basics of this method on the page on Ptolemy's_Theorem.

It would also be interesting to know what method was employed by early Chinese mathematicians such as Guo Shoujing. Did the Greek method find it's way to China via Islamic traders? Alternatively did Chinese mathematicians develop trig tables independently? The sudden appearance (in Europe) of trig tables using Arabic numerals is also something of a mystery if Europeans were using Greek sources.

Neil Parker (talk) 18:39, 2 August 2010 (UTC)


 * If I recall, Boyer argued that Hipparchus and Ptolemy used the half-angle and angle-addition formulas to generate the tables. I don't recall seeing a description of the methods used to generate later tables.  In any case, my understanding is that it is well documented that the Europeans trigonometry was heavily influenced by Arabic sources, which in turn built upon Greek and Hindu sources (although the Arabic authors added many contributions of their own), so there is not much mystery about trig tables using Arabic numerals.  (The word sine itself comes from a translation of Arabic.)  See History of trigonometry.


 * The Generating trigonometric tables article needs to be enhanced in many ways. It should mention more of the history (referring to other articles for details where appropriate), but the description of the methods also leaves much to be desired. — Steven G. Johnson (talk) 23:16, 2 August 2010 (UTC)

Given the importance of trig, the history thereof is - in general - somewhat sketchy. Have written a brief comment about that on the article's talk page.

Neil Parker (talk) 10:22, 3 August 2010 (UTC)

Reciprocity theorem
I think there is a problem with the redirect of Reciprocity theorem to Reciprocity (mathematics). For one thing Reciprocity Theorem redirects to somewhere else altogether. What I had in mind was writing an article on the simple version of electrical reciprocity theorem as commonly found in linear network analysis books. The reciprocity (electromagnetism) article is too deep for those who are just looking for two-port network reciprocity. I had in mind writing this at reciprocity theorem but clearly that would be unacceptabale now. Have you any ideas for an article title? Also reciprocity (mathematics) is an unsuitable title for the dab page if it is going to include electrical science entries, as it must if reciprocity theorem redirects there.  Sp in ni ng  Spark  19:46, 14 August 2010 (UTC)


 * I fixed the Reciprocity Theorem redirect. The point is that there are so many "reciprocity" theorems, it only makes sense to redirect to Reciprocity (mathematics) and let the reader pick which one she is interested in.


 * I think it is perfectly reasonable to have an article Reciprocity (electrical networks) or something like that, which is specifically on electromagnetic reciprocity applied to lumped-circuit models...although you need to name it carefully to distinguish it from Tellegen's theorem. I agree that the full Lorentz reciprocity theorem (of which the electrical-network application is a special case) is probably too much for readers just doing lumped circuit models (who aren't used to dealing with the full Maxwell equations).  (The electrical networks article could refer to the Reciprocity (electromagnetism) for the complete proof.)


 * Regarding the name of the dab page, perhaps it could be better named, although of course the electromagnetic theorem is a mathematical theorem as well. I would take it up on the Talk page for the dab article, though. — Steven G. Johnson (talk) 20:23, 14 August 2010 (UTC)


 * I've made a Reciprocity theorem (disambiguation) page. — Steven G. Johnson (talk) 22:34, 14 August 2010 (UTC)

Old St Paul's Cathedral
Hello, I hope you don't mind, but I've turned your picture of the 1908 cathedral model into a sortof "photographic artist's impression". What do you think? (I wouldn't recommend zooming into it, as the illusion slightly falls apart at full size.) Bob talk 22:33, 8 September 2010 (UTC)


 * I can't speak to the historical accuracy, but it looks beautiful! Thanks for letting me know!  — Steven G. Johnson (talk) 22:42, 8 September 2010 (UTC)

Plancherel and Parseval
Hi StevenJ - Am I missing something? Could you check the recent edits on Fourier transform in which an editor states that the Plancherel and Parseval theorems are equivalent? This seems trivially untrue. PAR (talk) 20:37, 22 September 2010 (UTC)


 * It's true. The abstract form of Parseval is that &lang;Fu,Fu&rang;=&lang;u,u&rang; for all u (the Fourier transform is an isometry), while Plancherel is that &lang;Fu,Fv&rang;=&lang;u,v&rang; for all u and v (inner products are preserved), where &lang;&sdot;,&sdot;&rang; is the usual inner product of functions and F is the Fourier transformation operator.  Plancherel obviously implies Parseval. The converse is less obvious, but is also true and in fact the two statements are equivalent for any operator on a Hilbert space.  For a proof, see for example section 10.3 ("Unitary Operators") of the book Basic Classes of Linear Operators by Gohberg et al.  For your information, I paraphrase it here:
 * Suppose &lang;Fu,Fu&rang;=&lang;u,u&rang; for all u. Then we first prove that F*F = I (identity), where F* is the adjoint (the operator such that &lang;F*u,v&rang;=&lang;u,Fv&rang; for all u,v).  Proof: &lang;(F*F-I)u,u&rang;=&lang;Fu,Fu&rang;-&lang;u,u&rang;=0.   Now, the desired result is obvious: &lang;Fu,Fv&rang;=&lang;F*Fu,v&rang;=&lang;u,v&rang;.  Q.E.D.
 * Probably, our unitary operator article should be updated to include the three equivalent properties of these operators U (isometry, preserving inner products, and U*U=I), citing e.g. Gohberg. — Steven G. Johnson (talk) 20:53, 22 September 2010 (UTC)
 * Thanks, Steven, for the explanation. I believe this means that the equivalent statement needs to be put into Discrete Fourier transform as well. It does not seem to be an issue with Fourier series and Discrete-time Fourier transform, but it might be useful to make that point for them as well. I would feel more comfortable if you make the changes, if you feel they are needed. PAR (talk) 01:38, 23 September 2010 (UTC)
 * You're right, there is an equivalent statement for the DFT, and in fact it is even clearer for finite matrices. The key point is that the following three properties are equivalent (and hence any of them can be used as the definition of the "unitary" property):
 * $$\Vert F u \Vert = \Vert u \Vert$$ for all u (preserving length), $$F^* = F^{-1}$$ (the traditional "unitary" definition), and $$\langle u, F v \rangle = \langle u, v\rangle$$ for all u,v (preserving dot products).
 * A general reference is Gohberg, but it would be nice to find a more accessible reference for the matrix case (since Gohberg handles the general case including infinite-dimensional operators and hence is quite sophisticated). None of the elementary linear algebra books on my shelf prove this fact, however; I'll ask around.
 * — Steven G. Johnson (talk) 21:33, 23 September 2010 (UTC)

T-integration
I created this article years ago as part of a project to include in Wikipedia an article on every topic that had an article on Wolfram Mathworld. Rick Norwood (talk) 13:19, 17 February 2011 (UTC)


 * Thanks for the info. Unfortunately, in this case Weisstein seems to have been persuaded to include a topic in Mathworld that is of dubious notability.  The Mathworld article itself is listed as being written in part by Smith, so I'm guessing Smith was the one who persuaded Weisstein to include the topic and Weisstein never checked it more closely.  — Steven G. Johnson (talk) 21:17, 17 February 2011 (UTC)

Differences in the Cooley-Tukey Methods within Mathematics Tools
Hi Steven. I have a few questions on the FFT and was hoping you could point me down the correct road. The definition presented by you on the Cooley-Tukey FFT is consistant with the definition in my undergraduate college textbook (Lathi), but both definitions seem to be inconsistant with how the mathamatics tools I use implement them, even though they all claim to be using the Cooley-Tukey algorithm. For example, MATHCAD uses the 1/N factor on the FFT, but not on the IFFT. It also does not produce the first value since the summation starts from 1 instead of from 0. On the other hand, MATLAB produces the first value and uses a factor of 1 on the FFT and the 1/N factor on the IFFT like here, but it uses a positive sign in the exponent in the exponential of the FFT rather than a negative sign, and the negative sign is used in the IFFT. So it is reversed. My questions are these: Are these variations simply wrong, or are they as correct as what is presented here and just dependant upon how they were derived? It seems that the Cooley-Tukey Method should only be one way. Am I wrong? Also, what is the impact of using these different tools and/or methods? I would suspect that cycling through the MATLAB version (i.e., time domain sampled function --> transform --> time domain sampled function) should give the same result as the version presented here. But what about the transforms while in the frequency domain? Is it comparing apples-to-apples? And then for the MATHCAD version, I suspect that this one would have to have some additional operations on it to yeild the same results as the other two, like adding back in the first term and changing the 1/N weighting by multiplying thorugh by N. Am I wrong in assuming the results would be very different? Why do they define it so differently across the board? I would think that a company selling a toolbox like this would want to stay consistant with what the math/engineering community expects. Does anyone have any insight into this? Thanks in advance. J.G. Lindley (talk) 02:16, 23 March 2011 (UTC)


 * Where you put the 1/N factor is simply a normalization convention, and differs between treatments. There is no single right answer, as long as you are internally consistent. You just have to keep in mind that the product of the normalizations for forward and backward transforms have to be 1/N if you want them to be inverses.  (Some FFT programs don't put in any normalization at all, and expect the user to include 1/N as desired.)


 * To be a DFT, you have N outputs for N inputs; you can't "skip" any outputs. However, some programs (e.g. Matlab) use 1-based indexing rather than 0-based indexing -- they are producing the same outputs (including the 0-th/DC output), just are indexing the arrays differently.  This is just a labelling convention and there are no right or wrong answers here, as long as you are internally consistent.


 * Also you should realize that your questions aren't really about the Cooley-Tukey algorithm or any FFT algorithm for that matter; you are asking about the mathematical definition of the discrete Fourier transform (DFT). The DFT is a transformation, and FFTs are merely algorithms to compute the transformation.  Any slight change in the conventions used for the DFT (normalization, indexing, sign conventions) can easily be incorporated as changes in the corresponding FFT algorithms. — Steven G. Johnson (talk) 16:25, 23 March 2011 (UTC)


 * Thanks, Steven, for your response. I do realize that the +/- reversals within the exponentials are simply a 180 degree phase reversal, so those did not bother me nearly as much as the differences in the weighting, or what I have perceived as a dropping of the first term. I'll go back and redo my math to make sure I can find the first term in Mathcad's definition. It seems when I went through the math the first time, it worked out that the first term of the summation was left out (the one where e^0 = 1 and you are left with just it's coefficient). Also, thanks for explaining that the 1/N is a normalization. I did not perceive it that way at first. J.G. Lindley (talk) 15:27, 25 March 2011 (UTC)

Regarding Mitzi Szereto
Hi Steven,

I blanked your comments on the talk page of this article based on a complaint and denial from the subject. I recommend in the future that we generally refrain from speculating about "self-promotion" as in many cases, misbehavior may be undertaken by a fan or family member rather than by the subject of the biography, and feelings can be hurt if such accusations are false. Additionally, speculating on who is behind promotional edits doesn't usually add much to simply noting that they are promotional edits.

Having said that, I ran a checkuser and your blocks on User:SindySays and User:81.98.138.108 were valid. Additionally, based on the checkuser results, I recommend that you extend the block to User:Writerlassuk. I would do it myself, but as you may know, I customarily refrain from using the block tool.

Your work on this page is good, and it was good of you to identify the problematic edits. Please don't take my blanking of your comments and my advice to refrain from speculation as a reprimand.--Jimbo Wales (talk) 13:54, 7 April 2011 (UTC)


 * Thanks Jimbo, this seems sensible. — Steven G. Johnson (talk) 20:55, 7 April 2011 (UTC)

Sliding DFT
Davidmeo (talk) 15:20, 22 April 2011 (UTC)Hello Steven, thank you for pointing me to the Sliding DFT - when was this method first published? This is regarding the Standing Wave Discrete Fourier Transform published in a 1990-- I'd like to review the original if it is posted also. If you're correct and they are the same (not sure I agree, perhaps portions are the same), the later publication should reference the earlier one - I would think in 1990 my UCSD professors would have pointed me in SWDFT to reference the Sliding DFT - it looks like Sliding DFT mentioned Wikipedia published 2003, is there an earlier publication?


 * I don't know for sure when the first publication of sliding DFTs was, although the 2003 article doesn't cite any earlier algorithm of that name if I recall correctly. (Update 5/2/2011: The article does indeed cite a 1975 book by Rabiner and Gold.) The basic ideas seem the same as your SWDFT: a circular buffer, adding in the effect of new samples and subtracting the effect of old ones.  The biggest difference was that your pseudocode didn't compensate for the overall time shift and so differs from a DFT of each window by a linear phase, but this is fairly minor — the difference is one parentheses.
 * You are certainly right that an earlier publication has a certain priority, and you are free to make this argument in appropriate venues. However, Wikipedia does not referee priority disputes, and by policy we describe things according to the most common name.  As far as I can tell, the "sliding DFT" terminology is nowadays extremely widespread, whereas I can't find any usage of SWDFT outside of your papers, so Wikipedia would have to go with that terminology and that formulation of the algorithm.  When we eventually do get a sliding DFT article, however, I'll try to look more into the history and cite your paper if it seems to be the first.  — Steven G. Johnson (talk) 17:46, 22 April 2011 (UTC)

Davidmeo (talk) 05:27, 23 April 2011 (UTC)Hello Steven, I left comment on my main page asking to remove the redirect Short Time Fourier Transform. It looks like due to the conflict of interest policy the article shouldn't be posted on Wikipedia also.

I don't think the SWDFT is the same as STFT. While I was a student a UCSD, Dr Alan Schneider the main professor who signed my thesis and I met with Dr Andrew Viterbi (from Qualcomm) to review the SWDFT. At the time, Dr Viterbi shared an office at UCSD with Jack Wolf (my Information Theory professor). Dr Laurence Milstein (my Communications professor) was one of my professors who signed the Thesis. No one mentioned the Sliding DFT or STFT, had they mentioned this I would have reviewed and provide references as appropriate.

While I haven't reviewed this closely enough, I believe the Standing Wave Discrete Fourier Transform (SWDFT) is not the same as the sliding DFT. The Sliding DFT (referenced in the STFT article) uses the circular phase shift property of the DFT to achieve the same order of computational efficiency as does the SWDFT which uses quadrature harmonic demodulators. Some differences to the Sliding DFT are that the SWDFT can use heterodyne Low-Pass filters to minimize spectral leakage and still identify multiplication redundancy for fewer multiplies, has a harmonic weighted memory method with no stability issues, has constant phase mapping, and the SWDFT structure can be modified to tune the quadrature heterodyne demodulators to address spectral leakage without stability issues for certain signals. All these methods are just different variations of Fourier analysis. David

Davidmeo (talk) 06:22, 2 May 2011 (UTC) Hello Steven G Johnson,

In reply to your comment about WP Spam, I reviewed an automated message in my Standing Wave Discrete Fourier Transform article which stated the article is an orphan because there are no Wikipedia links, and that automated message advised me to add links from other Wikipedia articles - that's what I did. I have a deeper concern as I intend to protect my original research, and have two questions for you: 1) Is the author R. Lyons of the 2003 Sliding DFT article referenced on Wikipedia STFT the same Dick Lyon on Wikipedia who has proposed for deletion the "Standing Wave Discrete Fourier Transform" (SWDFT) published in 1990?  2) You have commented that the Standing Wave Discrete Fourier Transform is "nearly identical to what is commonly known as the "sliding DFT" algorithm for the Short-time Fourier transform" - based on your comment, do you believe it is appropriate for Wikipedia STFT to reference the 2003 article?

E. Jacobsen and R. Lyons, The sliding DFT, Signal Processing Magazine vol. 20, issue 2, pp. 74–80 (March 2003).


 * I have no idea if Dick Lyon is the same as R. Lyons, although the spelling of the last name would lead me to guess that the answer is "no". But a quick search will verify that the Jacobsen and Lyons article has been widely cited and the terminology of "sliding DFTs" has been adopted by many authors in the signal-processing field.  So, as per Wikipedia policy (WP:COMMON and WP:RS), this is the terminology that should be used in Wikipedia, regardless of the identity of any given editor.   The point is that Wikipedia is not the appropriate place to promote work; by policy, we only write about things that are already widely accepted. — Steven G. Johnson (talk) 18:56, 2 May 2011 (UTC)

PS. As Dicklyon mentioned elsewhere, the sliding DFT indeed goes back well beyond the Jacobsen and Lyons paper (Dicklyon cited a book from 1975, I believe). The Jacobsen and Lyons article is merely a popular review of a well-known concept (although I could wish for a more thorough literature review, upon looking back at it the J&L article did indeed cite the same 1975 Rabiner and Gold book as Dicklyon). As another example, this 1997 article by Douglas and Soh (also cited by J&L) describes a similar sliding DFT, including the recursive updates, and cites work on it going back to the 1980s (although it doesn't claim to cite the originator of the algorithm, and seems to take it for granted that the concept is common knowledge). It is indeed unfortunate that your UCSD professors did not comment on this to you; perhaps they were not signal-processing people? — Steven G. Johnson (talk) 23:05, 2 May 2011 (UTC)

RV étiquette
Hi, I just wanted to tell you that simply RVing another editors changes can be a bit of a slap in the face. Starting a discussion on the topic seems like a minimal polite thing to do. 018 (talk) 21:59, 7 July 2011 (UTC)


 * The same could be said for deleting a large section of an article. Usually, the onus for starting a Talk discussion should be on the person proposing changes to the status quo. — Steven G. Johnson (talk) 00:54, 8 July 2011 (UTC)


 * Not really. WP:RV and WP:BRD are pretty much at opposite ends of what the meaning of a revert is, but there is no disagreement about WP:BOLD for good faith edits. 018 (talk) 02:56, 8 July 2011 (UTC)


 * You'll find that in just about any article, if you delete large sections of an article without discussion it will get immediately reverted. — Steven G. Johnson (talk) 15:33, 8 July 2011 (UTC)

In some cases an explanation in the edit summary is enough. Michael Hardy (talk) 21:46, 8 July 2011 (UTC)


 * Similar for reverts. The basic point here is that the pot is calling the kettle black: the person who makes a large deletion from an article (deleting not only content, but also interwiki tags etc.) without discussion can hardly complain when the deletion is reverted without discussion.  — Steven G. Johnson (talk) 23:44, 8 July 2011 (UTC)


 * Michael Hardy, I long ago realized that you think writing an edit summary is a really big deal and am made happy when you decide the rest of us are worthy of getting your effort in this area. 018 (talk) 01:16, 10 July 2011 (UTC)

User:A. Pichler
Hi Steven, I know you've had prior experience dealing with User:A. Pichler's edits to special functions articles. I find that they lack credibility, and are potentially very destructive to the project. Often, Pichler will replace identities that correspond to those in cited source with his own versions of them (without updating the references). When he does, it often seems to be the case that the cited reference doesn't support what he is adding. This is very problematic since the identities he adds are often impossible to verify without a reliable source. I have brought this up at WT:WPM once or twice in the past, but no one paid any attention at the time. Every few months, I need to do a (probably incomplete) audit of this user's edits. In my experience, about 80% of them need to be removed or reverted. Do you think it's worth pursuing some more permanent solution? If so, what can be done? Sławomir Biały (talk) 22:24, 18 September 2011 (UTC)


 * Unfortunately, my impression is that dealing with this kind of problematic editor takes a lot of time, no matter what course of action one decides on. The usual course is to try to persuade the user to follow Wikipedia policy (in this case, by including accurate citations), but this has been tried repeatedly with Pichler and he seems very reluctant to even acknowledge that there is a valid problem with his pattern of editing.  The next step might be a very short (e.g. 48-hour) block, but since Pichler is editing at a relatively slow rate over a long period of time rather than furiously edit-warring in a short period, my suspicion is that a short block will make little impression on him, nor is the blocking authority clear-cut.


 * I applaud you for doing an occasional audit of his edits, which is one approach (albeit time-wasting and probably incomplete as you note)&mdash;ultimately, this kind of approach can even succeed if you have enough patience, but may take literally years. Another approach might be to file some kind of arbitration request, but that also takes a lot of time and effort and has an uncertain outcome.  Dealing with well-meaning but uncooperative and long-lasting editors has always been a challenge on Wikipedia.  — Steven G. Johnson (talk) 23:02, 18 September 2011 (UTC)


 * This is one of the most frustrating aspects of Wikipedia for me. I have no doubt that Pichler's energies can be channelled in a constructive direction.  Perhaps it's true that all he needs is some guidance from more experienced editors. I will try to find someone to volunteer for the task.  Thanks for the advice.     Sławomir Biały  (talk) 01:43, 20 September 2011 (UTC)


 * In my perusal, I've noticed that the article Laguerre polynomials seems to be in particularly dire straits. It needs some expert attention (but I imagine that you have about as much time for Wikipedia these days as I do...)   Sławomir Biały  (talk) 03:51, 21 September 2011 (UTC)

MSU Interview
Dear Stevenj,

My name is Jonathan Obar user:Jaobar, I'm a professor in the College of Communication Arts and Sciences at Michigan State University and a Teaching Fellow with the Wikimedia Foundation's Education Program. This semester I've been running a little experiment at MSU, a class where we teach students about becoming Wikipedia administrators. Not a lot is known about your community, and our students (who are fascinated by wiki-culture by the way!) want to learn how you do what you do, and why you do it. A while back I proposed this idea (the class) to the community HERE, where it was met mainly with positive feedback. Anyhow, I'd like my students to speak with a few administrators to get a sense of admin experiences, training, motivations, likes, dislikes, etc. We were wondering if you'd be interested in speaking with one of our students.

So a few things about the interviews:
 * Interviews will last between 15 and 30 minutes.
 * Interviews can be conducted over skype (preferred), IRC or email. (You choose the form of communication based upon your comfort level, time, etc.)
 * All interviews will be completely anonymous, meaning that you (real name and/or pseudonym) will never be identified in any of our materials, unless you give the interviewer permission to do so.
 * All interviews will be completely voluntary. You are under no obligation to say yes to an interview, and can say no and stop or leave the interview at any time.
 * The entire interview process is being overseen by MSU's institutional review board (ethics review). This means that all questions have been approved by the university and all students have been trained how to conduct interviews ethically and properly.

Bottom line is that we really need your help, and would really appreciate the opportunity to speak with you. If interested, please send me an email at obar@msu.edu (to maintain anonymity) and I will add your name to my offline contact list. If you feel comfortable doing so, you can post your name HERE instead.

If you have questions or concerns at any time, feel free to email me at obar@msu.edu. I will be more than happy to speak with you.

Thanks in advance for your help. We have a lot to learn from you.

Sincerely,

Jonathan Obar --Jaobar (talk) 07:26, 12 February 2012 (UTC)

Young June Sah --Yjune.sah (talk) 21:01, 15 February 2012 (UTC)

Complaint at WP:COIN about your edits of the Karatsuba article


Hello Stevenj. Please see a report at Conflict of interest/Noticeboard, where an IP is complaining about your edits. Since he opened a discussion at COIN maybe he will follow up there and do some negotiating. If you are willing to explain the article issues there it would be helpful. I would warn the IP about his reverts of the article but it is not much use to leave messages for an IP-hopper. Thanks, EdJohnston (talk) 03:23, 10 March 2012 (UTC)
 * The Ip edits may be related to the accountUser:AliceNovak who restarted editing just after semiprotection was applied and filed a report at AIV that I declined as baseless warning the user instead. Nevertheless I should remind you in any case not to edit war and about the three revert rule. --Tikiwont (talk) 20:08, 10 March 2012 (UTC)
 * It seems clear that AliceNovak and the IP edits are one of several WP:SOCK accounts. I have not violated the 3RR as far as I can tell, but it seems that these sock accounts have (besides violating the sockpuppet prohibition).  — Steven G. Johnson (talk) 03:04, 11 March 2012 (UTC)
 * About the 3RR role that was just a reminder for you. I took the IP edits into account but AliceNovak not having received any previous 3rr warning or the like (and the IPs actually with red talk pages) I left a final warning and she got blocked subsequently. In any case please don't call other editors 'Trolls'. It hardly ever helps and is in this case also not really in line with your own analysis at WP:COI. Regards --Tikiwont (talk) 16:57, 11 March 2012 (UTC)

When I warned you above about edit warring and name cllaing, it didn’t even cross my mind you’d be able and willing block  yourself. This block goes clearly against WP:INVOLVED. There isn't even a block notice on her talk page. Please address as soon as possible. --Tikiwont (talk) 14:42, 14 March 2012 (UTC)


 * I assumed that inserting an editorial screed into multiple articles obviously fell under the "blatant vandalism" exception to the INVOLVED rule. If you disagree and feel that a block was somehow not obviously warranted here, go ahead and unblock her.   You're right that I should have posted a block notice, though.  — Steven G. Johnson (talk) 16:33, 14 March 2012 (UTC)
 * The posts were different than the previous ones, and there was no specific warning. More importantly I feel feel you personally shouldn't have acted. You've been having a conflict with her which makes it difficult for anybody judging things and are already seen by her as having a conflict of interest. So I've unblocked with a warning. As a general rule, I'd refrain from using administrative tools in areas such as this where you are involved in content work. Thanks for your understanding. --Tikiwont (talk) 18:54, 14 March 2012 (UTC)

Primary source: Britton
The problem with that primary source is not that it is simply a straightforward quote, it is part of a larger section that aims to prove that this is truly a mainstream historical belief. Yet, we have no idea how this quote from Britton was applied historically, how rapes were historically judged, etc. WP:SYNTH says "Do not combine material from multiple sources to reach or imply a conclusion not explicitly stated by any of the sources". Look at the very first example, 'simple example of original synthesis'. It is exactly the same thing that is occuring in that article. -- Avanu (talk) 01:02, 23 August 2012 (UTC)


 * Except that we already have secondary sources that argue that this is a widespread historical belief, and we already have secondary sources that point to medieval British jurisprudence as an example of the historical prevalence of the idea; the article is simply illustrating that point more vividly by a direct quotation from said jurisprudence. Besides which, the quotation "because no woman can conceive if she does not consent" is about as straightforward as you can get...we don't need sources on how this rule was applied to straightforwardly state that the impossibility of rape pregnancy was promulgated in a famous medieval legal text. — Steven G. Johnson (talk) 01:05, 23 August 2012 (UTC)

Pregnancy from rape‎
Thank you for all the hard work you've put into improving this article. --Mr. Vernon (talk) 14:58, 23 August 2012 (UTC)
 * Came here to say the same thing! Jokestress (talk) 10:43, 24 August 2012 (UTC)

Kudos
Hey, I just want to say that twice in the last week or so I've noticed edits that I thought were excellent and then seen that you made them. Totally unrelated articles (PPACA and James Leon Holmes). You rock! --Nstrauss (talk) 04:31, 24 August 2012 (UTC)

Disambiguation link notification for August 28
Hi. Thank you for your recent edits. Wikipedia appreciates your help. We noticed though that when you edited Pregnancy from rape, you added a link pointing to the disambiguation page Congo (check to confirm | fix with Dab solver). Such links are almost always unintended, since a disambiguation page is merely a list of "Did you mean..." article titles. Read the FAQ* Join us at the DPL WikiProject.

It's OK to remove this message. Also, to stop receiving these messages, follow these opt-out instructions. Thanks, DPL bot (talk) 13:52, 28 August 2012 (UTC)