Is The 'CSI Phenomenon' Good For Science?

Tycoon Guy writes "With CSI: Crime Scene Investigation airing its 100th episode this week, I wonder, how do Slashdot readers feel about the show, and its two spinoffs? On the one hand, they've caused a boom in the popularity of forensic science college courses, and they glamorize geeks bent over microscopes, rather than smarmy lawyers. On the other hand, they may also promote an inaccurate view of science: prosecutors throughout the country now worry about juries that refuse to accept eyewitness accounts or even outright confessions, and instead exclusively demand the kind of forensic evidence they see on CSI. But of course, in the real world, you don't get a test like that in mere seconds - or without spending a substantial amount of money. So where does CSI rate on the geek scale for you?"

Comments

Its good, look at what happened with OJ

[yorkpaddy]

Its good to have the public have some knowledge of forensics. The OJ jury didn't believe overwelming forensics and set him free. Juries should also be smart enough to know hen to believe eyewitness accounts. oops, hoping for to much, why should I expect juries to be smart

Re:Its good, look at what happened with OJ

[nfras]

Actually, the OJ Simpson case is a good example of when the jury used forensic evidence properly. The jury was presented with lots of DNA evidence, blood stains, foot prints etc. When Mark Furman was asked if he planted evidence, he pleaded the fifth amendment. All forensic evidence is therefore suspect and cannot be given any weight. No matter whether you think he did it or not, the jury had no option but to acquit.
CSI is a good show, but it's just that, a show. The photographic close ups are the best. I remember one where they had a photo of a girl, there was a blur in her eye which they managed to extrapolate into a picture of her killer, pin sharp. It just not feasible.
I also love the nice sharp finger prints they take off wood, no hint of wood grain.
A bit more realism would be nice.

Re:Grade

[ThomaMelas]

I hate how they show PC's being used in forensics work on TV. I work for a company that does DVRs for CCTV systems and a ton of people call up wanting a system that will take a compressed file and let you Zoom it 50x and read newsprint at 200 ft away.

Re:Genuine Fractals 3.5

[Jerf]

You can wavelet and fractalate and vigourously wave your hands in the air until the lift you generate pulls you alongside a cruising 747, but you can not get more information than exists out of an image.

Most zooming algorithms suck, compared to the true content of the image, which is why we can do much better with our eyes. We know that is a "car", so we don't interpolate, say, a tire with jaggy lines, we know it is round.

But ultimately, take a fuzzy, off-true "3" and "5" and zoom out/blur enough, and there is no difference between the two, thus, no way to "backtrack" to the original image. There is a fundamental limit, and CSI routinely passes it.

You can play with contrast and brightness and sometimes retrieve a number or something. But your human eyes are already as good as you can expect at extracting a "3" from an image with suitable brightness and contrast. If you can't already see it, no magic algorithm is going to help. (I'm confident in this case our brains are close enough to optimal on this problem that no significant improvement can be made, even in theory, on still images.)