ERDNASE

Discuss general aspects of Genii.
Leonard Hevia
Posts: 1951
Joined: January 17th, 2008, 12:00 pm
Favorite Magician: Dai Vernon, Frank Garcia, Slydini, Houdini,
Location: Gaithersburg, Md.

Re: ERDNASE

Postby Leonard Hevia » November 6th, 2017, 10:34 pm

lybrary wrote:
Leonard Hevia wrote:We can't be sure that Erdnase told Smith that he was actually related to Dalrymple.


Oh, I see. This argument requires for Smith to have incorrectly remembered something. What a novel thought! I will not argue against it. But I find the alleged comment rather unlikely in this situation. If Dalrymple would have actually drawn a nice full page portrait of his father then I guess you would have a point. But here his father was depicted as a tiny part of a larger image. It is silly to think that this would have been subject of a conversation.


No--my argument would not require Smith to have incorrectly remembered Erdnase's Dalrymple comment. I think Smith correctly remembered something that he might have misinterpreted back in 1901. Smith may have perceived Erdnase's comment to mean that he was related to Dalrymple by blood or marriage when Erdnase simply commented on a family connection.

Yes, it would be silly to think that this was the subject of conversation, but bear in mind that Sanders--if he was Erdnase--was in the presence of an artist sketching his hands. The thought that another artist also illustrated a family member years before might certainly have crossed his mind and be worth mentioning in passing to Smith.

Bill Mullins
Posts: 5911
Joined: January 17th, 2008, 12:00 pm
Location: Huntsville, AL
Contact:

Re: ERDNASE

Postby Bill Mullins » November 7th, 2017, 3:00 pm

For what it's worth, one of the co-authors of the Mosteller-Wallace paper we've been discussing died only last month.

Leonard Hevia wrote:Perhaps "Montana" was not intended to be W.F. Sanders, but it does look at least a little like him, and anyone who glances at a photo of W.F. Sanders and looks at that cartoon may believe that it is indeed him--including W.E. Sanders.


I suppose, then, we disagree. W. E. worked for his senator father in 1891-1892, and would have known some of the pictured senators. He would have been able to recognize which pictures were likenesses, meant to depict real people, and which were generic characters like the grizzled Montana miner.

I defy you to display a photo of W.F. next to the Dalrymple cartoon to family and friends and ask them if "Montana" resembles W.F. I eagerly await the results.


Whether or not a random person thinks "Montana" looks like W. F. doesn't make much difference. It's whether W. E. would have thought so, ten years later, so much that he'd tell Smith that he was related to the Dalrymple.

User avatar
Richard Kaufman
Posts: 27047
Joined: July 18th, 2001, 12:00 pm
Favorite Magician: Theodore DeLand
Location: Washington DC
Contact:

Re: ERDNASE

Postby Richard Kaufman » November 7th, 2017, 3:25 pm

Re Dalrymple, we can go on the noses alone. They are spot on, except for Sanders.
Subscribe today to Genii Magazine

Bill Mullins
Posts: 5911
Joined: January 17th, 2008, 12:00 pm
Location: Huntsville, AL
Contact:

Re: ERDNASE

Postby Bill Mullins » November 7th, 2017, 3:54 pm

At the top left of the Dalrymple cartoon is a man sticking his head out of a tent, named "Morton" (possibly "Horton"). I haven't been able to figure out who he is. There was no one in Congress by that name in 1891. Anyone have any ideas?

There's another reason to believe that "Montana" is not W. F. Sanders. Standing next to him is a miner representing the state of Washington. He is the only other one whose face is visible. Washington had two senators in 1891: Watson Squire and John Allen. "Washington" clearly doesn't resemble either of them; thus the mining characters are not meant to be representative of the senators of those states.

Tom Sawyer
Posts: 360
Joined: January 7th, 2012, 6:44 pm

Re: ERDNASE

Postby Tom Sawyer » November 7th, 2017, 4:44 pm

Pretty sure that's supposed to repesent Levi P. Morton, who (as everybody knows) was Vice President of the United States at the time of the cartoon.

Bill Mullins
Posts: 5911
Joined: January 17th, 2008, 12:00 pm
Location: Huntsville, AL
Contact:

Re: ERDNASE

Postby Bill Mullins » November 7th, 2017, 5:42 pm

Tom Sawyer wrote:Pretty sure that's supposed to repesent Levi P. Morton, who (as everybody knows) was Vice President of the United States at the time of the cartoon.


Everybody but me . . . .

Tom Sawyer
Posts: 360
Joined: January 7th, 2012, 6:44 pm

Re: ERDNASE

Postby Tom Sawyer » November 7th, 2017, 6:00 pm

Ha ha. Of course, in case anyone is wondering, until today I actually had no clue that one of our Vice Presidents had that name.

Leonard Hevia
Posts: 1951
Joined: January 17th, 2008, 12:00 pm
Favorite Magician: Dai Vernon, Frank Garcia, Slydini, Houdini,
Location: Gaithersburg, Md.

Re: ERDNASE

Postby Leonard Hevia » November 7th, 2017, 9:39 pm

I'm beginning to wonder if Dalrymple illustrated a caricature of W.F. in another issue of Puck. I'm aware that the characters in the front battle lines are symbols and not senators. The character representing Idaho has a long mullet and from photos I've seen of the Republican Senator for that state, his hair is cut short.

All that matters is the impression Sanders got from looking at Dalrymple's illustration.

Roger M.
Posts: 1598
Joined: January 17th, 2008, 12:00 pm

Re: ERDNASE

Postby Roger M. » November 7th, 2017, 10:38 pm

Richard Kaufman wrote:Re Dalrymple, we can go on the noses alone. They are spot on, except for Sanders.


This.

One thing artists with Dalrymple's talent don't do is screw up the noses of folks they draw.
In this case, the hooked nose is very obviously supposed to be a hooked nose.

Despite the potentially racially charged overtones of a drawing of a person with a hooked nose, this is obviously not the Senator from Montana.

Bill Mullins
Posts: 5911
Joined: January 17th, 2008, 12:00 pm
Location: Huntsville, AL
Contact:

Re: ERDNASE

Postby Bill Mullins » November 8th, 2017, 12:47 am

lybrary wrote: The MoWa paper does not say you must start at the known texts and then calculate the probability to the unknown.

Their paper does not say this, but their book, which amplifies the concepts, methods, and results initially described in the paper, explicitly says start with the known texts. (see Chap 3, Mosteller and Wallace, Applied Bayesian and Classical Inference: The Case of the Federalist Papers (1984))
"In Stage I, which has both logical and computational precedence, we analyze the papers whose authorship is known and convert the prior information about word rates into posterior distributions for these parameters."

Computational precedence -- you have to do this first.

However, with Erdnase the situation is entirely different.

That the Erdnase problem is different is an argument that the MW techniques aren't applicable, not that you should assume that the process as they describe it can be inverted and still be applied.

(And here you and I are in agreement -- The Erdnase problem is different. MW deals with a situation where a disputed text (or texts) is known to have been written by one of a group of authors, for all of whom we have known texts. Usually described as Authorship Attribution. The Erdnase problem is typically a situation where you have Erdnase and another text by a known author, and you want to know if they both are from the same hand. MW (1984) refers to this as a "Homogeneity Problem".)

User avatar
lybrary
Posts: 1169
Joined: March 31st, 2013, 4:59 pm
Contact:

Re: ERDNASE

Postby lybrary » November 8th, 2017, 9:04 am

Bill Mullins wrote:Their paper does not say this, but their book, which amplifies the concepts, methods, and results initially described in the paper, explicitly says start with the known texts. (see Chap 3, Mosteller and Wallace, Applied Bayesian and Classical Inference: The Case of the Federalist Papers (1984))

They do that for the practical reasons I explained above. For the Federalist Papers case you have to start with the known otherwise you cannot setup a large enough and consistent function word set. If you do not see the symmetry in their method, that it has nothing to do with which text is known and which text is unknown, then you do not understand the method. It comes down to the symmetry of if "A wrote B" then also "B wrote A".
Lybrary.com Magic & Gambling
preserving magic one book at a time

Bill Mullins
Posts: 5911
Joined: January 17th, 2008, 12:00 pm
Location: Huntsville, AL
Contact:

Re: ERDNASE

Postby Bill Mullins » November 8th, 2017, 1:33 pm

lybrary wrote: If you do not see the symmetry in their method, that it has nothing to do with which text is known and which text is unknown, then you do not understand the method. It comes down to the symmetry of if "A wrote B" then also "B wrote A".


The method is not symmetrical, and you can see this for yourself if you work the calculations in both directions. Suppose Erdnase used the word "Giraffe" 5 times in 52k words, that Teale used it 8 times in 85k words, and that Gallaway used it 3 times in 30k words.

The probability that Teale would use it 8 times if he wrote like Erdnase is 0.139.
The probability that Erdnase would use it 5 times if he wrote like Teale is 0.175.

The probability that Gallaway would use it 3 times if he wrote like Erdnase is 0.014.
The probability that Erdnase would use it 5 times if he wrote like Gallaway is 0.175.

If the process was symmetrical, you'd get the same numbers no matter which direction you went.

User avatar
lybrary
Posts: 1169
Joined: March 31st, 2013, 4:59 pm
Contact:

Re: ERDNASE

Postby lybrary » November 8th, 2017, 2:28 pm

Bill Mullins wrote:
lybrary wrote: If you do not see the symmetry in their method, that it has nothing to do with which text is known and which text is unknown, then you do not understand the method. It comes down to the symmetry of if "A wrote B" then also "B wrote A".


The method is not symmetrical, and you can see this for yourself if you work the calculations in both directions. Suppose Erdnase used the word "Giraffe" 5 times in 52k words, that Teale used it 8 times in 85k words, and that Gallaway used it 3 times in 30k words.

The probability that Teale would use it 8 times if he wrote like Erdnase is 0.139.
The probability that Erdnase would use it 5 times if he wrote like Teale is 0.175.

The probability that Gallaway would use it 3 times if he wrote like Erdnase is 0.014.
The probability that Erdnase would use it 5 times if he wrote like Gallaway is 0.175.

If the process was symmetrical, you'd get the same numbers no matter which direction you went.

Not true. You are looking at fine grained symmetry which is immaterial here. As I wrote earlier, you can apply the math, but you do not understand what it calculates. The method is obviously symmetric because if "A wrote B" then it follows that "B wrote A". That the specific numbers for each function word individually do change is simply a fact of the shape of the Poisson distribution which is not symmetric. But this asymmetry of the Poisson distribution does not change the symmetry of the MoWa method as a whole. If MoWa would give different results if applied from both ends then the method would be complete non-sense, which it is not. It has been shown to work surprisingly well in several cases.
Lybrary.com Magic & Gambling
preserving magic one book at a time

Jonathan Townsend
Posts: 8704
Joined: January 17th, 2008, 12:00 pm
Location: Westchester, NY
Contact:

Re: ERDNASE

Postby Jonathan Townsend » November 8th, 2017, 2:41 pm

I'd like a larger collection of text markers for analysis claims. The "(?)" itself is interesting but not IMHO persuasive. What's the significance of that marker given text over those years? A typographical fad?

The Federalist Papers came from a fairly specific context among a well known cohort of writers and editors. I'm not sure the "erdnase" text has such a literary context (responding to news items and arguments publishing over time etc).

? "A wrote B" -> author wrote text
which only by Borges or Eco permits the possibility of "B wrote A" -> text created author... unless that's the point of Erdnase?
What then is the role of the reader?
Or the limits of interpretation?

Our new abracadabra - created by words indeed.

JonT

Claims that saying "erdnase" backwards makes things vanish back to the fifth dimension. ;)
Mundus vult decipi -per Caleb Carr's story Killing Time

User avatar
lybrary
Posts: 1169
Joined: March 31st, 2013, 4:59 pm
Contact:

Re: ERDNASE

Postby lybrary » November 8th, 2017, 4:44 pm

Jonathan Townsend wrote:I'd like a larger collection of text markers for analysis claims. The "(?)" itself is interesting but not IMHO persuasive. What's the significance of that marker given text over those years? A typographical fad?

It has little to no meaning by itself. I simply pointed out a logical error in the argument how it was used in a past post, because applying the MoWa method resulted in exactly the opposite outcome than was argued.

Jonathan Townsend wrote:The Federalist Papers came from a fairly specific context among a well known cohort of writers and editors. I'm not sure the "erdnase" text has such a literary context (responding to news items and arguments publishing over time etc).

You are correct, the Erdnase case is of a somewhat different type. The question is essentially an open authorship verification or attribution case where we have to look at a number of possible authors but neither of these has to be Erdnase. The Federalist papers was a fairly simple closed cased with only two possible authors. Nevertheless, it is the same basic task: comparing texts based on usage frequencies of various linguistic features (ex. function words).

Jonathan Townsend wrote:? "A wrote B" -> author wrote text
which only by Borges or Eco permits the possibility of "B wrote A" -> text created author... unless that's the point of Erdnase?

I am simply using the abbreviation that the author name can either stand for the person or the text he wrote. So if you want to have it spelled out it would be:

- Author A wrote text a: known
- Author B wrote text b: known
If the result of the analysis is that author A also wrote text b, then it must immediately follow that author B also wrote text a.
Lybrary.com Magic & Gambling
preserving magic one book at a time

Jack Shalom
Posts: 1368
Joined: February 7th, 2008, 12:00 pm
Location: Brooklyn NY

Re: ERDNASE

Postby Jack Shalom » November 8th, 2017, 6:22 pm

Jonathan Townsend wrote:
JonT

Claims that saying "erdnase" backwards makes things vanish back to the fifth dimension. ;)

Hmm... new candidate put forward. Any extant comparison texts by MXYZPTLK?

Bill Mullins
Posts: 5911
Joined: January 17th, 2008, 12:00 pm
Location: Huntsville, AL
Contact:

Re: ERDNASE

Postby Bill Mullins » November 9th, 2017, 2:10 am

lybrary wrote: You are looking at fine grained symmetry which is immaterial here. As I wrote earlier, you can apply the math, but you do not understand what it calculates. The method is obviously symmetric because if "A wrote B" then it follows that "B wrote A". That the specific numbers for each function word individually do change is simply a fact of the shape of the Poisson distribution which is not symmetric. But this asymmetry of the Poisson distribution does not change the symmetry of the MoWa method as a whole. If MoWa would give different results if applied from both ends then the method would be complete non-sense, which it is not. It has been shown to work surprisingly well in several cases.


You are confusing the conclusion made by the user of the method, with the results of the method. Obviously, if A wrote B, then B wrote A. That's not an insight into MW's methods, it's a tautology.

But the methodology described by MW doesn't give you "A wrote B". It gives you a probability number. MW didn't decide that Madison wrote the disputed papers because that was the output of their method. The applied the method many times to multiple words, taken from large groups of texts, and got numerous probabilities. Only after examining all of these probabilities, and subjecting them to the Bayesian Analysis (which is the real point of their paper -- the demonstration of Bayesian Analysis on a statistical problem), do they come to the conclusion that Madison wrote the papers.

The case that's causing all the discussion, whether or not Erdnase is more similar to Gallaway than Teale, based on (?), doesn't have multiple data points. There's just the single marker -- (?). And applying the MW method to it in one direction (the correct one) suggests that Erdnase writes more like Teale than Gallaway (in that they both have non-zero usages of (?).) The actual probability that he would use it the way he did if his baseline was that of Teale is still quite small, though, so from the single test, it suggests that they are not the same. If you apply it in the wrong direction (like you did), it suggests that Gallaway writes more like Erdnase than Teale does. But the probability is still small, so it also suggest that they are not the same (again, the conclusion is symmetric, but not the output of the calculations).

Look at that again. Applying it in one direction tells you which of two authors a disputed author writes more like. Applying it in the other direction tells you which of two authors writes more like a disputed author. Two different, similar but non-symmetrical questions. The "amount" that Erdnase writes like Teale is not necessarily the same as the "amount" that Teale writes like Erdnase.

(I suspect that there are, but haven't found any, data sets that would indicate that it is likely that A wrote like B (probability > 50%), but it is unlikely that B wrote like A (probability < 50%). Such a situation would clearly be non-symmetric.)

Off on a tangent:
In Fred Mosteller's autobiography, The Pleasures of Statistics, is a chapter devoted to magic. It says:
When I worked in New York City in the 1940s, I came across an erudite book by Erdnase on card magic, or possibly how to cheat at cards. Jimmie Savage [a colleague] wanted to learn a little about magic, and so I lent him this book. Jimmie loved it for two reasons. First, the author said he wrote the book because he needed the money. Second, Erdnase treats each problem much like a mathematics text would. He has names for various devices such as false shuffling (riffling cards but leaving them in the original order), and he tells the reader exactly the sequence of use of these devices to produce the desired result. For anyone who hadn’t learned to manipulate cards at mother’s knee, too much skill is required. But for those who have it, Erdnase potentially moves their performance up many notches.

Too bad he wasted all that time on The Federalist Papers when, if things had gone just a little differently, he could have told us who wrote EATCT.

User avatar
lybrary
Posts: 1169
Joined: March 31st, 2013, 4:59 pm
Contact:

Re: ERDNASE

Postby lybrary » November 9th, 2017, 6:19 am

Bill Mullins wrote:The case that's causing all the discussion, whether or not Erdnase is more similar to Gallaway than Teale, based on (?), doesn't have multiple data points. There's just the single marker -- (?). And applying the MW method to it in one direction (the correct one) suggests that Erdnase writes more like Teale than Gallaway (in that they both have non-zero usages of (?).) The actual probability that he would use it the way he did if his baseline was that of Teale is still quite small, though, so from the single test, it suggests that they are not the same. If you apply it in the wrong direction (like you did), it suggests that Gallaway writes more like Erdnase than Teale does. But the probability is still small, so it also suggest that they are not the same (again, the conclusion is symmetric, but not the output of the calculations).

There is no correct or incorrect direction. Both directions are valid due to the symmetry of the problem. The Erdnase/Teale/Gallaway case can only be applied in the direction I did, because in the other the MoWa method is not defined due to the Gallaway frequency being zero. In the direction it can be applied Gallaway comes out much closer to Erdnase than Teale.

One gets the same result if newer stylometry methods are being applied to this case, for example Burrows Delta (2002), which is one of the most used and respected methods these days. Burrows does not use a Poisson distribution. He works with raw frequencies and a Manhattan distance metric. For the (?) case the numbers are:

Erdnase 3/52k = 0.058 (per 1k)
Teale 58/85k = 0.68 (per 1k)
Gallaway 0/30k = 0.0 (per 1k)

The Manhattan distances are:

Erdnase-Teale = 0.625
Erdnase-Gallaway = 0.058

Also with the newer state of the art stylo method Erdnase and Gallaway are much closer than Erdnase and Teale. MoWa agrees with Burrows. It also agrees with intuition. It is much more believable that somebody writing on magic using it 3 times, doesn't use it on a shorter text on a completely different subject, than using it 58 times writing on the same subject of magic.
Lybrary.com Magic & Gambling
preserving magic one book at a time

Bill Mullins
Posts: 5911
Joined: January 17th, 2008, 12:00 pm
Location: Huntsville, AL
Contact:

Re: ERDNASE

Postby Bill Mullins » November 9th, 2017, 2:24 pm

lybrary wrote:There is no correct or incorrect direction. Both directions are valid due to the symmetry of the problem. The Erdnase/Teale/Gallaway case can only be applied in the direction I did, because in the other the MoWa method is not defined due to the Gallaway frequency being zero. In the direction it can be applied Gallaway comes out much closer to Erdnase than Teale.


There is zero support in the literature for what you are saying. I've looked at many stylometry author identification/attribution studies over the years, and I've never seen one conducted where they used the unknown text as the baseline, and compared the known texts against it.

I've quoted the authors of the most important study in the field, Mosteller and Wallace, saying that you have to use the known texts as baselines, but you reject that.

The nomenclature for such studies has evolved so that the unknown text is called a "subject" text, and the known ones are called "training" texts. Vocabulary reflects practice, and the practice is to build baselines from the known texts.

All this tells me that when you say you can validly do it in the other direction, one of two things must apply:
1. You have stumbled onto a method that is otherwise unknown in the field.
2. You are wrong. (I think you can guess which way I lean on this).

Allowing for the possibility of #1 being correct, you should write this up and publish it. The Wasshuber inversion to the Mosteller-Wallace method may end up being what you are known for in years to come.

One gets the same result if newer stylometry methods are being applied to this case, for example Burrows Delta (2002), which is one of the most used and respected methods these days. Burrows does not use a Poisson distribution. He works with raw frequencies and a Manhattan distance metric. For the (?) case the numbers are:

Erdnase 3/52k = 0.058 (per 1k)
Teale 58/85k = 0.68 (per 1k)
Gallaway 0/30k = 0.0 (per 1k)

The Manhattan distances are:

Erdnase-Teale = 0.625
Erdnase-Gallaway = 0.058


Burrows Delta requires the use of the most frequently used words in the texts to be examined. By definition, a "word" like "(?)" which does not even appear in one of the texts is not appropriate for Burrows Delta analysis. The methodology isn't validated at usage rates = 0. Also, the Manhattan Distance is an ensemble measurement, taken over a number of words (typically 100 or more). The methodology isn't validated when used on only a single word.

When you so egregiously misapply the method, you don't prove anything.

Burrows does not use a Poisson distribution.

For that matter, neither did Mosteller-Wallace (despite what you've been implying). They used a negative binomial.

Bill Mullins
Posts: 5911
Joined: January 17th, 2008, 12:00 pm
Location: Huntsville, AL
Contact:

Re: ERDNASE

Postby Bill Mullins » November 9th, 2017, 2:59 pm

Bill Mullins wrote:Speaking of Dalrymple, Heritage Auctions currently has a piece of his art, used in Puck, up for auction.


Sold for $191 (including auction premium).

User avatar
lybrary
Posts: 1169
Joined: March 31st, 2013, 4:59 pm
Contact:

Re: ERDNASE

Postby lybrary » November 9th, 2017, 3:05 pm

Bill Mullins wrote:There is zero support in the literature for what you are saying. I've looked at many stylometry author identification/attribution studies over the years, and I've never seen one conducted where they used the unknown text as the baseline, and compared the known texts against it.

I've quoted the authors of the most important study in the field, Mosteller and Wallace, saying that you have to use the known texts as baselines, but you reject that.

Nowhere does it say that you cannot start with the unknown text. You start from the larger text (known or unknown) because that gives you the biggest set of function-words to work with. Once you actually implement this method and use it on different cases you quickly realize that. The typical test cases in the stylo literature have much more known text and that is why you see them start with it. Erdnase has written a lot more than several candidates and that is why starting with Erdnase is advantageous. No big insight there just obvious conclusions once one understands what the method actually calculates.

Bill Mullins wrote:Burrows Delta requires the use of the most frequently used words in the texts to be examined. By definition, a "word" like "(?)" which does not even appear in one of the texts is not appropriate for Burrows Delta analysis. The methodology isn't validated at usage rates = 0.

Again incorrect. It does not require the use of the most frequent words. What function words one uses with a Burrows Delta analysis is topic of research. One can use it with any words, frequent or not. But if frequent words are used they are typically calculated from the total ensemble of texts. 61 instances in about 160k words is frequent enough. Not all texts need to feature the word.

Bill Mullins wrote:Also, the Manhattan Distance is an ensemble measurement, taken over a number of words (typically 100 or more). The methodology isn't validated when used on only a single word.

Valid or not, the method is defined for a single word. I am not arguing that a single word stylometry makes a lot of sense. I am simply demonstrating that the argument that Teale is closer to Erdnase based on the single feature of (?) is incorrect, based on two methods MoWa and Burrows Delta, as well as intuition. Nothing more nothing less.

Bill Mullins wrote:
Burrows does not use a Poisson distribution.

For that matter, neither did Mosteller-Wallace (despite what you've been implying). They used a negative binomial.

Really?! They tried both the Poisson and the negative binomial but found that the added difficulties the negative binomial introduces is not worth the trouble. Here the quote from their paper:

"The entire study was carried out in parallel for the negative binomial and Poisson data distributions. The negative binomial introduces many complications that strongly influenced our allocation of effort, but few new ideas."

Ben Blatt in "Nabokov's Favorite word is Mauve" who tested MoWa on a set of 50 authors of novels and even more fan fiction authors also used the Poisson distribution and achieved above 96% accuracy.
Lybrary.com Magic & Gambling
preserving magic one book at a time

Tom Sawyer
Posts: 360
Joined: January 7th, 2012, 6:44 pm

Re: ERDNASE

Postby Tom Sawyer » November 9th, 2017, 8:19 pm

Bill, in connection with a hypothetical above, you said:

The probability that Gallaway would use it 3 times if he wrote like Erdnase is 0.014.
The probability that Erdnase would use it 5 times if he wrote like Gallaway is 0.175.

If you are saying "3 times in 30,000 words," and "5 times in 52,000 words," then the first part of the quotation sounds sorta counterintuitive to me.

Since, under the assumptions, they are both using the word "Giraffe" ROUGHLY once in 10,000 words, it would seem to me that 3 and 5, respectively, would be the most frequent "uses."

It thus seems weird (to me, anyway) that in the first case (Gallaway), he would use the word "3 times in 30,000 words" only 1.4% of the time -- and that in the other 98.6% of the 30,000-word samples he would use the word "Giraffe" zero, one, two, four, or greater-than-four times.

--Tom Sawyer

Bill Mullins
Posts: 5911
Joined: January 17th, 2008, 12:00 pm
Location: Huntsville, AL
Contact:

Re: ERDNASE

Postby Bill Mullins » November 9th, 2017, 11:50 pm

lybrary wrote:
Bill Mullins wrote:Burrows Delta requires the use of the most frequently used words in the texts to be examined.

Again incorrect. It does not require the use of the most frequent words.

Burrows, from his seminal 2002 paper: "The first step in the procedure is to establish a frequency-hierarchy for the most common words in a large group of suitable texts."

Bill Mullins wrote:Also, the Manhattan Distance is an ensemble measurement, taken over a number of words (typically 100 or more). The methodology isn't validated when used on only a single word.

Valid or not, the method is defined for a single word. I am not arguing that a single word stylometry makes a lot of sense. I am simply demonstrating that the argument that Teale is closer to Erdnase based on the single feature of (?) is incorrect, based on two methods MoWa and Burrows Delta, as well as intuition. Nothing more nothing less.

Except that you didn't use the MoWa method -- you inverted it -- and you didn't use Burrows Delta -- you used a single, non-common "word" instead of a group of most common words. And intuition tells us that two documents that have a feature in common are more alike that a third which does not share that feature. 0 for 3.

Bill Mullins wrote:
Burrows does not use a Poisson distribution.

For that matter, neither did Mosteller-Wallace (despite what you've been implying). They used a negative binomial.

Really?!

Really.
Mosteller, "A Statistical Study of the Writing Styles of the Authors of 'The Federalist' Papers," ProcAmerPhilSoc, v131n2, 6/1987
p. 136 "We investigated the usefulness of both the Poisson and the negative binomial. The negative binomial fits much better. . . .We used the negative binomial to fit the distributions of the words because of empirical evidence."
p. 137 "We used a substantial collection of words and the negative binomial distribution to generate the log odds in the manner just described."

Mosteller, The Pleasures of Statistics, (Springer, 2010) p. 57 "We had to complicate the theoretical model to allow for this, and the model we used is called the negative binomial distribution."

Mosteller-Wallace's book (Applied Bayesian and Classical Inference, Springer-Verlag, 1984) on the Federalist Papers study discusses this issue in detail and makes it clear that they did much of the work with both the Poisson and the negative binomial, and the negative binomial consistently was more accurate in describing the distributions. So their final conclusions were based on the negative binomial model.

Bill Mullins
Posts: 5911
Joined: January 17th, 2008, 12:00 pm
Location: Huntsville, AL
Contact:

Re: ERDNASE

Postby Bill Mullins » November 10th, 2017, 12:24 am

Tom -- I did make a mistake in the numbers I originally wrote. Where it says 0.014, it should say 0.223. Thanks for holding my feet to the fire.

And yes, low numbers like that do seem counter-intuitive, but that's the nature of the Poisson distribution. He'd use it 3 times 23% of the time, and 77% of the time, he'd use it 0, 1, 2, 4 or more times.

But my main point still holds. The direction you work the problem makes a difference in the numbers you get, so it isn't symmetrical.

Tom Sawyer
Posts: 360
Joined: January 7th, 2012, 6:44 pm

Re: ERDNASE

Postby Tom Sawyer » November 10th, 2017, 1:20 am

Bill, thanks for the clarification on that! --Tom

User avatar
lybrary
Posts: 1169
Joined: March 31st, 2013, 4:59 pm
Contact:

Re: ERDNASE

Postby lybrary » November 10th, 2017, 10:15 am

Bill Mullins wrote:Burrows, from his seminal 2002 paper: "The first step in the procedure is to establish a frequency-hierarchy for the most common words in a large group of suitable texts."

That was his original attempt and was simply a suggestion not a mandatory requirement. Many other researchers have used the same method for words of any frequency. It is still called Burrows Delta method because nothing in the math changes, just the list of words one decides to use. Including less frequent words has often improved accuracy. Some have shown best results with hundreds and even thousands of words, many of which only appear a few times.

Bill Mullins wrote:But my main point still holds. The direction you work the problem makes a difference in the numbers you get, so it isn't symmetrical.

The numbers change, the outcome which author is closest stays the same. If there is a case where the outcome would change then the result would be inconclusive because the authors style is too close to each other to make a decision.
Lybrary.com Magic & Gambling
preserving magic one book at a time

User avatar
Richard Kaufman
Posts: 27047
Joined: July 18th, 2001, 12:00 pm
Favorite Magician: Theodore DeLand
Location: Washington DC
Contact:

Re: ERDNASE

Postby Richard Kaufman » November 10th, 2017, 11:30 am

Is there anyone besides Chris who thinks Gallaway is a viable candidate?
Subscribe today to Genii Magazine

Roger M.
Posts: 1598
Joined: January 17th, 2008, 12:00 pm

Re: ERDNASE

Postby Roger M. » November 10th, 2017, 12:13 pm

Richard Kaufman wrote:Is there anyone besides Chris who thinks Gallaway is a viable candidate?


Looking back through the thread ... not a single poster has supported Chris's conclusion as it relates to Gallaway.

Gallaway as a viable candidate for Erdnase entered the realm of beating a dead horse many months ago.

Of course, new evidence is always welcome ... but to date it's simply been constant repetition of Chris's initial proposal.

User avatar
lybrary
Posts: 1169
Joined: March 31st, 2013, 4:59 pm
Contact:

Re: ERDNASE

Postby lybrary » November 10th, 2017, 2:31 pm

Roger M. wrote:Chris, I see what you mean by hearing Erdnase in the intro, it certainly seems more than just a similar style. Indeed it would be at the least a very similar style.
Lybrary.com Magic & Gambling
preserving magic one book at a time

User avatar
lybrary
Posts: 1169
Joined: March 31st, 2013, 4:59 pm
Contact:

Re: ERDNASE

Postby lybrary » November 10th, 2017, 2:35 pm

Richard Hatch wrote:
lybrary wrote: Here is a part from the beginning of the introduction to "Estimating for Printers"

"This is a practical book - it is not padded with ponderous editorial homilies, old newspaper clippings, interest tables or platitudinous dissertations on the uplift of the printing industry."


I agree with Chris that this does sound like it could have been written by the same author who wrote the "Professional Secrets" section of EATCT.
Lybrary.com Magic & Gambling
preserving magic one book at a time

Roger M.
Posts: 1598
Joined: January 17th, 2008, 12:00 pm

Re: ERDNASE

Postby Roger M. » November 10th, 2017, 4:42 pm

As I was one of those folks you quoted Chris, I'll just note that on an internet forum, the ultimate point is to engage in enjoyable conversation with other participants in the forum ... which is exactly was trying to do with you.
In the end though, trying to have a decent conversation with you was difficult, if not impossible, and I gave up.
I deduced quickly that disagreeing with any of your comments or statements resulted in your barrage of insults.

But as the post you dug up indicates quite clearly ... I did try.

Based on your TWO dug up posts ... you've essentially proven my point that nobody on this thread has ever agreed with you that Gallaway was Erdnase, or that Gallaway was likely to be Erdnase.

Richard Hatch
Posts: 2095
Joined: January 17th, 2008, 12:00 pm
Location: Providence, Utah
Contact:

Re: ERDNASE

Postby Richard Hatch » November 10th, 2017, 4:56 pm

Richard Kaufman wrote:Is there anyone besides Chris who thinks Gallaway is a viable candidate?

I think Chris' research and arguments as laid out in detail in his Hunt for Erdnase book make Gallaway viable. He was "at the scene of the crime" (McKinney's print shop), owned a copy of the first edition, sounds (to some) like Erdnase when writing, self published books in Chicago that bear some resemblance to the first edition (price on title page, for example). Chris makes a good circumstantial case for Gallaway, in my opinion. Lacking is any evidence of the skills described in the book, but that is true for all current publicly announced candidates except Milton Franklin Andrews and E. D. Benedict, I believe. I also don't think he sounds enough like Erdnase consistently to be convincing to me, and I frankly don't follow the stylometric discussions with any degree of understanding that leads to conviction! Here's a short list of the viable candidates that are known to me and publicly discussed:

Milton Franklin Andrews
Wilbur Edgerton Sanders
Edwin Sumner Andrews
E. S. Andrews (the con man unearthed by Todd Karr)
Edward Gallaway
E. D. Benedict

If I had to guess, I say there is probably a 50 percent or more chance that Erdnase was "none of the above". But I consider them all interesting and viable

Roger M.
Posts: 1598
Joined: January 17th, 2008, 12:00 pm

Re: ERDNASE

Postby Roger M. » November 10th, 2017, 5:52 pm

Viable = "feasible".

I don't believe MFA is feasible simply because the man wrote like an idiot, and no amount of "editing" could fix his incredibly poor grasp of the English language.

I also don't believe Gallaway is feasible, because as much as he was at the scene of the crime, so were perhaps 30 to 50 other people. He had a first edition, but so did (presumably) about 1000 other people at some point in time.
There is no evidence that has been brought forward about Gallaway that is in any way convincing.
Chris hasn't made a successful case such that he's received any confirmation of his newsletter declaration that he's "found Erdnase".

Richard is far more polite than I am, and although he may consider Gallaway a viable candidate ... I stronly stand by my earlier comment that nobody in this thread has ever stated they feel Gallaway is Erdnase, or that Gallaway is likely to be Erdnase.

The general consensus of this thread strongly, and consistently contrasting with Chris's declaration in his newsletter that "Erdnase has been found!".
That newsletter has, this week, broken new ground in the expression of sheer fantasy being sold as legitimate research ... however Chris finishes off todays newsletter with :

"Next week I will drop a bombshell ..."

So maybe Chris will present something more convincing than he has to date.

Bill Mullins
Posts: 5911
Joined: January 17th, 2008, 12:00 pm
Location: Huntsville, AL
Contact:

Re: ERDNASE

Postby Bill Mullins » November 10th, 2017, 5:59 pm

Richard -- Robert F. Foster has been proposed as a candidate, based on a perceived similarity of writing styles to Erdnase. If you accept him as "viable", then he'd certainly have some relevant skills, having written multiple books on card games, having been a member of the Parent Assembly in NY of SAM, and then of the Los Angeles Assembly, and having performed magic at SAM shows in NY.

(and while poking around about Foster, I found another relevant name-reversal. Eugene Homer performed as Rajah Remoh).

Richard Hatch
Posts: 2095
Joined: January 17th, 2008, 12:00 pm
Location: Providence, Utah
Contact:

Re: ERDNASE

Postby Richard Hatch » November 10th, 2017, 6:43 pm

Bill Mullins wrote:Richard -- Robert F. Foster has been proposed as a candidate, based on a perceived similarity of writing styles to Erdnase. If you accept him as "viable", then he'd certainly have some relevant skills, having written multiple books on card games, having been a member of the Parent Assembly in NY of SAM, and then of the Los Angeles Assembly, and having performed magic at SAM shows in NY.


L'Homme Masqué has also been proposed (by no less than Juan Tamariz) and clearly had many of the skills described in the book, being both an expert at sleight of hand and a compulsive gambler. But I don't consider either R. F. Foster or L'Homme Masqué viable, having long ago ruled out both on circumstantial grounds. I do still find them both interesting candidates, just not "viable".

Jeff Pierce Magic
Posts: 670
Joined: January 17th, 2008, 12:00 pm
Location: Orlando, FL
Contact:

Re: ERDNASE

Postby Jeff Pierce Magic » November 10th, 2017, 7:58 pm

I think that Chris has provided more evidence, if only circumstantial that Gallaway is Erdnase. But that's more evidence than most, so yes, Gallaway is a viable candidate.

Roger M.
Posts: 1598
Joined: January 17th, 2008, 12:00 pm

Re: ERDNASE

Postby Roger M. » November 10th, 2017, 8:01 pm

Jeff Pierce Magic wrote:I think that Chris has provided more evidence, if only circumstantial that Gallaway is Erdnase. But that's more evidence than most, so yes, Gallaway is a viable candidate.


Well Chris has presented exactly zero evidence that Gallaway is Erdnase ... but exactly what evidence did Chris present that convinces you that Gallaway is a "viable" candidate?

Be specific please.

User avatar
lybrary
Posts: 1169
Joined: March 31st, 2013, 4:59 pm
Contact:

Re: ERDNASE

Postby lybrary » November 10th, 2017, 8:14 pm

Bill Marquardt wrote:You have made an excellent case to include Gallaway as a candidate, and I am not really arguing against him being Erdnase.

Jonathan Townsend wrote:You're on the way to making a good case. Looking for meaningful measures is a big step toward credible argument.

Roger M. wrote:Chris has achieved the commendable undertaking of introducing a "new" candidate, and has had most of the regular posters to this thread accept this new candidate (Gallaway) as to be taken seriously as an addition to the somewhat short list of long established candidates.
Lybrary.com Magic & Gambling
preserving magic one book at a time

Roger M.
Posts: 1598
Joined: January 17th, 2008, 12:00 pm

Re: ERDNASE

Postby Roger M. » November 10th, 2017, 8:52 pm

lybrary wrote:
Roger M. wrote:Chris has achieved the commendable undertaking of introducing a "new" candidate, and has had most of the regular posters to this thread accept this new candidate (Gallaway) as to be taken seriously as an addition to the somewhat short list of long established candidates.


Like I said a few posts ago Chris, off the top you were well researched and polite.
Folks enjoyed engaging with you, myself included.

When your evidence wound up not going anywhere, and folks started pointing that out to you - you began insulting anybody who disagreed with your conclusions.

S. Tauzier
Posts: 31
Joined: January 16th, 2013, 8:36 am

Re: ERDNASE

Postby S. Tauzier » November 10th, 2017, 9:05 pm

None of you know who I am and Im sure you guys could care less what I think but I’ve studied the ERDNASE issue long enough to come to a conclusions.
I believe Chris is right.
You guys- you’ve spent too much time in love with the wrong angle. All that name backwards stuff- no.
Take all the Sanders and Andrews off the list.
I think that is a 100% miss!!
Reversing the name or changing letters around-
I just dont think thats it. If you were seriously trying to hide your identity- not so the public doesnt know but so anyone you may know personally doesnt figure it out- would you just reverse your name?
No, you would not. So take those cats off the list.
The thing that got me: very early on in Chris’ work he points out the German meaning of Erdnase.
To me that made more sense.
He just thought it was a good name to use as it was a word he was familiar with. Why cant it just be that simple?
I think the focus has slowly gone down the wrong road for a long time.
I know Im just pointing out one thing that really isnt proof but If we’re splitting hairs with Occam’s razor
I’m putting my money on Gallaway-
Chris’ research feels right.
So many things point to Gallaway.
I believe Chris hit the nail on the head.


Return to “General”