Too good to be true, but not "too true" to be in Science?

Of the Kim Tae-Kook fraud, one commenter at The Scientist wrote:

I listened to his talk at a conference. That was about 2 years ago in Hawaii, a good place, he was one of the invited speaker. He was talking about his Magic beads, they are antibody conjugated magnetic beads to pull down signaling molecules so that they are on one plane and then he can do confocal to figure out the signaling pathways. Some of the confocal images, I could not see anything but he claimed there was something there. I also checked his publications, a recent Science paper, a resume of Harvard, Cold Spring Harbor, etc. My feeling at that time was that is almost too good to be true. But I was like whatever. It is funny that at that time, his compatriats, Huang was making big news and people were talking about that at the conference. Now looking back, they have the same magnetic signaling pathways almost at the same time.

A different commenter wrote of fraudulent papers in journals:

I second the question about how often journals are fooled by researchers. I know of several cases that have not been made public despite NIH DOI being informed because the statute of limitations ran out. It is now set at 6 years. But in addition to that, there are quite a few reasons why such things get quashed. A few reasons are:

- They are embarassing to those related to them on the fraudulent and other publications.
- Staff in the lab will lose their jobs or graduate student status.
- Universities will get black eyes.
- Retaliation can occur from friends of the accused, and from the accused themselves.
- A perception that "everybody does it."

Also:

http://ipbiz.blogspot.com/2008/03/just-like-hwang-woo-suk-another.html

Not clear what the journal Science is doing about the fraudulent paper by Kim Tae-kook.