[Goose]

Great reply jonnyss. Sums up my thinking as well. 

MTL, in the best light, is still responsible for a paper with fraudulent data, even if his postdoc did all the photo editing without MTL's knowledge or direction. The postdoc is his trainee, and direct report in business parlance. If he screws up, the professor is still responsible on some level for not properly monitoring or training that person.

And it happened multiple times.

There is one very good reason to believe that the doctoring of the MTL images was in fact NOT blatant, really hard to miss. As you undoubtedly know, there papers are peer reviewed before publication. NONE of the reviewers raised questions about these figures. If it was so obvious why didn't the reviewers see it? Clearly, the idea that it was obvious is not supported by facts.

Peer review is simply not designed to pick up things like this. It's done on a volunteer basis in the spare time of usually overworked scientists, who generally trust that people aren't lying about their actual data.

The peer review process could certainly use some changes but that's ancillary to the discussion of MTL's culpability.

If the doctoring was "blatant", anybody who looked at the images at all critically should have picked it up. I would argue peer review is indeed designed to pick up blatant, obvious fraud. I do agree that the bias of the reviewers is that the data is genuine, but when evaluating the culpability of MTL the assertion that the fraud was "blatant" certainly would greatly increase his culpability. I expect the bias of the PI is that his people's data is genuine as well, probably more so. The reviewer's bias if anything should be less. There are also at least two reviewers, so one reviewer doing a slipshod job wouldn't explain not seeing a "blatant" fraud. The articles are also reviewed twice by the journal, once to decide if they merit being sent out for review at all and once to make sure the journal has assembled everything correctly before it is printed. Again, nobody found anything worthy of question. I could add that after printing (20 years ago) this allegedly "blatant" altering of the images was not noted by the readers of the journal at the time. However, the quality (and size) of the printed reproductions is often such that it is hard to tell what they show :-).

As I understand it, there are three images that the detection software indicates are "suspicious". I don't know that the software can "prove" they were altered, but it can make a case that some pixel pattern is "extremely unlikely in nature". Two of these anomalies are in areas of their corresponding images that don't make any difference to the conclusions in the paper. Why someone would go to the trouble to make such an edit is unclear. Practicing perhaps? Cleaning up an imperfection of some kind? We don't know. The third anomaly is in are area of an image that "matters".  We don't know what the alteration actually changed (because we don't have the original), only that it was in an important area.

If the "charge" against MTL is that he (and his lab) did not adequately review the images and "spot" the editing (if it was editing), that editing needs to be "obvious" enough that people (like the reviewers) who have experience looking at such images would reasonably be expected to pick it up. I haven't seen any evidence this is the case, in fact what evidence we have points in the opposite direction.

I do agree that if the data was edited, MTL is "responsible" because his postdoc produced the data, whether he knew about it or not. But then, in the same vein, so are the reviewers, and so is the journal that published it. It would be real interesting to know if anybody cares enough about this to repeat the experiment and see what the blot should have shown. That result wouldn't "exonerate" anybody who was fooled by a doctored image, but it would potentially increase the culpability if the data was false. So far, AFAIK, nobody has owned up to altering the images. As I have said before, MTL is certainly responsible in some sense if the data was altered. However, if he did everything that one could reasonably expect him to do, i.e. he wasn't "sloppy" about accepting a blatant fraud, I can't hold him morally responsible.

[Goose]

Great reply jonnyss. Sums up my thinking as well. 

MTL, in the best light, is still responsible for a paper with fraudulent data, even if his postdoc did all the photo editing without MTL's knowledge or direction. The postdoc is his trainee, and direct report in business parlance. If he screws up, the professor is still responsible on some level for not properly monitoring or training that person.

And it happened multiple times.

There is one very good reason to believe that the doctoring of the MTL images was in fact NOT blatant, really hard to miss. As you undoubtedly know, there papers are peer reviewed before publication. NONE of the reviewers raised questions about these figures. If it was so obvious why didn't the reviewers see it? Clearly, the idea that it was obvious is not supported by facts.

Peer review is simply not designed to pick up things like this. It's done on a volunteer basis in the spare time of usually overworked scientists, who generally trust that people aren't lying about their actual data.

The peer review process could certainly use some changes but that's ancillary to the discussion of MTL's culpability.

If the doctoring was "blatant", anybody who looked at the images at all critically should have picked it up. I would argue peer review is indeed designed to pick up blatant, obvious fraud. I do agree that the bias of the reviewers is that the data is genuine, but when evaluating the culpability of MTL the assertion that the fraud was "blatant" certainly would greatly increase his culpability. I expect the bias of the PI is that his people's data is genuine as well, probably more so. The reviewer's bias if anything should be less. There are also at least two reviewers, so one reviewer doing a slipshod job wouldn't explain not seeing a "blatant" fraud. The articles are also reviewed twice by the journal, once to decide if they merit being sent out for review at all and once to make sure the journal has assembled everything correctly before it is printed. Again, nobody found anything worthy of question. I could add that after printing (20 years ago) this allegedly "blatant" altering of the images was not noted by the readers of the journal at the time. However, the quality (and size) of the printed reproductions is often such that it is hard to tell what they show :-).

As I understand it, there are three images that the detection software indicates are "suspicious". I don't know that the software can "prove" they were altered, but it can make a case that some pixel pattern is "extremely unlikely in nature". Two of these anomalies are in areas of their corresponding images that don't make any difference to the conclusions in the paper. Why someone would go to the trouble to make such an edit is unclear. Practicing perhaps? Cleaning up an imperfection of some kind? We don't know. The third anomaly is in are area of an image that "matters".  We don't know what the alteration actually changed (because we don't have the original), only that it was in an important area.

If the "charge" against MTL is that he (and his lab) did not adequately review the images and "spot" the editing (if it was editing), that editing needs to be "obvious" enough that people (like the reviewers) who have experience looking at such images would reasonably be expected to pick it up. I haven't seen any evidence this is the case, in fact what evidence we have points in the opposite direction.

I do agree that if the data was edited, MTL is "responsible" because his postdoc produced the data, whether he knew about it or not. But then, in the same vein, so are the reviewers, and so is the journal that published it. It would be real interesting to know if anybody cares enough about this to repeat the experiment and see what the blot should have shown. That result wouldn't "exonerate" anybody who was fooled by a doctored image, but it would potentially increase the culpability if the data was false. So far, AFAIK, nobody has owned up to altering the images. As I have said before, MTL is certainly responsible in some sense if the data was altered. However, if he did everything that one could reasonably expect him to do, i.e. he wasn't "sloppy" about accepting a blatant fraud, I can't hold him morally responsible.