Comments

• Andy

  July 4

  Thanks for sharing this, and welcome to the forum!

  Let me be totally honest with you. My initial response to your writing is very critical. There is a lot of theoretical exposition without, as far as I can see in a cursory reading, much serious consideration of counterarguments, as if "instrumental conceptions" of technology and Bruno Latour's actor network theory (ANT) are the only two alternatives, with "instrumental conceptions" serving almost as a straw man. And once the straw man is knocked down, ANT dominates the essay without any acknowledgment (unless I missed it) of its limitations, especially how ANT overlooks the importance of human psychology.

  This sentence epitomizes the problem:

  Each platform durably encodes particular methods and assumptions about knowledge work, shaping what practitioners come to understand as effective knowledge management.

  Here platforms strangely seem to have more agency than people, as if people are being shaped by them without even knowing that it is happening. This may be true for some people, but not everyone is so ignorant, and some of us have developed sophisticated models of what is happening, so we need a more in-depth consideration of human differential psychology within the context of embodied cognitive science. You might object that this is not the subject of your essay, but acknowledging these areas of study and the challenge they pose to your application of ANT is part of the consideration of counterarguments.

  If you're not familiar with Ronald Giere, I'd suggest reading his criticism of Latour and similar theorists. I think Giere is basically right against you and Latour, so my dissatisfaction with your thesis is unsurprising. For example, here's Giere in "The problem of agency in scientific distributed cognitive systems", Journal of Cognition and Culture (4)3-4, 2004, 759–774:

  It may not be an exaggeration to say that for Latour there is no such thing as a cognitive agent. There are only "actants" connected in more or less tightly bound networks, transforming material representations, and engaged in agonistic competition with other networks. Actants include both humans and nonhumans in relationships that Latour insists are "symmetric." Thus, in his recent book, Pandora's Hope (1999, 90), Latour writes of Joliot trying to produce the first nuclear chain reaction: "Such was his scientific work: holding together all the threads and getting favors from everybody, neutrons, Norwegians, deuterium, colleagues, anti-Nazis, Americans, paraffin..." [...] Contrary to Latour, I think this asymmetry is fundamental for understanding science as a cultural activity. This is not to say that neutrons are not causally active. One might even say that they are causal agents, but this does not make them comparable to human agents. On the other hand, being a thoroughgoing naturalist, I would not claim that human agents operate outside the causal order of nature in some sort of Kantian “kingdom of ends.” It is enough that humans are a highly evolved form of life capable of sustaining a culture which has developed the conception of human agency outlined above.

  At the start of that paper, Giere said:

  From the perspective of cognitive science, it is illuminating to think of much contemporary scientific research as taking place in distributed cognitive systems. This is particularly true of large-scale experimental and observational systems such as the Hubble Telescope. Clark, Hutchins, Knorr-Cetina, and Latour insist or imply such a move requires expanding our notions of knowledge, mind, and even consciousness. Whether this is correct seems to me not a straightforward factual question. Rather, the issue seems to be how best to develop a theoretical understanding of such systems as distributed cognitive systems.

  Following Giere's line of thought here, and to repeat what I already said, I don't think you have an adequate theoretical understanding of PKM in your thesis because you don't pay enough attention to the human aspect of distributed cognitive systems, and specifically how people differentially model their cognitive systems.

  Some other publications by Giere that may be helpful:

  • Cynthia Passmore, Julia Svoboda Gouvea, & Ronald N. Giere (2014). "Models in science and in learning science: focusing scientific practice on sense-making". In: Matthews, M. R. (ed.), International Handbook of Research in History, Philosophy and Science Teaching (pp. 1171–1202). New York: Springer. https://doi.org/10.1007/978-94-007-7654-8_36
  • Ronald N. Giere (2012). "The role of psychology in an agent-centered theory of science". In: Proctor, R. W. & Capaldi, E. J. (eds.), Psychology of Science: Implicit and Explicit Processes (pp. 73–85). Oxford; New York: Oxford University Press. https://doi.org/10.1093/acprof:oso/9780199753628.003.0004
  • Ronald N. Giere (2012). "Scientific cognition: human centered but not human bound". Philosophical Explorations, 15(2), 199–206. https://doi.org/10.1080/13869795.2012.677850
  • Ronald N. Giere (2010). "An agent-based conception of models and scientific representation". Synthese, 172(2), 269–281. https://doi.org/10.1007/s11229-009-9506-z

  I expect that studying more psychology of science and technology would be helpful too, for example: Jason R. Finley, Farah Naaz, & Francine W. Goh (2018). Memory and Technology: How We Use Information in the Brain and the World. Cham: Springer. https://doi.org/10.1007/978-3-319-99169-6
• Andy

  July 5

  At the start of section 3.3 on "atomic notes", this thesis also repeats the misconception that Luhmann invented the practice of "writing discrete, interlinked notes". People have often said on the Internet that Luhmann invented this practice, but there is no need to repeat this misconception as uncontested fact. In fact, among English speakers, people were already using and theorizing about highly granular (atomic) knowledge elements several decades before Luhmann's slip-box became widely known. This section's history and conceptualization of "atomic notes" is just one among many possible such histories and conceptualizations, but it doesn't acknowledge that there could be alternatives.

  In my previous comment, I pleaded for more attention to how people differentially model their cognitive systems. I would add that this entire thesis is an example of its author modeling cognitive systems, but nothing is revealed about the author. Perhaps this is partly due to the choice to write in an "impersonal" academic style, and/or perhaps it is due to the theoretical deprecation of human agency that I mentioned in my previous comment? Does the author's theory apply to the author as well, and how (examples?), or is the theory only about other people's knowledge? Whatever the reasons for this lack of attention to the author, I don't consider it a virtue. In the book Qualitative Literacy (Small & Calarco, 2022), one of the criteria for high-quality qualitative research is self-awareness, the extent to which the researcher understands and communicates the impact of who they are, and their life path, on their research. Why did the author choose to apply ANT to PKM instead of (in my opinion) a better theory? Small & Calarco's criterion is specifically for ethnographic research, but I think it also applies to theory when a theory is about people who are doing more or less what the theorist is doing, as in a theory of PKM.
• Andy

  July 6

  Although I have been very critical of the thesis shared above in its present state, this morning I was thinking that it does point to a way to improve a template by Giere that I have mentioned previously in the forum, such as in a discussion on "how to improve thinking skills". There I quoted Giere, who wrote:

  I argue that we should focus on the pragmatic activity of representing, so that the basic representational relationship has the form: Scientists use models to represent aspects of the world for specific purposes.

  Giere replaced the variables in that sentence with letters to give the form:

  S uses X to represent W for purposes P.

  This formula can be considered an instance of what the thesis calls an "instrumental conception" of science and technology, because it abstracts away the effects (E) of technical artifacts and other things (T) on the people, models, representational processes, purposes, and perhaps even the aspects of the world being represented, depending on on what those aspects are. (Although it should be noted that Giere's conception of X already includes T.)

  So, despite my criticism of the thesis, it has been helpful in pointing out to me that this particular formula of Giere can be expanded to include other causal factors, perhaps like this:

  S uses X to represent W for purposes P, involving T that have effects E on S, X, W, and P.

  Where I still think that ANT is insufficient as a theory of PKM is where it assumes that we can ignore the first half of that sentence, the so-called "instrumental" half concerned with cognitive agents and their representational processes. That is still the most important part for a theory of knowledge, especially an epistemological theory of knowledge about knowledge or meta-modeling, although it's true that the effects of technical artifacts and other things shouldn't be ignored, and indeed are often impossible to ignore, as most of us have probably experienced when struggling with our technical artifacts.
• Andy

  July 7 edited July 7

  Welcome to Day 4 of my solo commentary!

  I wrote a couple of days ago:

  At the start of section 3.3 on "atomic notes", this thesis also repeats the misconception that Luhmann invented the practice of "writing discrete, interlinked notes". People have often said on the Internet that Luhmann invented this practice, but there is no need to repeat this misconception as uncontested fact. In fact, among English speakers, people were already using and theorizing about highly granular (atomic) knowledge elements several decades before Luhmann's slip-box became widely known. This section's history and conceptualization of "atomic notes" is just one among many possible such histories and conceptualizations

  I should have included a literature pointer in this previous comment.

  For me, the system of granular knowledge elements that most influenced me (but not until the 2000s) before I discovered zettelkasten.de was IBIS, developed by Horst Rittel et al. from the 1960s through the 1980s. It was comparable to other structured argumentation schemes like Stephen Toulmin's in The Uses of Argument (1958), but was more directly aimed at nonacademic applications.¹ IBIS started on index cards, and there were various computer implementations of it by academic researchers in the 1980s. One book from 1990 about hypertext programming that is largely devoted to IBIS contains a section heading that should sound familiar: "One Idea per Node": "'One Idea per Node' should serve as a good measure for the size of your nodes. If you have many small sized nodes, this will increase the degree of organization of information in your system."² Usually this was applied to design rationale, but it was also applied by some researchers to writing composition (which is not so different from design rationale if we think of writers as "designers of texts").³ One of those researchers, Jeff Conklin, spent more than two decades trying to popularize various implementations of IBIS and related methods. All of this was before Luhmann's Zettelkasten was widely known (at least, however widely known it has become) among English speakers.

  In the last five years or so, computer scientist Joel Chan has drawn upon the aforementioned research and has tried to popularize a rather IBIS-like system sometimes called QCE or "discourse graph". As I recall, I was introduced to Chan's work in a 2023 discussion in this forum, "Discourse graph and Zettelkasten". Apparently Chan was already well known at the time among Roam Research users for his Discourse Graph plugin. Chan is an example of someone who seems to have been more influenced by the aforementioned English-language structured argumentation research literature than by Luhmann's Zettelkasten. The reference lists of his papers are a good, if incomplete, guide to this literature.^45678910

  ---

  1. There is a partial history in: D. Scott McCrickard (2012). Making Claims: The Claim as a Knowledge Design, Capture, and Sharing Tool in HCI. San Rafael, CA: Morgan & Claypool. ↩︎

  2. Safaa H. Hashim (1990). Exploring Hypertext Programming: Writing Knowledge Representation and Problem-Solving Programs. Blue Ridge Summit, PA: Windcrest. ↩︎

  3. An early example of IBIS applied to writing composition software: Wolfgang Schuler & John B. Smith (1992) [1990]. "Author's Argumentation Assistant (AAA): a hypertext-based authoring tool for argumentative texts". In: Rizk, A., Streitz, N. A., & André, J. (eds.), Hypertext: Concepts, Systems and Applications: Proceedings of the First European Conference on Hypertext, INRIA, France, November 1990 (pp. 137–151). Cambridge, UK; New York: Cambridge University Press. ↩︎

  4. Xin Qian, Matt J. Erhart, Aniket Kittur, Wayne G. Lutters, & Joel Chan (2019). "Beyond iTunes for papers: redefining the unit of interaction in literature review tools". In: CSCW '19 Companion: Conference Companion Publication of the 2019 ACM Conference on Computer Supported Cooperative Work and Social Computing: November 9–13, 2019, Austin, TX, USA (pp. 341–346). Association for Computing Machinery. ↩︎

  5. Joel Chan (2020). "Knowledge synthesis: a conceptual model and practical guide". ↩︎

  6. Joel Chan, Xin Qian, Katrina Fenlon, & Wayne Lutters (2020). "Where the rubber meets the road: identifying integration points for semantic publishing in existing scholarly practice". Paper presented at the virtual conference, JCDL 2020 Workshop on Conceptual Modeling, August 1, 2020, Wuhan, China. ↩︎

  7. Xin Qian, Katrina Fenlon, Wayne G. Lutters, & Joel Chan (2020). "Opening up the black box of scholarly synthesis: intermediate products, processes, and tools". Proceedings of the Association for Information Science and Technology, 57(1), e270. ↩︎

  8. Joel Chan (2021). "Discourse graphs for augmented knowledge synthesis: what and why". ↩︎

  9. Joel Chan, Matthew Akamatsu, David Vargas, Lukas Kawerau, & Michael Gartner (2024). "Steps towards an infrastructure for scholarly synthesis". ↩︎

  10. Siyi Zhu, Rob Haisfield, Brendan Langen, & Joel Chan (2024). "Patterns of hypertext-augmented sensemaking". In: Yao, L., Goel, M., Ion, A., & Lopes, P. (eds.), UIST '24: Proceedings of the 37th Annual ACM Symposium on User Interface Software and Technology, October 13–16, 2024, Pittsburgh, PA, USA (pp. 1–17). Association for Computing Machinery. ↩︎
• dgarner

  July 8

  @Andy I'm enjoying trying to follow some of what you are saying. It's tough to get it through my old, thick head, but thanks for sharing it with us. I'm particularly interested in any aspects of automation regarding organizing thoughts.
• Andy

  July 8 edited July 8

  @dgarner, glad you're enjoying it. I'm not trying to be perfectly clear and exact here, as I'm assuming that anything that needs to be clarified could be clarified through further discussion. I already see some parts I would change, but I'll leave them as they are, as practice in resisting my perfectionism.

  Regarding automation of organization of thoughts (statements, claims): using a computer already greatly helps to automate indexing, and software such as DEVONthink, which I use, has long existed that can suggest categories for thoughts (although I don't use this feature). DEVONthink has also long been able to suggest that notes are related, and I sometimes consult this feature, but this does not automate the task of directly relating (linking) two thoughts using a schema of link types. It may be that current developments in generative AI have enabled automation of these kinds of relations, but I've never used applications with such a feature. D-Agree, a discussion support platform, reportedly can do automatic IBIS classification of thoughts in a discussion in real time. If it could also suggest refactoring of the organization of thoughts in the discussion, that would be very impressive.

  Totally automated organization of thoughts seems undesirable, as it would eliminate eufriction or desirable difficulty for the people who are doing knowledge work. If we didn't think people were important, then sure, why wouldn't we maximize automation and minimize the amount of thinking that people have to do? (Desirability and extent of automation would depend in part on people's purposes, Giere's P mentioned above.)

  Post edited by Andy on July 8