In the article about artist’s sketches matching the actual perp you bring up Ted Kaczynski.
If I remember correctly, the Unibomber sketch was not an eye-witness sketch but a sketch of what profilers thought he SHOULD look like.
John (JJDubs)
It’s always helpful to reference the column you are commenting on, in this case http://www.straightdope.com/columns/read/2937/how-accurate-are-police-sketches-of-suspected-criminals
Not a great column. “Judge for yourself” is not a good measure of how good an identikit picture is.
The question is: Given a picture of a suspect and a large pool of people, will people correctly find the member of the pool that matches the picture? That’s what the celebrity/student tests measured.
The question addressed by most of the column is: Given a picture and a photo, how likely are people to nod their heads and say “yep, those look like the same person”.
If that were the objective test of identikit, then we could optimise them for “generic” features that seem to match lots of people. Then we could have fun locking up all the innocent people who match the identikit.
Just as an FYI, this was the sketch that inspired me to send in today’s question:
Those who believe Kaczynski was a MKULTRA patsy claim the FBI manufactured sketches by morphing this photo until the witness agreed with the likeness.
Somewhat related question: how effective are standard survelliance cameras at IDing criminals? The few times I’ve seen video shown on TV news shows and the like, the picture is hardly useful to tell anything meaningful about the perp other than general build and appearance. It’s not like it includes a full-face shot like a portrait.
jjdubs said:
You do not remember correctly. There were 3 Unibomer sketches, all taken from 1 source - an eye-witness.
There was an article in Reader’s Digest many years back discussing a woman who had extraordinary success in sketching criminals from victim’s description. She did not show pictures to victims, or use the clip book of facial features, or ask whether the nose should be longer or narrower. Instead her technique was based on a discussion of the crime, picking up cues from the victim as they ‘relived’ the incident. She felt that showing pictures or prompting for descriptions would alter the victim’s memory. This link tells the story of successful find based on an artists age-progressed clay bust: John List - Wikipedia
Facial expressions seem (to me, anyhow) to affect the accuracy of the drawings. In the Berkowitz picture there’s a significant difference between the glance of the eyes that would make me think it was even less of a match.
“A nose, Q, not a banana.”
ed malin said:
See the link I posted. It is an excerpt from the book written by the lady who did the third (famous) Unibomber sketch, and how she had to get that from the eye-witness 7 years later, after two prior attempts that the witness didn’t accept but were used for years. That chapter from her book discusses the matter of corrupting the witness’s recollection and how to draw out the accurate information conversationally without confronting the matter directly.
Lute Skywatcher said:
Dude, I just discovered this while googling Ted Kaczinski sketch:
Dioptre said:
I try to think about it as, “If I saw this picture, and I knew or had seen the person responsible, would I be able to think of that person by that sketch?”
From Cecil’s list,
David Berkowitz, no
Ted Kaczinski, maybe (not older beard, but younger clean-shaven)
Timothy McVeigh, probably
Ted Bundy, no
Bruno Hauptmann, probably
To apply that to Cecil’s second link
- maybe
- probably not
- no
The problem is it is very hard to evaluate this from beforehand. I see a sketch - do I know this person? If not, how can I evaluate the likeness in the sketch?
Also, how much variance is allowed and still make you think of the actual person? How much leeway are people looking at the sketch giving to an artist trying to recreate what someone else saw?
From Cecil’s third link, didn’t read the paper, but the pics
- Not likely
- No way
- Not likely - actually, I think the realism hurts. It overly constrains the eye.
4 (composite) I doubt it, but it is better than the first three by miles.
UncleFred said:
From the Berkowitz example, I’m less thrown off by the glance from the eyes than the fact the sketch looks Chinese. The nose is wrong, the cheekbones are high, and the eyes aren’t scrunchy enough. The Ted Bundy pic says bushy hair and eyebrows.
Yeah, I found the same thing.
Let’s test this by doing a sketch of Cecil based on Una’s and others’ descriptions. 
Aye to that! And compare it to Ed’s! Exapno, you’re a genius!
Congratulations on being one of the select few to have a question answered by Unca Cece. I assume they will now show you the secret handshake, or something like that…