Hany Farid in IEEE Spectrum:
Altering digital imagery is now ubiquitous. People have come to expect it in the fashion and entertainment world, where airbrushing blemishes and wrinkles away is routine. And anyone surfing the Web is routinely subjected to crude photographic mashups like the Palin hoax, whose creators clearly aren’t interested in realism but in whatever titillation or outrage they can generate.
But other photo manipulations demonstrate just how difficult it has become to tell altered images from the real thing. For example, in 2005 Hwang Woo-Suk, a South Korean professor, published a paper in one of the most prestigious scientific journals, Science, claiming groundbreaking advances in stem-cell research. But at least 9 of the 11 uniquely tailored lines of stem cells that Hwang claimed to have made were fakes. Much of the evidence for those 9 lines of stem cells involved doctored photographs.
Apparently, Hwang’s fabrication was not an isolated occurrence. Mike Rossner, then the managing editor of The Journal of Cell Biology, estimated that 20 percent of the manuscripts his journal accepted contained at least one image that had been inappropriately manipulated. Since then, a number of scholarly journals have implemented new fraud-detection procedures, such as software that makes it easier to compare images within or between documents. The incidence of image fraud in scholarly publishing has not declined, though; indeed, it seems to be on the rise.