[dabigv13]

“Based on the guide, had the research been conducted at SLAC I imagine we would already be processing MTL’s paperwork for dismissal for cause,” Chassin wrote. “The fact that his past research activities did not conform with Stanford’s standards at that time is the most relevant question in my mind, and these revelations should disqualify him from holding any current position at Stanford or SLAC. The credibility of the entire organization is at stake and it should be painfully obvious to all that his personal privilege is subordinated to that of the organization he leads.”

Amen.

I'm not as concerned as Bik is that a former DA is running the investigation, as long as they involve qualified independent scientists as well. This will be a complicated investigation and I think a DA has more relevant experience.

as long as the DA understands that the standard of academic integrity is very different from the standard a DA uses when considering whether a criminal case goes to trial. i would presume the typical DA would look at this at say, what's the fuss about?

Anyone with a law degree understands the difference between the standard for civil (preponderance of evidence) and criminal (reasonable doubt). I'm confident that someone who is a former DA has been given guidance from BOT as to what the evidentiary standard is for this investigation

When depriving someone of their liberty, the standard must be very high ("beyond.")  When depriving someone of their property a little less so ("preponderance.")  But when depriving someone of a highly paid, privileged position of largely symbolic leadership--failure to model the highest ethical standards should be enough for removal.  For example, the fact that papers went out under his name which contained doctored images is on the face of it enough to remove him, full stop.  Questions such as whether he was personally responsible for the manipulation of the images or whether he was aware of it would be relevant to a criminal proceeding or a lawsuit, but they are beside the point when judging his suitability to be in his position of leadership.  Our standard for leadership is higher than "he was only sloppy with his proofreading of his colleague's contributions, not deliberately dishonest."

Furthermore, put this in the context of the crisis in scientific publications, in which as many as 50% of published results are turning out not to be reproduceable.  Confidence in the whole scientific enterprise was being called into question before this scandal hit, and we needed Stanford and its leaders to actually lead the way to restoring that confidence.  If the President of Stanford is allowed to remain in office after it turns out he has been part of the problem, it is a terrible message to science and universities far beyond Stanford University.  It is as if to say that playing fast and loose with scientific pubs is just the nature of the game--absolutely everyone does it--and those who do it well get ahead.

Excellent post that summarizes better than I could what matters in all this. It's untenable for him to be an example of scientific leadership and it's baffling to me that apparently most of the BOT doesn't grasp that.

[dabigv13]

“Based on the guide, had the research been conducted at SLAC I imagine we would already be processing MTL’s paperwork for dismissal for cause,” Chassin wrote. “The fact that his past research activities did not conform with Stanford’s standards at that time is the most relevant question in my mind, and these revelations should disqualify him from holding any current position at Stanford or SLAC. The credibility of the entire organization is at stake and it should be painfully obvious to all that his personal privilege is subordinated to that of the organization he leads.”

Amen.

I'm not as concerned as Bik is that a former DA is running the investigation, as long as they involve qualified independent scientists as well. This will be a complicated investigation and I think a DA has more relevant experience.

as long as the DA understands that the standard of academic integrity is very different from the standard a DA uses when considering whether a criminal case goes to trial. i would presume the typical DA would look at this at say, what's the fuss about?

Anyone with a law degree understands the difference between the standard for civil (preponderance of evidence) and criminal (reasonable doubt). I'm confident that someone who is a former DA has been given guidance from BOT as to what the evidentiary standard is for this investigation

When depriving someone of their liberty, the standard must be very high ("beyond.")  When depriving someone of their property a little less so ("preponderance.")  But when depriving someone of a highly paid, privileged position of largely symbolic leadership--failure to model the highest ethical standards should be enough for removal.  For example, the fact that papers went out under his name which contained doctored images is on the face of it enough to remove him, full stop.  Questions such as whether he was personally responsible for the manipulation of the images or whether he was aware of it would be relevant to a criminal proceeding or a lawsuit, but they are beside the point when judging his suitability to be in his position of leadership.  Our standard for leadership is higher than "he was only sloppy with his proofreading of his colleague's contributions, not deliberately dishonest."

Furthermore, put this in the context of the crisis in scientific publications, in which as many as 50% of published results are turning out not to be reproduceable.  Confidence in the whole scientific enterprise was being called into question before this scandal hit, and we needed Stanford and its leaders to actually lead the way to restoring that confidence.  If the President of Stanford is allowed to remain in office after it turns out he has been part of the problem, it is a terrible message to science and universities far beyond Stanford University.  It is as if to say that playing fast and loose with scientific pubs is just the nature of the game--absolutely everyone does it--and those who do it well get ahead.

Excellent post that summarizes better than I could what matters in all this. It's untenable for him to be an example of scientific leadership and it's baffling to me that apparently most of the BOT doesn't grasp that.

[Goose]

Excellent post that summarizes better than I could what matters in all this. It's untenable for him to be an example of scientific leadership and it's baffling to me that apparently most of the BOT doesn't grasp that.

I guess my reaction is a bit different. Does anybody seriously believe that the senior author on the vast majority of scientific publications has verified the provenance of all (actually any of) the data contained therein? They are not in the lab physically watching all the experiments that are done. In many cases they are collaborating with colleagues that are not even at the same institution. Some of the larger labs have 70 or 80 graduate students who are working under supervision of post-docs that the PI actually doesn't know all that well. While some may operate in the fantasy world where all the PIs are intimately involved in every experiment and actually know what is going on in all aspects of their labs, that is simply not how it is. I think in many cases the best one can hope for is that the PI has actually read the paper before it was submitted :-).

It isn't a good system in many ways, but it is the system we have. It relies on the honesty of a lot of people. In point of fact, artifacts are sometimes the basis of some paper's data. There are so many papers published that contain data that will never be important enough to reproduce that many such thing go unnoticed for years. I think there are probably more honest errors than fraud. How many papers are published that are the result of something finally "working" after many failures? How many PIs insist on the data being reliably reproducible?

[Goose]

Excellent post that summarizes better than I could what matters in all this. It's untenable for him to be an example of scientific leadership and it's baffling to me that apparently most of the BOT doesn't grasp that.

I guess my reaction is a bit different. Does anybody seriously believe that the senior author on the vast majority of scientific publications has verified the provenance of all (actually any of) the data contained therein? They are not in the lab physically watching all the experiments that are done. In many cases they are collaborating with colleagues that are not even at the same institution. Some of the larger labs have 70 or 80 graduate students who are working under supervision of post-docs that the PI actually doesn't know all that well. While some may operate in the fantasy world where all the PIs are intimately involved in every experiment and actually know what is going on in all aspects of their labs, that is simply not how it is. I think in many cases the best one can hope for is that the PI has actually read the paper before it was submitted :-).

It isn't a good system in many ways, but it is the system we have. It relies on the honesty of a lot of people. In point of fact, artifacts are sometimes the basis of some paper's data. There are so many papers published that contain data that will never be important enough to reproduce that many such thing go unnoticed for years. I think there are probably more honest errors than fraud. How many papers are published that are the result of something finally "working" after many failures? How many PIs insist on the data being reliably reproducible?

[81alum]

I guess my reaction is a bit different. Does anybody seriously believe that the senior author on the vast majority of scientific publications has verified the provenance of all (actually any of) the data contained therein? They are not in the lab physically watching all the experiments that are done. In many cases they are collaborating with colleagues that are not even at the same institution. Some of the larger labs have 70 or 80 graduate students who are working under supervision of post-docs that the PI actually doesn't know all that well. While some may operate in the fantasy world where all the PIs are intimately involved in every experiment and actually know what is going on in all aspects of their labs, that is simply not how it is. I think in many cases the best one can hope for is that the PI has actually read the paper before it was submitted :-).

It isn't a good system in many ways, but it is the system we have. It relies on the honesty of a lot of people. In point of fact, artifacts are sometimes the basis of some paper's data. There are so many papers published that contain data that will never be important enough to reproduce that many such thing go unnoticed for years. I think there are probably more honest errors than fraud. How many papers are published that are the result of something finally "working" after many failures? How many PIs insist on the data being reliably reproducible?

Thoughtful comments as I have come to expect of you, Goose.  I think I might be inclined to reluctantly agree if we were judging the career of a research scientist qua research scientist.  But when one becomes the President of a leading research university the standards go up.  As you say, it is not a good system, and we need leadership to change it and improve it.  That leadership needs to start at the top--and there is nowhere in the world closer to the top than a President of a premier research university.  Instead, we have leadership that is, by example and association, affirming the slimier side of the current broken system.

[81alum]

I guess my reaction is a bit different. Does anybody seriously believe that the senior author on the vast majority of scientific publications has verified the provenance of all (actually any of) the data contained therein? They are not in the lab physically watching all the experiments that are done. In many cases they are collaborating with colleagues that are not even at the same institution. Some of the larger labs have 70 or 80 graduate students who are working under supervision of post-docs that the PI actually doesn't know all that well. While some may operate in the fantasy world where all the PIs are intimately involved in every experiment and actually know what is going on in all aspects of their labs, that is simply not how it is. I think in many cases the best one can hope for is that the PI has actually read the paper before it was submitted :-).

It isn't a good system in many ways, but it is the system we have. It relies on the honesty of a lot of people. In point of fact, artifacts are sometimes the basis of some paper's data. There are so many papers published that contain data that will never be important enough to reproduce that many such thing go unnoticed for years. I think there are probably more honest errors than fraud. How many papers are published that are the result of something finally "working" after many failures? How many PIs insist on the data being reliably reproducible?

Thoughtful comments as I have come to expect of you, Goose.  I think I might be inclined to reluctantly agree if we were judging the career of a research scientist qua research scientist.  But when one becomes the President of a leading research university the standards go up.  As you say, it is not a good system, and we need leadership to change it and improve it.  That leadership needs to start at the top--and there is nowhere in the world closer to the top than a President of a premier research university.  Instead, we have leadership that is, by example and association, affirming the slimier side of the current broken system.

[TreesAndBirds]

Excellent post that summarizes better than I could what matters in all this. It's untenable for him to be an example of scientific leadership and it's baffling to me that apparently most of the BOT doesn't grasp that.

I guess my reaction is a bit different. Does anybody seriously believe that the senior author on the vast majority of scientific publications has verified the provenance of all (actually any of) the data contained therein? They are not in the lab physically watching all the experiments that are done. In many cases they are collaborating with colleagues that are not even at the same institution. Some of the larger labs have 70 or 80 graduate students who are working under supervision of post-docs that the PI actually doesn't know all that well. While some may operate in the fantasy world where all the PIs are intimately involved in every experiment and actually know what is going on in all aspects of their labs, that is simply not how it is. I think in many cases the best one can hope for is that the PI has actually read the paper before it was submitted :-).

It isn't a good system in many ways, but it is the system we have. It relies on the honesty of a lot of people. In point of fact, artifacts are sometimes the basis of some paper's data. There are so many papers published that contain data that will never be important enough to reproduce that many such thing go unnoticed for years. I think there are probably more honest errors than fraud. How many papers are published that are the result of something finally "working" after many failures? How many PIs insist on the data being reliably reproducible?

I know virtually nothing about how university research is conducted. But I do know that culture and values are set at the top. Signals are sent, sometimes subtlely, sometimes not, about what is most important in an organization or group. Things like profitability, meeting deadlines and carefulness/quality are often in conflict in the business world. The leader sets the tone on what to prioritize. Yes, the leader can not know everything that more junior people or colleagues are doing, but with the right messaging it seems unlikely that a disproportionate number of these errors would occur (assuming they were disproportionate). I have no view on whether anything was intentional, or whether MTL should have been more careful in reviewing these papers, but as President of a university setting the right culture and expectations is really important, and Im concerned that MTL has demonstrated weakness in that area.

[TreesAndBirds]

Excellent post that summarizes better than I could what matters in all this. It's untenable for him to be an example of scientific leadership and it's baffling to me that apparently most of the BOT doesn't grasp that.

I guess my reaction is a bit different. Does anybody seriously believe that the senior author on the vast majority of scientific publications has verified the provenance of all (actually any of) the data contained therein? They are not in the lab physically watching all the experiments that are done. In many cases they are collaborating with colleagues that are not even at the same institution. Some of the larger labs have 70 or 80 graduate students who are working under supervision of post-docs that the PI actually doesn't know all that well. While some may operate in the fantasy world where all the PIs are intimately involved in every experiment and actually know what is going on in all aspects of their labs, that is simply not how it is. I think in many cases the best one can hope for is that the PI has actually read the paper before it was submitted :-).

It isn't a good system in many ways, but it is the system we have. It relies on the honesty of a lot of people. In point of fact, artifacts are sometimes the basis of some paper's data. There are so many papers published that contain data that will never be important enough to reproduce that many such thing go unnoticed for years. I think there are probably more honest errors than fraud. How many papers are published that are the result of something finally "working" after many failures? How many PIs insist on the data being reliably reproducible?

I know virtually nothing about how university research is conducted. But I do know that culture and values are set at the top. Signals are sent, sometimes subtlely, sometimes not, about what is most important in an organization or group. Things like profitability, meeting deadlines and carefulness/quality are often in conflict in the business world. The leader sets the tone on what to prioritize. Yes, the leader can not know everything that more junior people or colleagues are doing, but with the right messaging it seems unlikely that a disproportionate number of these errors would occur (assuming they were disproportionate). I have no view on whether anything was intentional, or whether MTL should have been more careful in reviewing these papers, but as President of a university setting the right culture and expectations is really important, and Im concerned that MTL has demonstrated weakness in that area.

[Goose]

But when one becomes the President of a leading research university the standards go up.  As you say, it is not a good system, and we need leadership to change it and improve it.  That leadership needs to start at the top--and there is nowhere in the world closer to the top than a President of a premier research university.  Instead, we have leadership that is, by example and association, affirming the slimier side of the current broken system.

I agree that those in leadership positions have to aspire to a "higher standard". The problem is that as far as I know nobody has found a way that all the author's on a paper can be "sure" all the data is correct unless they themselves physically verified all the experiments in person. Doing so would bring research to a standstill. This problem is further exacerbated by the fact the number of publications as either a first or last author is used as a metric of "success". This is the kind of success that gets you "important" jobs. In addition, being the first author on a paper is often considered necessary to "meaningfully" complete your PhD (even if the paper is published in a third-tier journal). I am not sure how this system can be "reformed" and still function. The only thing I can suggest is post warning labels on all journal articles saying the authors cannot vouch for all the data contained therein, which isn't exactly a solution :-). Given that, while MTL falling on his sword may be a nice gesture, I am not sure it would really change things one iota.

Reminds me of the current state of college football (except in slow motion).

[Goose]

But when one becomes the President of a leading research university the standards go up.  As you say, it is not a good system, and we need leadership to change it and improve it.  That leadership needs to start at the top--and there is nowhere in the world closer to the top than a President of a premier research university.  Instead, we have leadership that is, by example and association, affirming the slimier side of the current broken system.

I agree that those in leadership positions have to aspire to a "higher standard". The problem is that as far as I know nobody has found a way that all the author's on a paper can be "sure" all the data is correct unless they themselves physically verified all the experiments in person. Doing so would bring research to a standstill. This problem is further exacerbated by the fact the number of publications as either a first or last author is used as a metric of "success". This is the kind of success that gets you "important" jobs. In addition, being the first author on a paper is often considered necessary to "meaningfully" complete your PhD (even if the paper is published in a third-tier journal). I am not sure how this system can be "reformed" and still function. The only thing I can suggest is post warning labels on all journal articles saying the authors cannot vouch for all the data contained therein, which isn't exactly a solution :-). Given that, while MTL falling on his sword may be a nice gesture, I am not sure it would really change things one iota.

Reminds me of the current state of college football (except in slow motion).

[jonnyss]

Excellent post that summarizes better than I could what matters in all this. It's untenable for him to be an example of scientific leadership and it's baffling to me that apparently most of the BOT doesn't grasp that.

I guess my reaction is a bit different. Does anybody seriously believe that the senior author on the vast majority of scientific publications has verified the provenance of all (actually any of) the data contained therein? They are not in the lab physically watching all the experiments that are done. In many cases they are collaborating with colleagues that are not even at the same institution. Some of the larger labs have 70 or 80 graduate students who are working under supervision of post-docs that the PI actually doesn't know all that well. While some may operate in the fantasy world where all the PIs are intimately involved in every experiment and actually know what is going on in all aspects of their labs, that is simply not how it is. I think in many cases the best one can hope for is that the PI has actually read the paper before it was submitted :-).

It isn't a good system in many ways, but it is the system we have. It relies on the honesty of a lot of people. In point of fact, artifacts are sometimes the basis of some paper's data. There are so many papers published that contain data that will never be important enough to reproduce that many such thing go unnoticed for years. I think there are probably more honest errors than fraud. How many papers are published that are the result of something finally "working" after many failures? How many PIs insist on the data being reliably reproducible?

sure, unnoticed by some busy scientists. but, if i understand the chronology, mtl had previously self-reported to a journal errors in images in one of his articles. i would certainly expect him to be vigilant for such errors in the future. that he was not seems quite problematic. to me, it looks like, when the first journal said, "don't worry about it," he felt emboldened either to allow falsification in the future or at least to look the other way.

elizabeth holmes also tried the "someone else was responsible for data integrity" strategy.

[jonnyss]

Excellent post that summarizes better than I could what matters in all this. It's untenable for him to be an example of scientific leadership and it's baffling to me that apparently most of the BOT doesn't grasp that.

I guess my reaction is a bit different. Does anybody seriously believe that the senior author on the vast majority of scientific publications has verified the provenance of all (actually any of) the data contained therein? They are not in the lab physically watching all the experiments that are done. In many cases they are collaborating with colleagues that are not even at the same institution. Some of the larger labs have 70 or 80 graduate students who are working under supervision of post-docs that the PI actually doesn't know all that well. While some may operate in the fantasy world where all the PIs are intimately involved in every experiment and actually know what is going on in all aspects of their labs, that is simply not how it is. I think in many cases the best one can hope for is that the PI has actually read the paper before it was submitted :-).

It isn't a good system in many ways, but it is the system we have. It relies on the honesty of a lot of people. In point of fact, artifacts are sometimes the basis of some paper's data. There are so many papers published that contain data that will never be important enough to reproduce that many such thing go unnoticed for years. I think there are probably more honest errors than fraud. How many papers are published that are the result of something finally "working" after many failures? How many PIs insist on the data being reliably reproducible?

sure, unnoticed by some busy scientists. but, if i understand the chronology, mtl had previously self-reported to a journal errors in images in one of his articles. i would certainly expect him to be vigilant for such errors in the future. that he was not seems quite problematic. to me, it looks like, when the first journal said, "don't worry about it," he felt emboldened either to allow falsification in the future or at least to look the other way.

elizabeth holmes also tried the "someone else was responsible for data integrity" strategy.

[BostonCard]

MTL reported all three of the known issues in the articles he was a senior author on. Cell told him “no big deal” and didn’t bother with a correction.  Science apparently accepted corrections to the two articles he had with them, but due to an issue at the journal, never published the errata.

I do wonder if this scandal would be a lot different if those errata had been published.

BC

[BostonCard]

MTL reported all three of the known issues in the articles he was a senior author on. Cell told him “no big deal” and didn’t bother with a correction.  Science apparently accepted corrections to the two articles he had with them, but due to an issue at the journal, never published the errata.

I do wonder if this scandal would be a lot different if those errata had been published.

BC

[Goose]

sure, unnoticed by some busy scientists. but, if i understand the chronology, mtl had previously self-reported to a journal errors in images in one of his articles. i would certainly expect him to be vigilant for such errors in the future. that he was not seems quite problematic. to me, it looks like, when the first journal said, "don't worry about it," he felt emboldened either to allow falsification in the future or at least to look the other way.

IMO the conclusion that he felt any way at all about it is (as of yet) unsupported by evidence. My inclination is to say that in neither case did he know there were any "irregularities" in the images until this was noted by others. As of yet, I know of no evidence that he knew about it before the articles were published, but we may find out otherwise. Right now, the worst that can be charged is not being adequately diligent in ensuring the data integrity of an article on which he was the last author. My point is that virtually the same charge can be levied against almost all such last authors, and to the extent they have not signed off on data that was in some way "altered", it is mostly luck, not diligence.

elizabeth holmes also tried the "someone else was responsible for data integrity" strategy.

Totally and completely different cases. Elizabeth Holmes clearly knew that their data did not indicate their machines "worked". She actively tried to suppress that knowledge. She also profited handsomely for so doing. It also wasn't just a few figures that were "doctored". It was the entire operational concept of their product line that simply didn't work, which she unarguably knew.

[Goose]

sure, unnoticed by some busy scientists. but, if i understand the chronology, mtl had previously self-reported to a journal errors in images in one of his articles. i would certainly expect him to be vigilant for such errors in the future. that he was not seems quite problematic. to me, it looks like, when the first journal said, "don't worry about it," he felt emboldened either to allow falsification in the future or at least to look the other way.

IMO the conclusion that he felt any way at all about it is (as of yet) unsupported by evidence. My inclination is to say that in neither case did he know there were any "irregularities" in the images until this was noted by others. As of yet, I know of no evidence that he knew about it before the articles were published, but we may find out otherwise. Right now, the worst that can be charged is not being adequately diligent in ensuring the data integrity of an article on which he was the last author. My point is that virtually the same charge can be levied against almost all such last authors, and to the extent they have not signed off on data that was in some way "altered", it is mostly luck, not diligence.

elizabeth holmes also tried the "someone else was responsible for data integrity" strategy.

Totally and completely different cases. Elizabeth Holmes clearly knew that their data did not indicate their machines "worked". She actively tried to suppress that knowledge. She also profited handsomely for so doing. It also wasn't just a few figures that were "doctored". It was the entire operational concept of their product line that simply didn't work, which she unarguably knew.

[martyup]

I'm no scientist and I don't have familiarity with the subject or contents of these papers.  My layman question is did the altered images give a false conclusion to the science?  In other words, if the images would not have been altered would the conclusions be the same?

[martyup]

I'm no scientist and I don't have familiarity with the subject or contents of these papers.  My layman question is did the altered images give a false conclusion to the science?  In other words, if the images would not have been altered would the conclusions be the same?

[murky]

I'm sympathetic to Goose's arguments and have had similar thoughts.  

I lead a team of engineers in product development, and see some parallels.  It's untenable to personally review everything.   I need to put trust in judgment and integrity of team members.

[murky]

I'm sympathetic to Goose's arguments and have had similar thoughts.  

I lead a team of engineers in product development, and see some parallels.  It's untenable to personally review everything.   I need to put trust in judgment and integrity of team members.

[washingtonismoney]

I know virtually nothing about how university research is conducted. But I do know that culture and values are set at the top. Signals are sent, sometimes subtlely, sometimes not, about what is most important in an organization or group. Things like profitability, meeting deadlines and carefulness/quality are often in conflict in the business world. The leader sets the tone on what to prioritize. Yes, the leader can not know everything that more junior people or colleagues are doing, but with the right messaging it seems unlikely that a disproportionate number of these errors would occur (assuming they were disproportionate). I have no view on whether anything was intentional, or whether MTL should have been more careful in reviewing these papers, but as President of a university setting the right culture and expectations is really important, and Im concerned that MTL has demonstrated weakness in that area.

To wit, one telling detail for me is that corrections were submitted, but not followed up upon, for years, if I'm recalling the sequence correctly. Speaking of my own business, you pretty much floor it until that correction gets into the paper.

[washingtonismoney]

I know virtually nothing about how university research is conducted. But I do know that culture and values are set at the top. Signals are sent, sometimes subtlely, sometimes not, about what is most important in an organization or group. Things like profitability, meeting deadlines and carefulness/quality are often in conflict in the business world. The leader sets the tone on what to prioritize. Yes, the leader can not know everything that more junior people or colleagues are doing, but with the right messaging it seems unlikely that a disproportionate number of these errors would occur (assuming they were disproportionate). I have no view on whether anything was intentional, or whether MTL should have been more careful in reviewing these papers, but as President of a university setting the right culture and expectations is really important, and Im concerned that MTL has demonstrated weakness in that area.

To wit, one telling detail for me is that corrections were submitted, but not followed up upon, for years, if I'm recalling the sequence correctly. Speaking of my own business, you pretty much floor it until that correction gets into the paper.

[dabigv13]

I'm no scientist and I don't have familiarity with the subject or contents of these papers.  My layman question is did the altered images give a false conclusion to the science?  In other words, if the images would not have been altered would the conclusions be the same?

In at least one case, yes. 

It's fine for MTL to continue to be a researcher and professor going forward even if he a) photoshopped b) told someone else to photoshop or c) didn't notice when other people photoshopped, several times. But not to be president of Stanford. He has to be a leader and a leader sets an example.

It would not be difficult for Stanford to find another excellent academician to be president whose name is not on falsified papers, who would do a fine job of being president. MTL is not irreplaceable.

[dabigv13]

I'm no scientist and I don't have familiarity with the subject or contents of these papers.  My layman question is did the altered images give a false conclusion to the science?  In other words, if the images would not have been altered would the conclusions be the same?

In at least one case, yes. 

It's fine for MTL to continue to be a researcher and professor going forward even if he a) photoshopped b) told someone else to photoshop or c) didn't notice when other people photoshopped, several times. But not to be president of Stanford. He has to be a leader and a leader sets an example.

It would not be difficult for Stanford to find another excellent academician to be president whose name is not on falsified papers, who would do a fine job of being president. MTL is not irreplaceable.

[Goose]

I'm no scientist and I don't have familiarity with the subject or contents of these papers.  My layman question is did the altered images give a false conclusion to the science?  In other words, if the images would not have been altered would the conclusions be the same?

In at least one case, yes. 

It's fine for MTL to continue to be a researcher and professor going forward even if he a) photoshopped b) told someone else to photoshop or c) didn't notice when other people photoshopped, several times. But not to be president of Stanford. He has to be a leader and a leader sets an example.

It would not be difficult for Stanford to find another excellent academician to be president whose name is not on falsified papers, who would do a fine job of being president. MTL is not irreplaceable.

I agree with your conclusion that MTL is not irreplaceable. In fact, I think that there are good reasons to question his leadership. That said, I most emphatically do not agree that it would be acceptable for him to remain a professor going forward if he photoshopped data or suggested that somebody else photoshop data. Those would be intentional breaches of ethics for which he would be responsible. Those things can and should get you dismissed with prejudice from the faculty.

Your characterization of "not noticing" that something was photoshopped is IMHO misleading. It is not as if MTL had seen the "real" data, the data in the paper was "different", and he should have detected that difference. As far as we know, the only thing he ever saw was the "tainted" data. I find it difficult to actually fault him much if that were the case. I do understand the argument that the President of the University must be above reproach, and I could even understand the argument that the President, much as the captain of a ship, can be dismissed for misdeeds that he could not have reasonably prevented. I just can't consider him morally responsible for such a situation.

[Goose]

I'm no scientist and I don't have familiarity with the subject or contents of these papers.  My layman question is did the altered images give a false conclusion to the science?  In other words, if the images would not have been altered would the conclusions be the same?

In at least one case, yes. 

It's fine for MTL to continue to be a researcher and professor going forward even if he a) photoshopped b) told someone else to photoshop or c) didn't notice when other people photoshopped, several times. But not to be president of Stanford. He has to be a leader and a leader sets an example.

It would not be difficult for Stanford to find another excellent academician to be president whose name is not on falsified papers, who would do a fine job of being president. MTL is not irreplaceable.

I agree with your conclusion that MTL is not irreplaceable. In fact, I think that there are good reasons to question his leadership. That said, I most emphatically do not agree that it would be acceptable for him to remain a professor going forward if he photoshopped data or suggested that somebody else photoshop data. Those would be intentional breaches of ethics for which he would be responsible. Those things can and should get you dismissed with prejudice from the faculty.

Your characterization of "not noticing" that something was photoshopped is IMHO misleading. It is not as if MTL had seen the "real" data, the data in the paper was "different", and he should have detected that difference. As far as we know, the only thing he ever saw was the "tainted" data. I find it difficult to actually fault him much if that were the case. I do understand the argument that the President of the University must be above reproach, and I could even understand the argument that the President, much as the captain of a ship, can be dismissed for misdeeds that he could not have reasonably prevented. I just can't consider him morally responsible for such a situation.

[BostonCard]

Agree with Goose. If the post-doc did this without his knowledge, it’s a forgivable breach of oversight, but, to be honest, even the most diligent PI is not going to run forensic software on everything their grad students and post-docs work on.  If he had knowledge or instigated it, he should be done for, period.

That’s why the investigation is so important.

WIM, FYI, an academic journal is a bit different than most, because the paper authors don’t have editorial control.  Errata are few and far between, mostly because people publish and move on.  The fact that MTL notified the journal and submitted an erratum is a significant mitigating factor.  That he didn’t bug the journal about its status is a non-issue.

BC

[BostonCard]

Agree with Goose. If the post-doc did this without his knowledge, it’s a forgivable breach of oversight, but, to be honest, even the most diligent PI is not going to run forensic software on everything their grad students and post-docs work on.  If he had knowledge or instigated it, he should be done for, period.

That’s why the investigation is so important.

WIM, FYI, an academic journal is a bit different than most, because the paper authors don’t have editorial control.  Errata are few and far between, mostly because people publish and move on.  The fact that MTL notified the journal and submitted an erratum is a significant mitigating factor.  That he didn’t bug the journal about its status is a non-issue.

BC

[jonnyss]

so what do folks think of chimera's post that, when she worked in a similar lab, the pi and junior scientists and techs met for hours looking at their images and debating whether the results were significant? 

these images are not like graphs or figures representing the data. these are the data.

i, for one, trust her personal experience that neuroscientists don't just publish these images without reviewing them very very carefully.

[jonnyss]

so what do folks think of chimera's post that, when she worked in a similar lab, the pi and junior scientists and techs met for hours looking at their images and debating whether the results were significant? 

these images are not like graphs or figures representing the data. these are the data.

i, for one, trust her personal experience that neuroscientists don't just publish these images without reviewing them very very carefully.

[dabigv13]

I worked in the McConnell Lab at Stanford, a developmental neurobiology lab, for a couple years as an undergraduate.

The sorts of data that ends up as figures published in a paper (especially a Cell or Science type paper) would typically have been looked at in lab meetings over a good period of time and reviewed by a lot of folks in the lab, not just the authors. Not always but I would say usually.

Some labs that are collaborating across continents or institutions, maybe not all the individual pictures get analyzed closely by each author. But a 1st author, senior author, should have a very close handle on all figures.

[dabigv13]

I worked in the McConnell Lab at Stanford, a developmental neurobiology lab, for a couple years as an undergraduate.

The sorts of data that ends up as figures published in a paper (especially a Cell or Science type paper) would typically have been looked at in lab meetings over a good period of time and reviewed by a lot of folks in the lab, not just the authors. Not always but I would say usually.

Some labs that are collaborating across continents or institutions, maybe not all the individual pictures get analyzed closely by each author. But a 1st author, senior author, should have a very close handle on all figures.

[Goose]

so what do folks think of chimera's post that, when she worked in a similar lab, the pi and junior scientists and techs met for hours looking at their images and debating whether the results were significant? 

these images are not like graphs or figures representing the data. these are the data.

i, for one, trust her personal experience that neuroscientists don't just publish these images without reviewing them very very carefully.

As I understand it, the claims made in the paper are supported by the images. That isn't the problem. The problem is that the images are not totally "original", and in one case the area that is altered is in a location that may have contained a band indicating the presence of a protein that would contradict the paper's conclusions. Reviewing that image very, very carefully would have led to (and did support) the paper's conclusions. Only if one examined the image with special software would it's provenance been in question. It is also probable at the time this data was published that software didn't yet exist. Was there a band there? We don't know.

If one looks at the lab notebooks kept by most graduate students, there are lots of things missing that should be there. They still get signed. Even if one looks at the lab notebooks of most biotech companies there are plenty of things not done perfectly or completely. Only things that are going to be externally audited are going to be a priority to get right. It is also often true that the things one wants to know retrospectively about a certain experiment were not recorded because it wasn't at all obvious they mattered. People end up interrogating the technicians hoping they recall the missing pieces. These things happen even when everybody is doing their best. If somebody is photoshopping, the odds of getting caught aren't all that great.

[Goose]

so what do folks think of chimera's post that, when she worked in a similar lab, the pi and junior scientists and techs met for hours looking at their images and debating whether the results were significant? 

these images are not like graphs or figures representing the data. these are the data.

i, for one, trust her personal experience that neuroscientists don't just publish these images without reviewing them very very carefully.

As I understand it, the claims made in the paper are supported by the images. That isn't the problem. The problem is that the images are not totally "original", and in one case the area that is altered is in a location that may have contained a band indicating the presence of a protein that would contradict the paper's conclusions. Reviewing that image very, very carefully would have led to (and did support) the paper's conclusions. Only if one examined the image with special software would it's provenance been in question. It is also probable at the time this data was published that software didn't yet exist. Was there a band there? We don't know.

If one looks at the lab notebooks kept by most graduate students, there are lots of things missing that should be there. They still get signed. Even if one looks at the lab notebooks of most biotech companies there are plenty of things not done perfectly or completely. Only things that are going to be externally audited are going to be a priority to get right. It is also often true that the things one wants to know retrospectively about a certain experiment were not recorded because it wasn't at all obvious they mattered. People end up interrogating the technicians hoping they recall the missing pieces. These things happen even when everybody is doing their best. If somebody is photoshopping, the odds of getting caught aren't all that great.

[dabigv13]

As I understand it, the claims made in the paper are supported by the images. That isn't the problem. The problem is that the images are not totally "original", and in one case the area that is altered is in a location that may have contained a band indicating the presence of a protein that would contradict the paper's conclusions. Reviewing that image very, very carefully would have led to (and did support) the paper's conclusions. Only if one examined the image with special software would it's provenance been in question. It is also probable at the time this data was published that software didn't yet exist. Was there a band there? We don't know.

Not really sure what you're saying here. But some important things A) even "cleaning up" images is bad, important to understand. Not behavior to emulate. And B) in the Cell paper, some conclusions rest on the presence of a certainly falsified band on the Western blot. He was senior author.

Last week, Bik found another manipulated image in that paper, which she says appears to be “deliberately changing the results” and could indicate data falsification. Cell Press said the “new concerns … warrant a closer look” and it now plans to investigate. The only authors on all three challenged papers are Tessier-Lavigne and a postdoc then at UCSF who apparently hasn’t published a paper since 2012. She did not respond to emails from Science.

From above linked Science editorial.

MTL is not innocent here, that much is really beyond dispute. It's just a matter of how guilty.

[dabigv13]

As I understand it, the claims made in the paper are supported by the images. That isn't the problem. The problem is that the images are not totally "original", and in one case the area that is altered is in a location that may have contained a band indicating the presence of a protein that would contradict the paper's conclusions. Reviewing that image very, very carefully would have led to (and did support) the paper's conclusions. Only if one examined the image with special software would it's provenance been in question. It is also probable at the time this data was published that software didn't yet exist. Was there a band there? We don't know.

Not really sure what you're saying here. But some important things A) even "cleaning up" images is bad, important to understand. Not behavior to emulate. And B) in the Cell paper, some conclusions rest on the presence of a certainly falsified band on the Western blot. He was senior author.

Last week, Bik found another manipulated image in that paper, which she says appears to be “deliberately changing the results” and could indicate data falsification. Cell Press said the “new concerns … warrant a closer look” and it now plans to investigate. The only authors on all three challenged papers are Tessier-Lavigne and a postdoc then at UCSF who apparently hasn’t published a paper since 2012. She did not respond to emails from Science.

From above linked Science editorial.

MTL is not innocent here, that much is really beyond dispute. It's just a matter of how guilty.

[winflop]

I'm sympathetic to Goose's arguments and have had similar thoughts.  

I lead a team of engineers in product development, and see some parallels.  It's untenable to personally review everything.   I need to put trust in judgment and integrity of team members.

OK, but if one of them screws up, aren't you still accountable for that?

[winflop]

I'm sympathetic to Goose's arguments and have had similar thoughts.  

I lead a team of engineers in product development, and see some parallels.  It's untenable to personally review everything.   I need to put trust in judgment and integrity of team members.

OK, but if one of them screws up, aren't you still accountable for that?

[BostonCard]

I'm sympathetic to Goose's arguments and have had similar thoughts.  

I lead a team of engineers in product development, and see some parallels.  It's untenable to personally review everything.   I need to put trust in judgment and integrity of team members.

OK, but if one of them screws up, aren't you still accountable for that?

Yes, but it is a different level if you are complicit than if one of your reports does it without your knowledge.  MTL has probably had between 50 and 100 grad students and post docs in his lab over his career.

But all that is predicated on the theory that there was one bad apple who was responsible.  We will see what the investigation shows.

BC

[BostonCard]

I'm sympathetic to Goose's arguments and have had similar thoughts.  

I lead a team of engineers in product development, and see some parallels.  It's untenable to personally review everything.   I need to put trust in judgment and integrity of team members.

OK, but if one of them screws up, aren't you still accountable for that?

Yes, but it is a different level if you are complicit than if one of your reports does it without your knowledge.  MTL has probably had between 50 and 100 grad students and post docs in his lab over his career.

But all that is predicated on the theory that there was one bad apple who was responsible.  We will see what the investigation shows.

BC