There are possibilities for less biased orderings which are also less drama-prone. For example, choose a day 1, then use alphabetical order, but advance the first letter of the alphabet for each a day after day 1. Today is the alphabet for author ordering starts with A, tomorrow with B, and so on. If that still introduces a bias, then perhaps the alphabet should start with the same letter on consecutive days, but alternate between going forwards and backwards.
If you want an more-or-less unbiased but deterministic way to do this, you could sort the authors by whose birthday is closest (in either direction) at time of publishing. This additionally makes it so the precise date doesn’t matter too much. Making it closest upcoming birthday would be simpler, but if a colleague’s birthday is one day before you it kind of sucks.
But probably the random-order idea, as suggested by the article, would be even easier.
The reason I wanted something deterministic is that I wasn’t convinced that scientists would generally trust something that looked random with the stakes being somewhat high. When I think about the amount of scientific fraud, I’m not even sure that they should trust each other.
My alphabetical scheme isn’t ideally random—it gives an advantage to authors whose names begin with unusual letters.
Thanks for the information that date of publication can be somewhat foggy.
There are possibilities for less biased orderings which are also less drama-prone. For example, choose a day 1, then use alphabetical order, but advance the first letter of the alphabet for each a day after day 1. Today is the alphabet for author ordering starts with A, tomorrow with B, and so on. If that still introduces a bias, then perhaps the alphabet should start with the same letter on consecutive days, but alternate between going forwards and backwards.
In the article Hanson proposes a simpler method.
If you want an more-or-less unbiased but deterministic way to do this, you could sort the authors by whose birthday is closest (in either direction) at time of publishing. This additionally makes it so the precise date doesn’t matter too much. Making it closest upcoming birthday would be simpler, but if a colleague’s birthday is one day before you it kind of sucks.
But probably the random-order idea, as suggested by the article, would be even easier.
The reason I wanted something deterministic is that I wasn’t convinced that scientists would generally trust something that looked random with the stakes being somewhat high. When I think about the amount of scientific fraud, I’m not even sure that they should trust each other.
My alphabetical scheme isn’t ideally random—it gives an advantage to authors whose names begin with unusual letters.
Thanks for the information that date of publication can be somewhat foggy.