Tuesday, November 18, 2014

Critic of Polygraph Tests Accused of Teaching People to Lie to Government

When I read this story about a man who teaches people how to pass a polygraph test, I thought of my own multiple experiences with being polygraphed.

Regular readers know that I preach reliable / repeatable science. Polygraph exams are not scientific, they're art forms. They're investigative tools. I remember my last test. Here I am, all 6'7" and 345 lbs, with a blood pressure cuff that doesn't fit so it's down around my wrist, pneumo-tubes that are stretched beyond belief around my very large chest, and squirming in a special chair that I don't fit. I'm thinking, if the apparatus makes me this uncomfortable, and I can't sit comfortablly, how is this reliable or repeatable?

It's a game. It's a trick. The device is not a "lie detector." It just registers your biorhythms. It's the examiner who chooses to quiz you and drill down into certain questions. It's an investigative tool - nothing more.

To me, training someone to pass a polygraph is the same as those books that teach novice Photoshop users to pass the Photoshop ACE exam. It's hard for me to take the government seriously on this. Polygraph exams are tricks, nothing more. They play their games, and the examinee plays his. Fair's fair, I guess.

Monday, November 17, 2014

Shooting Incident Reconstruction

Over the weekend, I was honored to be invited to attend a Shooting Incident Reconstruction seminar taught by local firearms expert, Dr. Bruce Krell, PhD. I wasn't sure if I could spare the time to make it to the class, but the course announcement really got me interested. Here's a snippet:

During the morning, we will be in an air conditioned classroom.
During the afternoon, we will be out on a private shooting range.

NOTE: We don’t take a lunch break. The range does not have food facilities.
Bring your own lunch and plan to eat during the last half hour of the lecture.

The afternoon session may be hot and may also be windy.
My guess right now is that shorts and t-shirt might be appropriate.
But, bring a sweater in case you are too cold in the classroom.

Wear tennis shoes or hiking boots.

Please bring sun block and a baseball cap or a hat.

Bottled water would also be a good idea.

If you have shooting ear muffs, please bring them.
If not, don’t worry. We will have plenty of the soft ear plugs.

Some of you will be participating in some of the measurement tasks.
Some of you will be participating as actors in some of the reconstructions.
Some of the reconstructions may involve being on the ground, so if
you might be willing to volunteer, dress so that you don’t’ mind getting dirty.

We will be discussing some of the strategies for photo perspectives.
So, I will be taking photos of some sections of the reconstructions.

The first half of the course featured a ton of slides on the math and stats of trajectories, ejector patterns, ricochets, and so forth. He walked the class through several old cases, demonstrating how reconstructions are performed using all of the available reports and on-scene work. Then the class moved up to the range for some practical exercises.

In all, it was an enjoyable class.

If you're in the Los Angeles area and can make it to one of Dr. Krell's classes, you'll get a ton of useful information. Plus, you'll never look at a shooting incident in the same way again.

Friday, November 14, 2014

How Courts Miss Bad Forensics

As a follow-up to yesterday's post, I received a link to this story - How Courts Miss Bad Forensics.

This story will really piss you off. It pissed me off. Here are some excerpts:

"Despite Hayne’s impossible workload (over about 20 years he performed on average 1,200 to 1,800 autopsies per year, by his own admission), his lack of board certification, and the fact that he has on multiple occasions given testimony that other medical examiners have said ranged from implausible to malpractice, to date no court has rejected Hayne as an expert witness. While some courts have overturned a handful of convictions that were based on his testimony, they’ve only done so in the most egregious instances. Where Hayne has given plausible testimony, or even implausible-but-not-completely-nutty testimony, the courts have generally refused to intervene.

But if Hayne isn’t a credible witness, he isn’t a credible witness. If he has shown that he’s willing to say outrageous things in a few cases, has lied about his certification, and has been shown to be sloppy and unprofessional in his work, the cases in which he gave plausible but debatable testimony (and was opposed by a more competent medical examiner) should be seen just as tainted as those in which his testimony was transparently ridiculous."

"So far, the courts haven’t agreed. But a two-word phrase makes last week’s ruling different than all of the others.

The evidence shows the witness for Louisiana, Dr. Steven Hayne, a now-discredited Mississippi coroner, lied about his qualifications as an expert and thus gave unreliable testimony about the cause of death.

To my knowledge, this is the first time a court has acknowledged that Hayne has been broadly “discredited.” The acknowledgment is significant because of what the panel does next. Under federal law, in order to obtain a new trial based on newly discovered evidence, a convicted person must show that the evidence is either new or could not have been discovered at trial, that had the evidence been available at trial the jury would likely have convicted, and must file his petition based on the new evidence within a year of when the evidence “could have been discovered through the exercise of due diligence.”

The essence of the ruling in this case is that because the defendant missed a deadline, it simply doesn’t matter that Hayne may not be a credible witness. Shocking and sad.

Read the whole article. See for yourself what happens when the courts fail to police themselves.

Thursday, November 13, 2014

George Reis and the Frye Hearing

Along the lines of who are these people anyways, George Reis writes about his experiences with one of the members of the LA County Superior Court's list.

"That ‘expert’ has testified 25 times in the past, but has had no training in Forensic Video Analysis or in any Comparative Science. His work experience was only peripherally related to video analysis. He does not have any certifications in video analysis, photographic analysis, or anything related to these fields.

His methods in video analysis and in comparisons are not considered best practices by any peers, any forensic organization, or in any publications. Additionally, his comparison method is a method that the Facial Identification Scientific Working Group (FISWG) specifically states is not appropriate for comparisons. Further, he identified artifacts as being features, which the Scientific Working Group for Imaging Technology (SWGIT) points out is something that should not be done."


"So, the next time a public defender in Los Angeles needs an expert, he or she will take a look at the approved list and see this expert’s name. That public defender may not have the time to search out the background of that expert, but instead may just assume that because he is on the list, he must be competent. It is a frightening situation. I hope that the LA Superior Courts will review the criteria used for placing (and retaining) individuals onto their experts panel."


Wow. Where do we start with the needed reforms?

Wednesday, November 12, 2014

It's free, but ...

On Monday, I wrote about Ski's article on reducing JPEG artifacts. As a follow up, I went to the Foray web site and downloaded their free ForayJPGArtifactReductionTool. It's a saved action file that you can load into Photoshop that automates the steps outlined in Ski's article.

When I loaded the action file, I noticed a few things about the steps that I want to share with you.

  • All of the work is done to the Background layer. Can you articulate why this is a bad idea?
  • The image is resized. Can you articulate why?
  • Since all the work is done to the Background layer, how are you controlling the complimentary color artifacts that result from Unsharp Mask? I would have expected that the script would create a copy of the layer and changed the blending mode, or worked in LAB mode and just worked on the L channel.
  • You already know how I feel about Auto Adjustments
I would have hoped that the Layers Panel looked something like this when the action was completed.

Sadly, it wasn't.

So, caveat emptor. Just because it's free doesn't mean you should use it.