iiinova — where your digital self keeps evolving

Bryan Gee @koskinonodon.bsky.social · 6h

No doubt, I think that's where the point about repeating authorship comes in - someone who has done it before knows how to write a cover letter, sell a big picture, etc. There's a bunch of citations about that in the wall of text

100

Bryan Gee @koskinonodon.bsky.social · 6h

To me, the issue is more that the royal "we" put a lot of weight on the brand name of these journals and fund, hire, etc. in part based on perceived prestige/reputation, so I think there's a risk that people just consolidate on studying a few charismatic clades, otherwise they end up unemployed

120

Bryan Gee @koskinonodon.bsky.social · 6h

I don't think that's a problem - as with museums or modern conservation, leaning on the most charismatic taxa as a gateway to a broader discipline is really useful. Although I would also say it risks creating a self-reinforcing loop (unpopular taxa can't become popular if rarely publicized)

110

Bryan Gee @koskinonodon.bsky.social · 6h

If you send the survey, we can co-author the paper

100

Bryan Gee @koskinonodon.bsky.social · 6h

Yeah but keep going back further and you don't get any stem tet work of any kind til Gaiasia in 2024, then Qikiqtania in 2022. I think old fish are more in the "description of known taxa" camp

100

Bryan Gee @koskinonodon.bsky.social · 6h

As another point, one could perhaps argue that paleoproteomics should be discarded because that's mostly applied to mammals, and then we start getting into evo-devo studies and the problem I ultimately ran into, which is what is definitely paleo, definitely not paleo, kind of paleo, etc.

120

Bryan Gee @koskinonodon.bsky.social · 6h

I agree it's an unusual case, but there are Nature papers on hominins that are basically the same as non-hominin descriptive work and dealing with the same themes (taxonomy, biogeography, etc.) - cutting out some of that work has to draw some arbitrary line at some homin* level

121

Bryan Gee @koskinonodon.bsky.social · 7h

This is sort of true but I did some cursory passes at trying to sample non-naming papers (it could be hard to determine what was/wasn't paleo) and the skew got worse in some cases, mainly because of how many studies on hominins there are (more than dinos) compared to new species (rare)

211

Bryan Gee @koskinonodon.bsky.social · 7h

I didn't go into the literature on extant taxa or specific types of studies on them too much, but I know a few in this area exist. Example: doi.org/10.1016/j.an...

010

Bryan Gee @koskinonodon.bsky.social · 10h

Single-spaced, 10 it font no less

010

Bryan Gee @koskinonodon.bsky.social · 11h

Got this email this morning, I'm not even sure I understand the product

Spam email from someone advertising a product called Caelrith that scrapes preprints for...something

100

Bryan Gee @koskinonodon.bsky.social · 11h

Some techbro will feed all my data into an LLM and then sell a product for researchers to tell you this lol

110

Bryan Gee @koskinonodon.bsky.social · 11h

Yeah there's probably some eay to at least attempt it, though I'm thinking about how juat the survey design could get complicated fast (do you only list charismatic clades, how granular into say Dinosauria do you get, etc.). Also thinking about all the surveys I ignore...

000

Bryan Gee @koskinonodon.bsky.social · 11h

Yeah I ran out of predefined color palettes, so the colors stopped holding consistent once I started doing clade-level splits 😅

110

Bryan Gee @koskinonodon.bsky.social · 11h

It's an interesting question but a hard one to test given many researchers work on multiple groups, so setting a threshold on # of papers or something else to qualify as a specialist would be arbitrary. Also have the issue of how to systematically compile even a list of candidate researchers

100

Bryan Gee @koskinonodon.bsky.social · 11h

Yup, both in there, though they're actually 50% of all pan-arthropods named in Sci Advances writ large (there are only two insects in Sci Adv, both scorpionflies from Myanmar amber)

110

Bryan Gee @koskinonodon.bsky.social · 13h

Lastly, *no*, the preprint is not "really" 308 pages. There are ~1,000 lines of text meant for reading. The length is b/c I included the reference for every naming paper in the dataset so that when this gets picked up by Google Scholar, those people will get credit (see below) = 243 pgs of refs.

doi.org

Increasing the equitability of data citation in paleontology: capacity building for the big data future | Paleobiology | Cambridge Core

Increasing the equitability of data citation in paleontology: capacity building for the big data future - Volume 50 Issue 2

041

Bryan Gee @koskinonodon.bsky.social · 13h

I think there are other related research projects that can be done on different/broader scopes, but these will require a lot more work (both to design and to execute), like trying to sample all anatomical descriptive work (reliable search terms, does histo count, what about mixed-method studies).

121

Bryan Gee @koskinonodon.bsky.social · 13h

I am also not saying that dinosaur people can do crap work and get it published in Nature. Most dinosaur work is not published in Nature. But I agree with this post by @daveyfwright.bsky.social, and I do think that "is dinosaur" is a huge asset at selective journals. bsky.app/profile/dave...

131

Bryan Gee @koskinonodon.bsky.social · 13h

Also! This study is not me, a specialist of decidedly uncharismatic taxa, having sour grapes. I like dinosaurs, I like that the public likes dinosaurs, I have done months of fieldwork on dinosaurs, etc.

120

Bryan Gee @koskinonodon.bsky.social · 13h

To me, this is not really an issue of journal editorial practices - journals can publish whatever they want - but rather the academic system in which we accord a lot of value to pubs in Nature, Science, etc. just by virtue of that name brand.

131

Bryan Gee @koskinonodon.bsky.social · 13h

So while I think that this kind of bibliometric work is just interesting on the premise of testing a widely held belief among academics, it points to potential systemic issues that can have real adverse effects on the field and our ability to reconstruct a comprehensive history of life.

121

Bryan Gee @koskinonodon.bsky.social · 13h

Because of how much weight we accord to publication in high-IF journals, even when studies are widely refuted or retracted, bias toward certain study systems in these high-value venues, in a taxonomy-rooted field like paleo, risks progressive consolidation around those few study systems.

121

Bryan Gee @koskinonodon.bsky.social · 13h

Ultimately, I think the cause of taxonomic bias in publishing is less important than the fact that bias exists. It's well-known that publishing in big-name journals gives you an advantage in other areas (hiring, grants, future manuscripts) - i.e. the Matthew Effect.

doi.org

The Matthew Effect in Science

This account of the Matthew effect is another small exercise in the psychosociological analysis of the workings of science as a social institution. The initial problem is transformed by a shift in the...

131

Bryan Gee @koskinonodon.bsky.social · 13h

That doesn’t mean journals/editors aren't contributing to researchers self-selecting. Having had some of my work get desk-rejected by society journals w/ IFs < 3 because it's "not of sufficiently broad interest," lower-IF journals can create experience/perception of bias, leading to self-selection.

static.klipy.com

Mean Girls Rachel Mc Adams

Alt: Mean Girls "You can't sit with us" GIF

121

Bryan Gee @koskinonodon.bsky.social · 13h

Another explanation, which I tend to favor, is that researchers self-select on an assumption that there is bias in editorial decisions, so people who work on less charismatic clades don't even think about submitting to higher-IF journals unless they have something amazing.

static.klipy.com

Toy Story Rex: I Can't Take Rejection!

Alt: Toy Story Rex: I Can't Take Rejection!

121

Bryan Gee @koskinonodon.bsky.social · 13h

What is the causal mechanism? It could be editors using subjective critera like "broader interest", but you can't test "editors are the cause" without detailed data on journal submissions and why some are rejected (desk or not) and others are accepted. Journals are never going to share that data.

static.klipy.com

Purplecrumbs: That's Confidential Puppet

Alt: Purplecrumbs: That's Confidential Puppet

221

Bryan Gee @koskinonodon.bsky.social · 13h

Something important to note is that this analysis has only demonstrated a correlation between journal prestige (via IF) and relative proportion of new species belonging to clades that we think of as being relatively charismatic. And as we all know, correlation =/= causation!

130

Bryan Gee @koskinonodon.bsky.social · 13h

(For those curious as to who are the most frequent repeat authors, it's in the public dataset: doi.org/10.5281/zeno.... Xu Xing is at the top, with between 18 [counting by unique ORCID] and 20 [counting by unique name] naming papers in Nature & Science within the sampled temporal range).

doi.org

120

Bryan Gee @koskinonodon.bsky.social · 13h

Something I only briefly touch on is how authorship/research collabs could lead to repeat publication (which could then result in taxonomic skews in high-IF journals where there are few paleo pubs). About 10% of authors with a Nature/Science naming paper in the dataset are listed on 2+.

Horizontal bar graph depicting the percent of authors of naming papers in Nature or Science who are listed on 1, 2-4, or 5+ articles for each of the journals.

Horizontal bar graph depicting the percent of authors of naming papers in Nature or Science who are listed on 1, 2-4, or 5+ articles between the journals.

120

Bryan Gee @koskinonodon.bsky.social · 13h

So TLDR, I think there is good evidence for taxonomic bias in high-IF journals on the basis of naming papers, specifically towards avemetatarsalians (most of which are non-avian dinosaurs) and mammaliamorphs (many of which are primates or stem mammals)!

121

Bryan Gee @koskinonodon.bsky.social · 13h

I did this kind of breakdown for a few other categories as well (sorry trilobite people, the big-name journals don't like your study system).

Stacked horizontal bar graph showing the relative proportion of different pan-arthropod clades between all sampled journals.

Stacked horizontal bar graph showing the relative proportion of different insect clades between all sampled journals.

231

Bryan Gee @koskinonodon.bsky.social · 13h

Worth noting that I am well-aware that not all {insert members of clade} are equally charismatic. The data show that non-avian theropods are more common than ornithischians in high-IF journals, for example. Likewise for primates vs. other mammals (but non-mammalian mammaliaforms are hot too).

Stacked horizontal bar graph showing the relative proportion of different avemetatarsalian clades between all sampled journals.

Stacked horizontal bar graph showing the relative proportion of different mammaliamorph clades between all sampled journals.

121

Bryan Gee @koskinonodon.bsky.social · 13h

Here's the relationship b/t IF and charismatic clade proportion as a simple regression (L = all journals; R = without JVP). The r2 is ~0.32 with JVP and ~0.65 without. Gondwana Research and its affinity for Myanmar inverts are the green dot far down the y-axis.

Paired scatterplots with linear regressions depicting the log10-transformed Journal Impact Factor (x-axis) against the percent of new species that belong to Avemetatarsalia or Mammaliamorpha. The left graph is for all 13 journals sampled here; the right graph omits the Journal of Vertebrate Paleontology.

120

Bryan Gee @koskinonodon.bsky.social · 13h

To that point, I stopped data collection on 9/4 and just in the month since, we've gotten a new eutriconodont (Dongoconodon) in Nature Comms, a new microraptorine (Norellraptor) in Nature Comms, a new oviraptorosaur (Infernogallus) in Nature Comms, and a new docodont (Megacauda) in Nature.

130

Bryan Gee @koskinonodon.bsky.social · 13h

This graph, which keeps avemetatarsalians and mammaliamorphs separate and puts everything else in one category, summarizes the primary findings. Journals with higher IFs publish new species from these clades more often. JVP has a lot too but relatively few primates and non-avian avemetatarsalians.

Stacked horizontal bar graph showing the relative proportion of avemetatarsalians, mammaliamorphs, and all other taxa between all sampled journals.

121

Bryan Gee @koskinonodon.bsky.social · 13h

The control group is a lil wonky because I included JVP (verts only, obviously), but most new species in these journals are non-vertebrates. In the 4 non-JVP journals, verts are only ~20-25% of all new species. Mammals are common (JVP), but primates are rare. Non-avian avemetatarsalians also rare.

Horizontal bar graph depicting the percent of new species in each of the 20 categories used in this study, grouped across low-JIF journals sampled here.

Horizontal bar graph depicting the percent of new species in each of the 20 categories used in this study, separated across low-JIF journals sampled here.

121

Bryan Gee @koskinonodon.bsky.social · 13h

Despite only representing 1.5% of all holotypes in the entire dataset, Myanmar represents 15.5% of all holotypes in the medium-IF journal group specifically. Note that we don't see this skew in Nature or Science.

Horizontal bar graph depicting the percent of new species in each of the 20 most productive countries in terms of number of holotypes for new species sampled in this study, across all journals sampled here.

121

Bryan Gee @koskinonodon.bsky.social · 13h

What accounts for the representation of insects, which are probably of middling charisma if we're being generous? New species from Myanmar amber. For Current Bio, Gondwana Research, & Nature Comms, the most common combo of age+category+geography is Cretaceous insects from Myanmar 🇲🇲.

Table showing the most common (taxonomic) category, the most common combination of category and age, and the most common combination of category, age, and geography for each journal.

120

Bryan Gee @koskinonodon.bsky.social · 13h

We sort of see the same thing when looking at the medium-profile group, but there are some notable differences. Insects are really common! Non-primate mammaliamorphs are actually underrepresented relative to the control group here.

Horizontal bar graph depicting the percent of new species in each of the 20 categories used in this study, grouped across medium-JIF journals sampled here.

Horizontal bar graph depicting the percent of new species in each of the 20 categories used in this study, separated across medium-JIF journals sampled here.

121

Bryan Gee @koskinonodon.bsky.social · 13h

We can compare the proportion that each category represents in Nature & Science vs. the proportion in the control group. If (+), that category is overrepresented compared to the control; if (-), it's underrepresented. Avemetatarsalians 🦖 & mammaliamorphs 🦣are way overrepresented (sorry 🦪).

Horizontal bar graph depicting the difference in proportions between high- and low-JIF journals with respect to each of the 20 categories used here.

131

Bryan Gee @koskinonodon.bsky.social · 13h

Let's get into results! Firstly, as surprises no one, avemetatarsalians and mammaliamorphs are by far the most common groups in Nature and Science.

Horizontal bar graph depicting the percent of new species in each of the 20 categories used in this study, grouped across high-JIF journals sampled here.

Horizontal bar graph depicting the percent of new species in each of the 20 categories used in this study, separated across high-JIF journals sampled here.

141

Bryan Gee @koskinonodon.bsky.social · 13h

From these ranks, I divided species into 20 categories (one is a catch-all). Yes, these are skewed towards verts. Sorry if your favorite clade is not its own category. I spent a lot of time adjusting and prioritized more charismatic groups and keeping the catch-all 'Other' to under 10%.

Table depicting the 20 categories used in this study.

Horizontal bar graph depicting the percent of new species in each of the 20 categories used in this study, across all journals sampled here.

140

Bryan Gee @koskinonodon.bsky.social · 13h

I then used a script to retrieve taxonomic rank info from the PBDB API for each unique genus. Yes, I know (perhaps better than many) that the PBDB is not perfect. For the purposes of "are dinosaurs more common than dinoflagellates in Nature," it works fine (actually much better than fine).

150

Bryan Gee @koskinonodon.bsky.social · 13h

Why only new species? These are easy to find (have to designate with terms like 'sp. nov.'), are relatively attention-grabbing, and taxonomy is a primary component of the work. Finding & categorizing studies is harder with work on known taxa, multi-method studies, and mixed fossil-extant studies.

150

Bryan Gee @koskinonodon.bsky.social · 13h

I examined all papers naming new species (>4,200 papers, >8,300 species) across 13 journals between 2000 to ~Sept 2026, which include high-profile journals (Nature/Science), 6 medium-profile ones (e.g., PNAS, Nature Comms), and a control of 5 paleo journals (e.g., JVP, Sys Palaeo, J Paleo).

static.klipy.com

Brooklyn Nine-Nine: Amy Santiago in Paper Pile

Alt: Brooklyn Nine-Nine: Amy Santiago in Paper Pile

161

Bryan Gee @koskinonodon.bsky.social · 13h

(If this sounds familiar, I did post about this a few months back when it was just a pet project, and then a lot of people told me I should actually publish it): bsky.app/profile/kosk...

160

Bryan Gee @koskinonodon.bsky.social · 13h

But how do we know that it's not just confirmation bias - seeing a few dinosaurs in big name journals and over-extrapolating that as proof of systemic bias against a backdrop of zero {insert non-charismatic clade} in those same journals? We can, and should, get data to test this!

151

Bryan Gee @koskinonodon.bsky.social · 13h

I think most academics would say "yes" - here's a quote from Benton (2010; JVP) to that point: "[...] a new species of dinosaur is more likely to grace the pages of Nature or Science, and to attract wide press interest, than a new species of almost any group other than fossil hominids."

doi.org

Naming dinosaur species: the performance of prolific authors

Many current debates about biodiversity and large-scale evolution have identified the need for comprehensive species inventories. Such species lists may be incomplete because more collecting is n...

171

Bryan Gee @koskinonodon.bsky.social · 13h

There's no doubt that dinosaurs (and a few other clades) dominate the pop culture landscape and are of greater interest to the public, but does this skew translate to academic publishing, which, in theory, should be more objective in not overvaluing some taxa simply on the basis of broad popularity?

static.klipy.com

Littlefoot with a Tree Star Leaf

Alt: Littlefoot with a Tree Star Leaf

131