It’s been a long day (Sunday). Starting off with my jumping the gun several times. And in opposite directions. biggrin

It was time I stepped back, took a few deep breaths, read through the entire techdude report, and spend the next number of hours mulling it all over.

So, starting nearly at the top:

By separating the certificate issue dates into groups the pre-2006 certificates show a clearly different security border design than the more recent 2006 through 2008 design.

I’m rather dumbfounded at how techdude says the pre-2006 certificates show a celarly different security design than the more decent 2006 through 2008 design so nonchalantly. This from the same person who had previously said:

If there was a new or modified certificate security pattern introduced there would also be a different revision date indicating that it had been revised after November 2001.

As consistency and precision are the keys to spotting forgeries it is extraordinarily unlikely – perhaps even to the point of impossibility – that the revision numbers would be the same while containing different “certificate patterns”.

Let’s say that again.

Extraordinarily unlikely – perhaps even to the point of impossibility – that the revision numbers would be the same while containing different “certificate patterns.”

However as I pointed out in a previous post, the revision numbers in the 2008 Michele COLB (and presumably the 2006 and 2007 anonymous COLBs as well), are the same as the revision numbers in the 2002 DeCosta COLB with the “clearly different security border design.”

Yet in his final report, he notes the different security border design as if he’d never uttered the words above and instead of it being extraordinarily unlikely – perhaps even to the point of impossibility – it’s as if it’s a common, everyday occurrence.

So why did he say what he did originally? Was there simply no truth to it?

Or was there?

While the borders in the images of the DeCosta and pre-2006 COLBs are clearly different, were the borders of the actual physical documents in fact different? I’m not so sure, but I’ll get to that later.

I won’t comment on the heat maps except to say that what I said I would get to later may offer a plausible explanation for the supposed differences.

Now on to what techdude calls “…the most damning piece of evidence – the remnants of the previous security border.”

By connecting the points together the original placement of what appears to be part of the original security border becomes apparent. The placement of the lines matches the expected location of a good security border based on the known placement from the known good certificates. The width and spacing of the highlighted areas also match a 1:1 scale overlay from a section of a known good security border.

I don’t know exactly where techdude expected the location of the suspected phantom border to be, but his red connect-the-dots lines don’t jibe with any known good certificates.

I took techdude’s image15 and overlaid the DeCosta COLB, lining it up with the actual borders of the Obama COLB. I then shifted it over to the right so that the left side of the DeCosta COLB’s border was centered between the two vertical red lines techdude had drawn in to indicate the position of the phantom border’s left side.

phantom.jpg

As can clearly be seen, the outside edge of the DeCosta COLB’s right border (indicated by the red arrow) falls well short of the red line drawn where techdude says the outside edge of the phantom border is located.

And with that, techdude’s theory and “the most damning piece of evidence” seems to have flown out the window.

Moving on.

A close examination of the security border itself reveals several repeating inconsistencies. Among them is a “weak line” which repeats once after every second bold line and a slightly downward curving end point where a straight line should end.

The “third line” phenomenon is something I have mentioned before. As for the “slightly downward curving end point,” I don’t think techdude is seeing anything here which isn’t also evident in the 2008 COLB and is simply a phenomenon of the acute side of the angle at which the lines terminate with the border.

Because they’re bitmaps, they have a finite amount of resolution. On the acute side of the angle, you have an ever-narrowing point as the line terminates with the border’s edge. However the pixels are of a fixed size and the ever-narrowing point has to be quantized. The end result is that the ever-narrowing point ends up being rounded off, making it appear that the acute angle has been filleted which can give an impression of a change of direction of the line.

You can see the effect in this image from the 2008 COLB (may need to squint your eyes to blurry things up a bit):

lines.jpg

Upon very close inspection portions of the security border also repeat every 240 pixels. By mapping this repeating pattern it becomes apparent that the pattern is laid out as a 240 x 240 pixel square that can be accurately extrapolated to the next position by simply counting 240 pixels. This type of tiling effect is commonly seen when an image has been modified by filling an area using an image editing application’s tile or pattern fill function.

Of course any repeating pattern can be infinitely expanded using squares of sufficient size. However I see nothing in the Obama COLB image which delineates any 240 pixel section from any other 240 pixel section, which is what would be needed to establish that the pattern has been tiled.

Techdude provides no evidence whatsoever to support his claim here. And until he does, this claim shouldn’t be given any credence.

During the course of my analysis several calls were made to various departments in the Hawaiian State Government in an attempt to better understand the process and procedures used to create, print, and distribute copies of the COLB form. While I was brushed off or hung up upon by almost all of the people I contacted I did manage to talk with a computer technician who was familiar with the computers and printers used by the Department of Health and the clerk’s offices. He was unwilling to give any specific details but did provide enough information to work with.

Conveniently this person isn’t named, making it impossible to verify what he is claimed to have said, or that there was ever any such person in the first place.

When asked if a COLB can be printed off center he said it was not possible and any misfeed would simply jam in the printer.

This only addresses a mechanical issue, and doesn’t rule out a software issue, such as printer margin settings, etc. So until software issues can be ruled out, how the Obama COLB came to be printed off center is far from being resolved.

When asked if he had seen the images on-line he replied that he had – and that there is “no way” they had printed something that looked like that which further backed up my conclusions.

This person who has yet to be identified.

Well I talked to someone there too. And they say “Yeah, we do that every day.” eek

Now let’s start to put the pieces of the puzzle together.

Let’s.

The KOS image security border pattern does not match any known specimen from any known year. It does not match the pre-2006 nor does it match the post-2006 certificate patterns.

The key word here being “image.” I’ll get to this later.

The placement of the text in all of the pre-2006 and post-2006 certificates are almost identical pixel location matches while the image’s text placement does not match any known specimen from any known year.

If by “the image” it is meant the KOS image, then to say the text placement in the KOS image does not match any known specimen from any known year is flat out false. While the text placement in the KOS image does not match the placement of text in the COLBs sourced by way of either TexasDarlin or Techdude, the text placement in the KOS image does indeed match that of the DeCosta, Smith and Tomoyasu COLBs.

Here are images of the DeCosta, Smith and Tomoyasu COLBs overlaid on the Obama COLB with approximately a 50% transparency:

obamadecosta.jpg

obamasmith.jpg

obamatomoyasu.jpg

Looking at this another way.

The text placement of the Obama COLB was consistent with the text placement of other COLBs, some known to have existed on the Internet before the whole Obama COLB issue came up. However the text placement of these new, anonymous COLBs, sourced through either TexasDarlin or Techdude, is inconsistent with all of the other known COLBs, even though all COLBs bear the same form revision dates.

The shape and kerning of the fonts used in the 2006 through 2008 certificates are identical while the shape and kerning of the fonts used in the image does not match any known specimen.

Yet one could also say the shape and kerning of fonts used in the new, anonymous COLBs do not match any known specimen. But it would be just as misleading as to say what Techdude says here. That’s because none of the other known specimens are of sufficient resolution to make such a comparison.

So as with text placement, all that can really be said is that the shape and kerning of the fonts in the KOS image doesn’t match that of the new, anonymous COLBs, and not “any known specimen.”

The KOS image shows clear signs of tampering such as the mismatch in RGB and error levels…

While AJStrata already offered up a plausible reason for why the RGB and error levels may differ, I have another, which I’ll get to later.

…visible indications of the previous location of the erased security border…

And your evidence of that was shown here not to support this theory.

…easily detectable patterns of repeating flaws around the new security border…

Unless I missed something in Techdude’s report, he hasn’t provided a single shred of evidence to support this claim. Just a low resolution image with some blue lines drawn on it.

…EXIF data that says the image was last saved with Photoshop CS3 for Macintosh…

I’m not sure what is so revelatory about this. SOME form of graphics software had to interface with the scanner to acquire the image and save it to the hard drive. Typically a scanner is installed as a TWAIN device, which allows various software to interface with it. You don’t just put something in a scanner, press a button, and zip bam boom you have an image file on your hard drive. Some form of software has to interface with the scanner.

So the EXIF data saying it was last saved using PhotoShop, a natural program to interface with a scanner, is not remarkable here.

At least the KOS image has EXIF data so that one can know this information. None of the MicheleCOLBs released have any EXIF data. Why is that? What software was used to acquire the images and save them to disk?

…and finally a technician from Hawaii who confirms it just looks wrong.

A technician from Hawaii who is not identified and for all anyone knows does not exist. Versus a named spokesperson, Janice Okubo, Communications Director for Hawaii’s Department of Health, who stated that she saw no obvious problems with it.

There are two obvious scenarios used to create the image that can be ascertained from evidence.

But first one has to actually have the evidence.

Either a real COLB was scanned into Photoshop and digitally edited…

For which this report offers no compelling evidence to support.

or a real COLB was first scanned to obtain the graphic layout then blanked by soaking the document in solvent to remove the toner. After rescanning the blank page to a separate image the graphics from the previously obtained scan could then be easily applied to the blank scan after some editing and rebuilding.

Again, for which this report offers no compelling evidence to support.

So much for that.

So where does this leave things? This gets us to what I said I would get to later.

The one thing that’s left out of all of this for which there is no argument is that the borders in the Obama COLB are clearly visibly different from all others.

So just what the hell is it with them damn borders?

I find them unbelievable. In two respects.

In one respect, I find them unbelievable because they simply don’t look like any others. The lines that make up both the outline and the hatch pattern inside them are are not distinct, seemingly being made up of as much or more green as black. Most of the lines making up the hatch pattern exhibit the phenomenon of having every third line being lighter and less distinct, in some spots leaving little evidence of anything. Also, the pitch of the hatch pattern that is visible is greater than the pitch of the hatch pattern in the Michele COLB. And finally, there seems to be an overall gray wash over the whole border.

In another respect, I find them unbelievable because I simply can’t conceive of anyone making any real attempt to forge or alter such a document producing one with such unbelievable borders. For that matter, I can’t conceive of anyone not bothering to even make a half-assed attempt to forge or alter such a document producing one with the borders seen in this image. They might just as well have stamped it “FORGERY” in big red letters.

It simply doesn’t make any sense. If the document was forged, then the person forging it obviously had to know what a real document looked like. How could they possibly produce this document if the hadn’t? And obviously knowing what a real document looked like, how could they possibly have produced the borders seen in the KOS image?

Again, it simply doesn’t make sense.

But in spite of that, there you have it. The image is there for all to see.

So what gives?

A clue may be found in the fact that Techdude used the words “security border” or “security borders” no less than 32 times in his report (give or take one or two).

But are they in fact security borders? One would seem to think so. The green weave pattern on the paper combined with the crosshatch pattern in the borders probably aren’t there for purely decorative purposes. But if the Michele COLB is any example, an image was obtained that presumably is a highly accurate representation of what the original document looks like in real life.

So where’s the security in that if you can just stick it into a scanner and get such accurate results?

Hmmm. Scanner.

Now, I’m no expert in the field here, but I seem to recall seeing, hearing or reading that there are document security features that are designed in conjunction with the manufacturers of photocopiers who design (at least some of) their machines such that they will not accurately reproduce those security features. And that scanners typically have no such features built in.

It’s been generally assumed that the KOS image was scanned from an original document. But was it?

Is it not inconceivable that Obama, instead of using his original COLB, made a color photocopy instead to be passed along to whomever was responsible for having the KOS image produced? And is it not inconceivable that the supposed security border, in conjunction with the photocopier all did what they were intended to do, and that this resulted in the unbelievable border seen in the KOS image?

I don’t know, but I find this scenario much more plausible than even a rank amateur forger coming up with such an unbelievable border.

Anyway, that’s my two cents worth.

k

Advertisements