All-inclusive authorship on large scientific papers can a be dangerous practice

Alert: potential unpopular opinion
Thread 1/n

Many argue that authorship should be an all-inclusive mechanism to assign credit, e.g.
https://twitter.com/xhimichka/status/1327652998781218816?s=09
https://twitter.com/martin_hebart/status/1315232143170125826

I don't disagree with this. As scientific projects grow, larger and larger teams with varied expertise are necessary to do the work.
2/n

This includes the very practical (technical aspects of data collection) and the more abstract (managing the logistics and. All of this should ofc be credited.

3/n

But: authorship is *not* just for assigning credit (and all the career benefits it brings).

It's also about indicating responsibility.
4/n

Can I take responsibility for the integrity and quality of the data? For the rationale behind these analyses? For the conclusions drawn from the data? For a fair coverage of previous literature?

Now, that's a whole other story.
5/n

I've certainly struggled with authorship on projects where I had done enough to earn credit, but not enough to take responsibility (due to lack of expertise, time investment, or simply insufficient interest to stand behind the full project and its implications).

6/n

The traditional first-author-did-the-work / last-author-supervised handles this well: there are a limited number of people who get most credit, and also assume most responsibility for the final work.

Call me old-fashioned, but I think there is merit to this.
7/n

However, implicit responsibility assignment can break down even in small, traditional teams.

In cases of scientific fraud, co-authors are 'dragged in'; they put their name on a paper to earn credit, and later deal with consequences of assumed responsibility.

8/n

Thinking through all pieces of a project and how they fit together is a major undertaking, and can be quite separate from contributing concrete parts.

Prioritizing credit over responsibility may lead to massive, but incoherent or internally inconsistent scientific output.

9/n

What to do?
I'm a big fan of detailed contribution tables e.g. https://doi.org/10.1101/2020.01.17.909838 (see https://twitter.com/SteinmetzNeuro/status/1147241128858570752 for more).

Perhaps we need another dimension for levels of responsibility, to complement https://casrai.org/credit/ 
10/n

I hope journals will allow increasingly complex authorship indications, with separate indications for those authors who assume responsibility for the manuscript.

Can be used to check for full responsibility coverage, making it clear who to contact with questions.

11/n

Until then, long author lists should be accompanied not just by extensive credit breakdown, but also responsibility breakdown. These will correlate, but not overlap - and they will help when honest bugs, strange omissions or suspected fraud come to light.
12/n

Curious what you think!
Disclaimer: opinions are my own, and don't necessarily represent @IntlBrainLab, @CSHL or @LeidenPsy.

Fin.