Ivan Oransky

So when a paper is retracted, and it's probably worth defining that, a retraction is a signal to the scientific community or really to any readers of a particular paper. peer-reviewed journal article, that you should not rely on that anymore, that there's something about it that means you should, you know, you can not pretend it doesn't exist, but you shouldn't base any other work on it.

Right. So when you retract a paper, you're supposed to put a retraction notice on it, the same way when you correct an article in the newspaper, you're supposed to put a correction notice on it. But when you actually read these retraction notices, and to be fair, this has changed a fair amount in the 13 years that we've been doing this, sometimes they include no information at all.

Right. So when you retract a paper, you're supposed to put a retraction notice on it, the same way when you correct an article in the newspaper, you're supposed to put a correction notice on it. But when you actually read these retraction notices, and to be fair, this has changed a fair amount in the 13 years that we've been doing this, sometimes they include no information at all.

Right. So when you retract a paper, you're supposed to put a retraction notice on it, the same way when you correct an article in the newspaper, you're supposed to put a correction notice on it. But when you actually read these retraction notices, and to be fair, this has changed a fair amount in the 13 years that we've been doing this, sometimes they include no information at all.

Sometimes they include information that is woefully incomplete. Sometimes it's some version of you know, getting Al Capone on tax evasion. You know, they fake the data, but we're going to say they forgot to fill out this form, which, you know, is still a reason to retract, but isn't the whole story.

Sometimes they include information that is woefully incomplete. Sometimes it's some version of you know, getting Al Capone on tax evasion. You know, they fake the data, but we're going to say they forgot to fill out this form, which, you know, is still a reason to retract, but isn't the whole story.

Sometimes they include information that is woefully incomplete. Sometimes it's some version of you know, getting Al Capone on tax evasion. You know, they fake the data, but we're going to say they forgot to fill out this form, which, you know, is still a reason to retract, but isn't the whole story.

So we think that... Probably 2% of papers should be retracted for something that would be considered either out-and-out fraud or maybe just severe bad mistake. According to our data, which we have the most retraction data of any database, about 0.1% of the world's literature is retracted. So one in a thousand papers. We think it should be about 20 times that, about 2%.

So we think that... Probably 2% of papers should be retracted for something that would be considered either out-and-out fraud or maybe just severe bad mistake. According to our data, which we have the most retraction data of any database, about 0.1% of the world's literature is retracted. So one in a thousand papers. We think it should be about 20 times that, about 2%.

So we think that... Probably 2% of papers should be retracted for something that would be considered either out-and-out fraud or maybe just severe bad mistake. According to our data, which we have the most retraction data of any database, about 0.1% of the world's literature is retracted. So one in a thousand papers. We think it should be about 20 times that, about 2%.

There's a bunch of reasons, but they come down to one. There was a survey back in 2009, which has been repeated and done differently and come up with roughly the same number, actually even higher numbers recently, that says 2% of researchers, if you ask them anonymously, they will say, yes, I've committed something that would be considered misconduct.

There's a bunch of reasons, but they come down to one. There was a survey back in 2009, which has been repeated and done differently and come up with roughly the same number, actually even higher numbers recently, that says 2% of researchers, if you ask them anonymously, they will say, yes, I've committed something that would be considered misconduct.

There's a bunch of reasons, but they come down to one. There was a survey back in 2009, which has been repeated and done differently and come up with roughly the same number, actually even higher numbers recently, that says 2% of researchers, if you ask them anonymously, they will say, yes, I've committed something that would be considered misconduct.

Of course, when you ask them how many people they know who have committed misconduct, it goes much, much higher than that. And so that's one line of evidence, which is, you know, admittedly indirect.

Of course, when you ask them how many people they know who have committed misconduct, it goes much, much higher than that. And so that's one line of evidence, which is, you know, admittedly indirect.

Of course, when you ask them how many people they know who have committed misconduct, it goes much, much higher than that. And so that's one line of evidence, which is, you know, admittedly indirect.

The other is that when you talk to the sleuths, the people doing the real work of figuring out what's wrong with literature and letting people know about it, they keep lists of papers they've flagged for publishers and for authors and journals. And routinely, most of them are not retracted. Again, we came to 2%. Is it exactly 2%? And is that even the right number?

The other is that when you talk to the sleuths, the people doing the real work of figuring out what's wrong with literature and letting people know about it, they keep lists of papers they've flagged for publishers and for authors and journals. And routinely, most of them are not retracted. Again, we came to 2%. Is it exactly 2%? And is that even the right number?

The other is that when you talk to the sleuths, the people doing the real work of figuring out what's wrong with literature and letting people know about it, they keep lists of papers they've flagged for publishers and for authors and journals. And routinely, most of them are not retracted. Again, we came to 2%. Is it exactly 2%? And is that even the right number?

No, we're pretty sure that's the lower bound. Others say it should be even higher.