I made two mistakes yesterday. First I installed DISQUS to replace this blog’s commenting system. It was a noble essay, but for several reasons it just didn’t work out. My apologies to the fans of DISQUS, something with obvious advantages, but it’s not something that works for me here. Probably a number of comments from yesterday are lost for good.
The other mistake is more embarrassing. I said over at John Woodman’s blog, and I think here also, that I had scanned my own birth certificate using Adobe Acrobat, and that the optimization created layers and that different areas of the certificate had their own layer. I did that experiment a year ago, and what I just wrote is true. However, recently I said that more than one of those layers was a 1-bit layer and I can’t back that up. There were multiple layers in obvious color and multiple layers in obvious not color, but I cannot say that two of them are 1-bit. I’m downloading a trial copy of Adobe Illustrator now to see for sure. Previously I had just used Acrobat to export the images, but I know that one doesn’t get them all. With Acrobat image export, one doesn’t get any 1-bit layers at all (the background is 8-bit grayscale). The point is not whether I was right or wrong, but whether I reported the experiment accurately, and it turns out that based on what I had to work with at the time, I could not have supported the conclusion that I remembered.
I’m not going to lie to you, but I make mistakes and it is always appropriate to ask me for sources or evidence to back up a claim.
I was delivered by Dr. Webb and on my birth certificate the doctor’s name is “We” on the background layer and “bb” on a foreground layer. That’s the same kind of broken words and random layer assignment that we see on the Obama form. This is one reason that I believed last year, and still believe today that the layers are on the Obama certificate are too crazy to be the result of intentional human action.
Garrett Papit, the latest darling of the birther image analysis faction, ignores the crazy aspects of why single letters get broken into different layers while focusing on certain logical units of information that do hold together, such as the Registrar’s stamp. This is the operation of confirmation bias: accepting things that seem to support the bias, and downplaying or ignoring things that go against it. The registrars stamp is a distinct area surrounded by a broad boarder of background and a software algorithm isolating it doesn’t seem all that strange, nor the smoking gun of human intervention.
My birth certificate is in two parts. The top is the certification text with the signature of the state registrar when the certificate was printed for me and the seal. The bottom is the image of the original certificate. It’s all on a single sheet of security paper. When I scan this document using the HP Photosmart scanner utility, it thinks that I have two documents and it tries to save them as separate images. This isn’t PDF software, but it is an example of software somehow recognizing a logical division in a document, a registrar’s stamp even (!), without being asked to. I’ll show you a screen shot of what I mean:
By the way, my birth certificate is the only thing I’ve ever scanned that the software splits up this way. I had always considered it an annoyance until now.
My resume can be written to sound as impressive as that of Mr. Papit, and under scrutiny neither of us is an expert. But it’s not my word against his. It’s my sources against his. While I have, for example, found articles describing a halo effect around text in a compressed PDF file, he simply asserts that there aren’t any. I’m no more an expert than Papit is, and given his interest, he probably knows more than I do about PDF optimization. However, I don’t rely on my own expertise, and just about everything he says is on his own authority. So when I say that you should ask for sources and evidence, you should ask Papit too.