[neverdem]

How to Spot a Fraud You'll have to go to the actual Times' webpage. i.e. URL = http://www.nytimes.com/2006/01/24/science/24frau.html?_r=1 , to click on the link so that the captions can be read. 56K dialup beware. It's one of those huge astronomy graphics that displace the format when browsing comments.

[coconutt2000]

HOW TO SPOT A FRAUD 101

1) Go to "http://www.nytimes.com/".
2) Take a good look.
3) Think on it.

[Oztrich Boy]

Gonna need a sharpeyed guard dog to spot that

[supercat]

I downloaded the above picture to the left from whitehouse.gov on February 16, 1996. I wonder what brightness and contrast adjustment will do to it...

Is that a leash?

[Rocky]

I'm glad the NYT is all over this scientific fraud thing. Wouldn't want the public to be misled, now would we?

(/sarcasm)

[lowbridge]

[Eastbound]

Piece of cake. But the bottom left photo must have been at lower resolution than their original. Couldn't get the backbround right.

[neverdem]

Tobacco Drug May Treat Cancer

Diabetes From Plastic? Estrogen Mimic Provokes Insulin Resistance Comment 24 links the original article.

FReepmail me if you want on or off my health and science ping list.

[Gamecock]

At least these are for real:

[R. Scott]

Remember when photographs didn’t lie?

[syriacus]

some authors had yielded to the temptation of Photoshop's image-changing tools to misrepresent the original data.

Cookbook for Flim-Flam

The pressure for cloning's so great
That a researcher might take the bait
And "doctor" his finds to mislead weaker minds.
Which proves -- greed is a strong human trait

[LurkedLongEnough]

The NYT: "We are learning new technologies so we can prevent future boo-boos..."

Give this one to the NYT. It would be better for their case to imply that image fraud happens more on TV.

[tallhappy]

The advent of scanners really allowed this sort of abuse.

Journals should only accept photos -- but the trend is the opposite, purely e-submissions.

That's as it should be, tbut integrity of data is harder to guarentee.

In talking with friends the actual thing nowadays is that one can't believe anything in any journal necessarily.

[editor-surveyor]

What I find interesting, is that while pointing out the fraud, they still want to call them scientists. This is where we get evolution from.

[aruanan]

Not surprising. In my photos I've used the Photoshop contrast tools, but these are applied to the entire image, not just to part of it. I've removed fluorescent debris that would detract from an otherwise good image. I've removed lanes in a DNA gel photo (because they were on the gel but not part of the experiment or not relevant to that particular portion of the experiment), but not bands within a single lane. I think a basic rule should be this: if one adds something that isn't there to make it appear that it is or removes something that is there that invalidates the claim being made, then it's fraud; otherwise, it's cosmetic.

Making a composite photo of, say, several different exposures to show what's actually there across the field when there's not enough range within any exposure to show everything is okay if it's stated right up front. This is done in astronomy. Some very faint dust bands may show up only after a very long exposure which overexposes other parts of the field, obliterating detail. Combining various portions of the field at various levels of exposure to reveal the most detail is okay to do if you're stating that you're doing it and detail how it was done.

[Mamzelle]

bumpmark

[Elsie]

It May Look Authentic; Here's How to Tell It Isn't

Hey!!

I thought this was gonna be a Pamela Anderson thread!!!

[Doctor Stochastic]

Note that none of the "Creationist Scientists" nor "ID Scientists" caught this. It was caught by peer review and efforts of other scientists.

[js1138]

The best way to authenticate a digital image would be to calculate a checksum using a crc32 or similar algorithm on the file right from the memory card. Store this in a secure place. The crc could be recalculated at any future time to match the original.

Any digital information could be secured from undetected change this way. It's done all the time -- every time you use a computer for anything.