Fun ways of deciding authorship order (2016)

Posted by qifzer 3 days ago

Fun ways of deciding authorship order (2016)(dynamicecology.wordpress.com)

72 points | 34 comments

jpmattia 3 days ago|

The wordpress post is old, and so the author didn't have the chance to include my favorite method:

Every Author as First Author: (pdf) https://arxiv.org/pdf/2304.01393

madcaptenor 3 days ago||

This is biased in favor of authors with long names (which they mention at the beginning of their Future Work section)

rendaw 3 days ago||

That's addressed in "Future work"!

JohnKemeny 3 days ago|||

Extra funny, considering it's written by Demaine and Demaine.

Etheryte 3 days ago|||

The only shortcoming is that they currently use opacity even if there is only one author. In that case, it would seem natural to render the text as-is.

MortyWaves 3 days ago||

Why does it contain half-redacted words?

maartin0 3 days ago||

They're not redacted. If you zoom in you'll just see a very large number of stacked names which is ironic (I thought the same for a second)

jghn 3 days ago||

TFA touches on this, but one thing I initially found surprising is how few people understand different domains have different best practices around authorship order. It does make sense, people are typically not as involved in other domains and not exposed to those papers. But I do still find it surprising how different the practices can be overall.

madcaptenor 3 days ago|

The American Mathematical Society has a statement basically saying “in math, authorship order is alphabetical” that people going up for tenure can put in their files, in case people involved in the decision come from other disciplines which have other conventions.

https://www.ams.org/learning-careers/leaders/CultureStatemen...

setopt 3 days ago|||

Alphabetical order sounds interesting when you mix non-anglicized international names. Do you go by Unicode sort order?

I’m in physics, we have this thing where the first author did the most and the last author supervised the most, and the person in the middle just had an occasional coffee with them.

Wevah 3 days ago||

I’d guess it’d be a particular collation, rather than Unicode order…otherwise ö would always come after z (which is incorrect for English, but correct for, e.g., Swedish).

thaumasiotes 3 days ago||

> otherwise ö would always come after z

Really? Wouldn't that depend on how you spelled it and what kind of Unicode ordering you specified?

Which comes first?

    '\u0047\u00f6\u0064\u0065\u006c'
    '\u0047\u006f\u0308\u0064\u0065\u006c'

dmurray 3 days ago|||

Someone should study whether that statement is disproportionately included by authors with names later on in the alphabet.

xenonite 3 days ago||

Also I suppose that Z authors might have more publications because they are nice to collaborate with.

dhosek 3 days ago||

Kind of reminds me of the system we used in my band in the 90s: The person who brought the initial idea to the band gets to be first. After that, it was based on importance of contributions as determined by myself as the benign dictator, but if I contributed, my name always came last (unless I was the one who brought the idea to the band).

jvanderbot 3 days ago||

This is exactly how most professors and managers do it. Unless they themselves do the majority of the writing they are last by convention so it actually has some prestige to be last.

What you don't want is second-to-last on a paper w 4 or more authors. That's the worst.

bix6 3 days ago||

This is the method I use at work, feels right, puts the team first.

setgree 3 days ago||

Another idea is to only co-author with people with your last name, as in "A Few Goodmen: Surname-Sharing Economist Coauthors" by Allen C. Goodman, Joshua Goodman, Lucas Goodman, and Sarena Goodman:

> We explore the phenomenon of coauthorship by economists who share a surname. Prior research has included at most three economist coauthors who share a surname. Ours is the first paper to have four economist coauthors who share a surname, as well as the first where such coauthors are unrelated by marriage, blood or current campus.

https://scholar.harvard.edu/files/joshuagoodman/files/goodma...

dfltr 3 days ago||

My partner is a mathematician who realized (along with the other members of their working group) that if they were to deviate from the standard alphabetical authorship order, they could author a paper on the DILF Theorem.

setopt 3 days ago|

Sounds like they’re on track for a FILD medal with that paper.

hinkley 3 days ago||

> the order of their authorship was determined by executing the following commands in R:

    set.seed (7998976/5271)
    x ‹- sample (c("Anne", "Peder"), 1)
    print (paste ("The winner is", x, "!"))

But who picked the seed, Anne? And how do we know they didn’t have their thumb on the scale?

zvorygin 3 days ago||

My guess is that each simultaneously picked a number, one for numerator and one for denominator.

hinkley 3 days ago||

Hopefully. Though I wonder if that can be gamed if you pick the denominator and the value is rounded. Eg a large prime.

fph 3 days ago|||

That is weird; from what I understand (not an R expert), set.seed takes an integer, so I assume that number gets truncated / approximated to an integer. That means that all nearby seeds give the same result: for instance changing 7998976 to 7998977 or 7998975 makes no difference, up to the next multiple of 5271. This makes the result look a lot less random. Was Anne cheating?

madcaptenor 3 days ago||

From the documentation at https://stat.ethz.ch/R-manual/R-devel/library/base/html/Rand..., seed is "a single value, interpreted as an integer, or NULL (see ‘Details’)." From some quick testing it appears that "interpreted" means "truncated".

madcaptenor 3 days ago|||

I was wondering if this number was somehow significant in the paper, but it doesn't seem to be: https://journals.sagepub.com/doi/pdf/10.1177/251524591877096...

_Algernon_ 3 days ago||

Maybe we can solve this with blockchain?

JohnKemeny 3 days ago||

Or maybe quantum?

MortyWaves 3 days ago||

What surprising timing! I have started making a bookmarks page on my personal site and soon realised that any papers I linked to would need to deal with this. I couldn’t find a reliable answer so decided I would simply have authors listed in exactly the same order as found on the paper/site/wherever.

Der_Einzige 3 days ago||

You should in general prefer to give people as much credit as possible. in AI/ML we have the astrik of "equal contribution" which can be used to make N authors technically "first author".

tpoacher 3 days ago|

Yes but you'd still cite the paper as "as shown in FirstFirstAuthor et al (2024)" which rather defeats the purpose. And citing as "FirstFirstAuthor and SecondFirstAuthor and [...] et al" is both impractical and petty.

Which is what caused the whole "race to first authorship" mentality in the first place.

Der_Einzige 2 days ago||

No, you cite the very first author only. The "multiple first author" stuff is for when you talk about your paper to others.

tpoacher 3 days ago|

I once co-wrote a paper with a person who insisted on alphabetical order as a means of ensuring fairness.

Except his surname was Aaronson and mine starts with a P.

More comments...