Monday, May 4, 2015

More on PLoS reviewing from Ewan

Here is an update from Ewan. Thx.


Quick update on PLoS.

First off, they've responded (not directly to our letter, which you still have a chance to sign, but to the bevy of similar comments on the internet):

        LINK HERE

They've given that editor the boot, removed the reviewer from their database, sent the paper to another editor for re-review, and removed the review from the record (presumably/hopefully this just means they've removed it from having any bearing on accept/reject for the submission, not that they've erased it from history). They explicitly acknowledge the distress this caused the authors and that the review contained objectionable language, and they assert their commitment to "fair," "civil," "constructive," and "unprejudiced" review processes.

And they are reviewing their review process, with the main direction they're going seemingly being ways of releasing the reviewer identities to the authors.

So if you haven't already signed the letter, which is partly written as a letter of thanks for this process which we knew they'd already started, please do so.

        LINK HERE

Second, early on as this was circulating, I got a couple of comments that it was strange that we brought up that the reviewer had looked at the authors websites, since after all this is a common practice, and there might even be legitimate reasons to do this. It's been tweaked now in a way that's consistent with the original but doesn't make you feel guilty for Googling anyone. (I've done this too, and I still think that in the majority of cases there probably wasn't a good reason, and I could well have been biased in the end. But, being pretty explicitly pro-hypocrisy, saying I was against this didn't bother me.)

The reason for including this in the first place was that the authors had written that they weren't comfortable with the fact that the reviewer was able to confirm that they were women this way, and, especially, that the reviewer was able to confirm that they were junior, and make disparaging comments about that. It seemed pretty clear to me that this ought to be unacceptable and was a commonsense argument for a veil of ignorance.

But, in the context of this letter, which is just trying to thank PLoS and get them to make sure they engaged with all the issues this raised, it was a bit out of place. Now it simply says that using personal identity as a review criterion isn't okay, which I think anyone can agree with, and which at least hints at why the problems aren't trivial.

Finally, some thoughts of my own on the direction that PLoS seems to be going, and why I think it's still important to push them farther during their ongoing changes. Exposing reviewers publicly - for example by making them sign their reviews - and even having further mechanisms for holding them accountable - really isn't the same as protecting rights to fair treatment.

This might not be obvious, but I think it would be to anyone who's had to seriously deal with problems of harassment in online communities. Here's a comment thread from the Rust programming language community (i.e., their online forums where people discuss programming issues) that exposes a similar discussion to one we might have about peer review:

        LINK HERE

The gist is that, while there are people, vaguely well-meaning, who suggest that the solution is to have "many eyes" on commenters, and free commentary so that the good can drive out the bad, the bottom line is this:

"Even EFF [Electronic Frontier Foundation], free-speech champion, feels that this is a nuanced topic requiring a mixture of technical and social responses, not a 'just add more speech' free-for-all."

By "technical responses," they mean having rules and preventive measures that administered top-down, rather than just expecting that, if everything is open to the public, bad behavior will stop. In the context of reviewer accountability, that would suggest that there need to be consequences for a bad reviewer beyond just public embarrassment, and a mechanism for enforcing those consequences, beyond just having an editorial hierarchy with the power to step in in egregious cases. They need to know exactly what to do, be ready and willing to do it, and anyone signaling problems has to be able to do so freely and without fear.

What's been found further online is that there's a real resistance amongst those charged with reining in harassment to get really specific about what is and isn't okay. But hoping for people, even moderators (read: editors) to have "common sense" just doesn't work. It's too easy to lean with your bias on marginal cases, or to not recognize when other people are feeling harassed. Successful codes of conduct for behaviour are highly specific.

        LINK HERE

The existence only of a vague commitment to "fair and civil treatment" is arguably precisely why the editor failed to stop this. A serious approach would be like the Typelevel code of conduct. It would say, "Harassment in reviews includes: ... " . In fact, it's worse, at PLoS ONE and at most journals. When I've reviewed for this or any other journal, I don't generally get anything that warns me against unfair and uncivil treatment. Not having a specific code of conduct in giant bold letters has, in my opinion, no excuse. (PLoS ONE has a statement that says that the journal is committed to "rigorous review," and that's it. To be fair, PLoS ONE still has an extremely forward-thinking set of review criteria, in which vague things like "novelty" are supposed to go out the window, and you're meant to consider only the technical soundness of the work. This incident suggests not many people read that, including the editor responsible for this.)

These sorts of considerations are also exactly why people like double blind reviews. At the end of the day, we have unconscious biases that are sometimes too subtle to recognize, and sometimes we wind up being uncertain about what's okay and what's not.

I can see why people think double blind review is pointless in the era of the internet. While I don't personally agree with that, I'm guessing that there probably isn't going to be a huge wave of journals that never had going ahead with adopting it. But, particularly if that's going out the window, I think we have to at least recognize that crossing our fingers for good behavior - either in reviewers' unconscious evaluation criteria or in the comments they make in the reviews - is not an option.

No comments:

Post a Comment