This is a really useful contribution. I can see two next steps for this: first, poke and prod the nine categories you have to make sure they are comprehensive and well-defined. Maybe some could be combined, or a tenth needs to be added. It would be interesting to run every R&D organization we can collectively think of to see if every one of those fits somewhere, somehow in this taxonomy.
Second (this is where it gets interesting -- and provide insight into what additional models we should be developing): work through the connections between characteristics to see if gaps actually matter. As you point out, a simple presence/absence across ten parameters yields 1,024 types, and since you've got nine parameters that aren't binary there is some unknown but larger than 1,024 types. But that probably doesn't matter. For example, and just to completely make something up, say there is no organization that has zero motivation by user specific needs (parameter 2) but exceptional focus on one mission (parameter 6). Maybe that particular configuration doesn't exclude useful forms of scientific inquiry, but maybe it does. The only way to do that is to trace the connections between each of the nine parameters.
The way this can be done could be based on this approach:
That table works through a taxonomy of eight parameters with between two and four positions on the spectrum, drawing connections between all of them, creating a set of models at the bottom. The artwork is a bit clunky, but it's pretty easy to follow what the authors are getting at: they show that there are no systems where the liability structure (their parameter 2) also has an active quantitative investment strategy (their parameter 3). Once these connections, or lack of connections, is made visible, it becomes simpler to identify if it's a relevant gap.
Thanks for pulling all this together. Great stuff.
So much great thinking in this! Your question at the end about untried or incoherent combinations made me wonder about a third-ish option: how often are parameter values held together by an equilibrium?
Some of the bureaucratic friction around budgets is endogenous. Ex. a small private funder can delegate substantial resources through fairly direct trust whereas a federal agency distributing public money probably needs scalable proxies for it.
If so, some attractive combinations may just be unstable. Basically some of the undesirable parameter values might enable desirable values elsewhere in the system.
My youngest brother and I went to school ten years apart, in the same family. Mine was public: weapons checks, shooter drills, dogs at the lockers — an institution that processed me, kindly where it could manage kindness. His was private: he was treated as a legitimate member of an evolving community and taught to lean into discomfort rather than script around it. That is your governance question, run on children. Identical mission; radically different answer to who counts as a participant.
I'm for this taxonomy. I work in research support and have moved between institutes with very different resources; your nine parameters expertly describe real differences I watch operate daily, and "diversity of roles" as a predictor of the range of projects an org can undertake is quietly one of the best lines in the piece. But look at who appears in the taxonomy from outside the organisation: users, who can say no; funders, who set agendas. The only outsiders in the frame seem to be the ones with power. The people research runs on and through — support staff, technicians, the communities data comes from — aren't very salient. So, a candidate tenth I would propose is standing. Whom does an org treat as a legitimate interlocutor, and whom as throughput? In my understanding of his work, Musa al-Gharbi (https://www.youtube.com/watch?v=34_iFzRXWxQ) observes that many communities trust science while distrusting scientists, because they experience research as something done on and from them — they give and receive but hold no equity. That is not an abstraction. During the West African Ebola epidemic, data gathered on the ground in brutal circumstances was processed abroad and published behind paywalls that the researchers who survived could not afford — collected from them, then sold back to them. The First Nations OCAP principles — ownership, control, access, possession — exist precisely because being defined by data you cannot govern makes it harder, not easier, to participate in the science defining you. (Your ninth parameter, openness and IP strategy, already brushes against this; standing is what it looks like from the other side of the consent form.) Chana Teeger (https://behavioralscientist.org/what-does-boredom-teach-us-about-how-we-engage-with-history/) found that bored students in history classrooms weren't failing to engage; boredom was a produced distancing, a rational reading of their position in the room. Disengagement from science works the same way. So does disengagement inside our labs. I work on my own resentments here, not always successfully.
Standing is measurable in your spirit: what share of roles carry it; who can propose, and who can veto; how much unhurried conversation an org makes room for with people who don't look useful on first pass. Bell Labs' lunch tables were exactly that instrument.
You close by looking for bright minds. Some are already filing the grant paperwork and translating the impact sections, inside the fence, rarely counted among the gardeners. Which parameter would find them?
I believe the correct response is "Hnnnnnnnnghhhhh" :D
either way, it sounds like you're being strangled! :)
yet I still feel like your comment was a compliment!
Words were simply not adequate to convey my level of elation! ;)
is that a silent gh or the one that sounds like "f"?
Further research is needed to
This is a really useful contribution. I can see two next steps for this: first, poke and prod the nine categories you have to make sure they are comprehensive and well-defined. Maybe some could be combined, or a tenth needs to be added. It would be interesting to run every R&D organization we can collectively think of to see if every one of those fits somewhere, somehow in this taxonomy.
Second (this is where it gets interesting -- and provide insight into what additional models we should be developing): work through the connections between characteristics to see if gaps actually matter. As you point out, a simple presence/absence across ten parameters yields 1,024 types, and since you've got nine parameters that aren't binary there is some unknown but larger than 1,024 types. But that probably doesn't matter. For example, and just to completely make something up, say there is no organization that has zero motivation by user specific needs (parameter 2) but exceptional focus on one mission (parameter 6). Maybe that particular configuration doesn't exclude useful forms of scientific inquiry, but maybe it does. The only way to do that is to trace the connections between each of the nine parameters.
The way this can be done could be based on this approach:
https://www.oecd.org/content/dam/oecd/en/publications/reports/2014/03/oecd-journal-financial-market-trends-volume-2013-issue-2_g1g23271/fmt-v2013-2-en.pdf
Table 2 on pg 111.
That table works through a taxonomy of eight parameters with between two and four positions on the spectrum, drawing connections between all of them, creating a set of models at the bottom. The artwork is a bit clunky, but it's pretty easy to follow what the authors are getting at: they show that there are no systems where the liability structure (their parameter 2) also has an active quantitative investment strategy (their parameter 3). Once these connections, or lack of connections, is made visible, it becomes simpler to identify if it's a relevant gap.
Thanks for pulling all this together. Great stuff.
So much great thinking in this! Your question at the end about untried or incoherent combinations made me wonder about a third-ish option: how often are parameter values held together by an equilibrium?
Some of the bureaucratic friction around budgets is endogenous. Ex. a small private funder can delegate substantial resources through fairly direct trust whereas a federal agency distributing public money probably needs scalable proxies for it.
If so, some attractive combinations may just be unstable. Basically some of the undesirable parameter values might enable desirable values elsewhere in the system.
My youngest brother and I went to school ten years apart, in the same family. Mine was public: weapons checks, shooter drills, dogs at the lockers — an institution that processed me, kindly where it could manage kindness. His was private: he was treated as a legitimate member of an evolving community and taught to lean into discomfort rather than script around it. That is your governance question, run on children. Identical mission; radically different answer to who counts as a participant.
I'm for this taxonomy. I work in research support and have moved between institutes with very different resources; your nine parameters expertly describe real differences I watch operate daily, and "diversity of roles" as a predictor of the range of projects an org can undertake is quietly one of the best lines in the piece. But look at who appears in the taxonomy from outside the organisation: users, who can say no; funders, who set agendas. The only outsiders in the frame seem to be the ones with power. The people research runs on and through — support staff, technicians, the communities data comes from — aren't very salient. So, a candidate tenth I would propose is standing. Whom does an org treat as a legitimate interlocutor, and whom as throughput? In my understanding of his work, Musa al-Gharbi (https://www.youtube.com/watch?v=34_iFzRXWxQ) observes that many communities trust science while distrusting scientists, because they experience research as something done on and from them — they give and receive but hold no equity. That is not an abstraction. During the West African Ebola epidemic, data gathered on the ground in brutal circumstances was processed abroad and published behind paywalls that the researchers who survived could not afford — collected from them, then sold back to them. The First Nations OCAP principles — ownership, control, access, possession — exist precisely because being defined by data you cannot govern makes it harder, not easier, to participate in the science defining you. (Your ninth parameter, openness and IP strategy, already brushes against this; standing is what it looks like from the other side of the consent form.) Chana Teeger (https://behavioralscientist.org/what-does-boredom-teach-us-about-how-we-engage-with-history/) found that bored students in history classrooms weren't failing to engage; boredom was a produced distancing, a rational reading of their position in the room. Disengagement from science works the same way. So does disengagement inside our labs. I work on my own resentments here, not always successfully.
Standing is measurable in your spirit: what share of roles carry it; who can propose, and who can veto; how much unhurried conversation an org makes room for with people who don't look useful on first pass. Bell Labs' lunch tables were exactly that instrument.
You close by looking for bright minds. Some are already filing the grant paperwork and translating the impact sections, inside the fence, rarely counted among the gardeners. Which parameter would find them?