Sunday, October 03, 2004
Hailey Halloween Horror
I'm working on a little fisk of David E. Hailey's paper (revised web, revised PDF), though it's hardly worth it; I dont' think anyone takes it seriously anymore.
Whether he was fooling himself or attempting to fool others is probably unanswerable at this point, but LORD! I have never read such unsupported gobbledy-gook passed off as research in all my life.
I don't really give him a pass on attempting to persuade others with fakery.
I think he was trying to make himself a little Charles-Johnson style "convincer" here (orig psd on his site here), by tweening each character in photoshop. Here's an animated gif I made that highlights some of the more obvious text "tweens".
When I try to make a convincing photomash, if there is text to match, I often resort to cutting and pasting letters from the original image, instead of picking a near-match font and using the photoshop fonts and text tools to make my doctored copy. If you cut letters out of an existing image, you can kern the letters by hand, and get a perfect color and font match. Using selections of cut text, however, has its downside.
Hailey falls victim to it - the bottom of the text line is uneven, and he did not clean up his selections well enough to keep from cutting off adjacent letters, and he inserted artifacts of cutting and pasting around the edges of his selection.
He thought better of putting this "reproduction" in his paper, apparently. However, he did the same thing in his figure five "convincer" (gif) (psd) that did make it into his paper.
Hailey was capable of producing a "perfect match" , with a digital "typewriter-style" font, as long as he changed the spacing with cut-and-paste, and inserted certain letters and numbers from entirely different fonts. Therefore a typewriter produced the memos.
If you didn't follow that, you needn't bother with Hailey's arguments.
Still, the fellow does have the power to amuse. I was looking through papers in Hailey's CV and ran across this gem :
"Online Education Horror Stories Worthy of Halloween: A Short List of Problems and Solutions in Online Instruction" [Computers and Composition 18 (2001) pp. 387-397] (pdf) , wherein he discusses some of the pitfalls an instructor might face teaching online courses, based on his own personal experience.
Listen to this lament about rogue online students from the journal page 389:
And p.391:
That dastardly interweb! On the bright side, he should be able to write another great paper.
Whether he was fooling himself or attempting to fool others is probably unanswerable at this point, but LORD! I have never read such unsupported gobbledy-gook passed off as research in all my life.
I don't really give him a pass on attempting to persuade others with fakery.
I think he was trying to make himself a little Charles-Johnson style "convincer" here (orig psd on his site here), by tweening each character in photoshop. Here's an animated gif I made that highlights some of the more obvious text "tweens".
When I try to make a convincing photomash, if there is text to match, I often resort to cutting and pasting letters from the original image, instead of picking a near-match font and using the photoshop fonts and text tools to make my doctored copy. If you cut letters out of an existing image, you can kern the letters by hand, and get a perfect color and font match. Using selections of cut text, however, has its downside.
Hailey falls victim to it - the bottom of the text line is uneven, and he did not clean up his selections well enough to keep from cutting off adjacent letters, and he inserted artifacts of cutting and pasting around the edges of his selection.
He thought better of putting this "reproduction" in his paper, apparently. However, he did the same thing in his figure five "convincer" (gif) (psd) that did make it into his paper.
Hailey was capable of producing a "perfect match" , with a digital "typewriter-style" font, as long as he changed the spacing with cut-and-paste, and inserted certain letters and numbers from entirely different fonts. Therefore a typewriter produced the memos.
If you didn't follow that, you needn't bother with Hailey's arguments.
Still, the fellow does have the power to amuse. I was looking through papers in Hailey's CV and ran across this gem :
"Online Education Horror Stories Worthy of Halloween: A Short List of Problems and Solutions in Online Instruction" [Computers and Composition 18 (2001) pp. 387-397] (pdf) , wherein he discusses some of the pitfalls an instructor might face teaching online courses, based on his own personal experience.
Listen to this lament about rogue online students from the journal page 389:
...two or more such students working together to complain about the teacher can create a much more dangerous situation. Enlisting even one ally can greatly increase the credibility of a student's complaints if they are directed outside the class, where a student found a confederate and both wrote not only to the teacher but also to the department head and the state governor. [They can even do so wearing pajamas!!! ed. note - SW]
And p.391:
Complaints in traditional, face to face classes are more likely to stay within the class of department bounds, where it is easier to for the teacher to control...
..By contrast fires in an online class can spread quickly and unconrollably...
..Multiple and strongly worded complaint messages from more than one student can blow a minor problem out of proportion and can damage a teacher's careerbefore the teacher even knows that he or she is in trouble - especially if the addressees include powerful political figures such as a state governor who has a particularly strong interest in online education....p. 394 -
Teachers who elect to teach online may inadvertantly be putting themselves and their career at risk
That dastardly interweb! On the bright side, he should be able to write another great paper.
Comments:
Links to this post:
<< Home
I already 'fisked' Hailey when he released his garbage, maybe my comments are in line with what you are working on, here it is:
First, one must be careful to ignore Hailey's conclusions. Whenever one tries to evaluate an expert, their conclusions must be pierced and the underlying analysis/evidence the conclusions are based on must be examined. Conclusions are nothing more than opinion. This is an example of one such conclusion: "The critical arguments of the above document experts are both spurious and uninformed. The ability of the military to produce the proportional text with a superscript “th” with a typewriter is beyond question." The evidence proves this statement false. No typewriter was capable of all of those things, although a typesetter might have been. Since so many of Hailey's conclusions are so wrong, I could fill 10 pages with a deconstruction of his conclusions, replete with citations to the other experts linked by allahpundit, but that would be long, boring, and few would read it. Instead I will only address the new evidence Hailey uses.
#1. Figure 5: the flawless replica**
Not so fast. He doesnt tell us how he created it. That is a very suspicious omission. Until he does, and until other can reproduce this 'exact match' by following the same steps he did, this "evidence" is just an opinion. Looking at samples of American Typewriter Condensed, which Hailey claims to be using, yields very different results. Note the exaggerated tails, absent from Hailey's "reproduction".
http://www.myfonts.com/fonts/agfa/itc-american-typewriter/condensed/testdrive.html?s=You+are+ordered+to+report+to+commander&p=24
Hailey writes "My recollection is shaky, but I believe the font was called “American Typewriter.” The fact that the Hailey's memory is so fuzzy in regard to his most crucial evidence is highly suspicious. It is even more suspicious when Hailey clearly hides behind a whole family of fonts instead of picking just one and sticking with it.
#2. Table 1 and Figure 6: the "1" issue
The 1s in the CBS document are so badly distorted that attempts to nitpick from among similar fonts based on very tiny visual details is ludicrous. NONE of the 'fresh' 1s match the distorted CBS documents, even the American Typewriter Condensed 1s are unable to reproduce the disortion. Although the undistorted American Typewriter Condensed 1s seem more similar to the distorted CBS document 1s than the undistorted Times New Roman 1s, this would change after introducing distortion. Therefore this "evidence" actually compares apples to oranges.
#3. Figure 6: Comparing the fonts
Note that the Times New Roman example is BOLD. Why? Note that it does not match in width to the CBS memo, even though we know that MS Word will create a copy that lines up exactly (thanks to the LGF graphic). Note the "in a" is shortened to just "in". This is a rigged comparison.
#4. Figure 7: "Flight"
Note the many differences between the typewriter condensed font and the CBS memo example: (1) the top tip of the "L" is far too deep in the "F". (2) The top tip of the "I" extends too far left. (3) The base of the "I" is too wide. (4) the 'tail' on the upper-right "G" is too large. (5) The "H" is too strongly curved. (6) The cross on the "T" is to wide and too thick. (7) All serifs (the 'feet') are too thick and too wide.
#5. Figure 13: Defective "E"s
Note that the flaws in the letters are not consistent, as one would expect if the same damaged letter block typed them. Note that one of the "E"s does NOT show any wear on the upper left. Not that Hailey neglected to point out the "E" that disproves his assertion.
- Kaltes written 9/29/04
** UPDATE: Hailey now claims to use ITF American Typewriter Condensed. Use the link to this font and compare it to Hailey's 'reproduction'. THEY ARE NOT THE SAME. Look at the lowercase 'a' in particular, notice the lack of a tail (aka curved serif).
Post a Comment
First, one must be careful to ignore Hailey's conclusions. Whenever one tries to evaluate an expert, their conclusions must be pierced and the underlying analysis/evidence the conclusions are based on must be examined. Conclusions are nothing more than opinion. This is an example of one such conclusion: "The critical arguments of the above document experts are both spurious and uninformed. The ability of the military to produce the proportional text with a superscript “th” with a typewriter is beyond question." The evidence proves this statement false. No typewriter was capable of all of those things, although a typesetter might have been. Since so many of Hailey's conclusions are so wrong, I could fill 10 pages with a deconstruction of his conclusions, replete with citations to the other experts linked by allahpundit, but that would be long, boring, and few would read it. Instead I will only address the new evidence Hailey uses.
#1. Figure 5: the flawless replica**
Not so fast. He doesnt tell us how he created it. That is a very suspicious omission. Until he does, and until other can reproduce this 'exact match' by following the same steps he did, this "evidence" is just an opinion. Looking at samples of American Typewriter Condensed, which Hailey claims to be using, yields very different results. Note the exaggerated tails, absent from Hailey's "reproduction".
http://www.myfonts.com/fonts/agfa/itc-american-typewriter/condensed/testdrive.html?s=You+are+ordered+to+report+to+commander&p=24
Hailey writes "My recollection is shaky, but I believe the font was called “American Typewriter.” The fact that the Hailey's memory is so fuzzy in regard to his most crucial evidence is highly suspicious. It is even more suspicious when Hailey clearly hides behind a whole family of fonts instead of picking just one and sticking with it.
#2. Table 1 and Figure 6: the "1" issue
The 1s in the CBS document are so badly distorted that attempts to nitpick from among similar fonts based on very tiny visual details is ludicrous. NONE of the 'fresh' 1s match the distorted CBS documents, even the American Typewriter Condensed 1s are unable to reproduce the disortion. Although the undistorted American Typewriter Condensed 1s seem more similar to the distorted CBS document 1s than the undistorted Times New Roman 1s, this would change after introducing distortion. Therefore this "evidence" actually compares apples to oranges.
#3. Figure 6: Comparing the fonts
Note that the Times New Roman example is BOLD. Why? Note that it does not match in width to the CBS memo, even though we know that MS Word will create a copy that lines up exactly (thanks to the LGF graphic). Note the "in a" is shortened to just "in". This is a rigged comparison.
#4. Figure 7: "Flight"
Note the many differences between the typewriter condensed font and the CBS memo example: (1) the top tip of the "L" is far too deep in the "F". (2) The top tip of the "I" extends too far left. (3) The base of the "I" is too wide. (4) the 'tail' on the upper-right "G" is too large. (5) The "H" is too strongly curved. (6) The cross on the "T" is to wide and too thick. (7) All serifs (the 'feet') are too thick and too wide.
#5. Figure 13: Defective "E"s
Note that the flaws in the letters are not consistent, as one would expect if the same damaged letter block typed them. Note that one of the "E"s does NOT show any wear on the upper left. Not that Hailey neglected to point out the "E" that disproves his assertion.
- Kaltes written 9/29/04
** UPDATE: Hailey now claims to use ITF American Typewriter Condensed. Use the link to this font and compare it to Hailey's 'reproduction'. THEY ARE NOT THE SAME. Look at the lowercase 'a' in particular, notice the lack of a tail (aka curved serif).
# posted by 
: 7:27 PM
: 7:27 PM Links to this post:
<< Home
