3 ms·

'pro and con' for case is a very strong design choice. What if it's an open problem. What if It's a wrong question, based on poor representation of the problem?

• mistermann 6y ago
  > 'pro and con' for case is a very strong design choice. What if it's an open problem. What if It's a wrong question, based on poor representation of the problem? Ok I see, and yes I agree. Although to me, I don't think this is necessarily as big of a problem as people seem unable to stop making it out to be, which I tried to point out in my comment: https://news.ycombinator.com/item?id=25904058 https://news.ycombinator.com/item?id=25904058 As I see it, the fundamental problem is the infinitely dimensional (recursively self-referencing, paradoxical, etc) nature of reality. A "right way" to model it is impossible - the best we can do is try to find a "good enough" (across all variables) approach, acknowledge (and gather opinions on) where we have fallen short and could improve, and iterate as you go. Kind of like life itself, "coincidentally". In the super short run, even adding a "middle" column would be a huge improvement. > I'm actually making tool for this exact same thing and the issue is most noticeable in kialo (which I argue is the same thing, with more mature UI). If you made all the research, conversations and planning to know, that answering this 'A or B' problem solves it, then Kialo (or that app) are perfectly fine tools. But most of the time people don't. > https://www.loomio.org https://www.loomio.org is another attempt at solving the same problem. It does provide some improvements, but also fails in interesting ways (mostly, I think, through not considering temporality very seriously). I think this is the most interesting (and largely unrecognized) niche out there - in no small part due to the obvious inherent complexity and no obvious business model, but I feel like those may not even be the main reason - I truly that this space is "hard to see" for most people. As for design of your product, what do you think of these pie-in-the-sky ideas (from the link above): > I think that approaches like this website are a not yet a perfect solution, but it feels like a step in the right direction. An improved implementation would have some clever means of allowing a crowd-sourced decomposition of the issue into it's constituent entities and behaviors, graphically displayed in some magical way to maximize the likelihood of everyone seeing the big picture free from misunderstanding, which "all" parties could agree upon (no agreement means more refinement is needed) and refer to in their arguments (so we're all talking about the same thing), and arguments/thinking could take place with precision (attached to the the specific entities and behaviors where the true disagreement resides). > A lot of people may have a strong aversion to this approach as it would turn easy argumentation into a painstakingly laborious & pedantic process, which is a natural consequence of using slow but accurate cognitive processing, rather than fast but unreliable heuristic processing (the root problem in the first place). Arguments under such a methodology would also more closely resemble the scientific method, or systems analysis in software (where one does not simply throw together an analysis and insist that it is 100% free of material flaws). So if you take Trophi as a working prototype, could we add a second phase onto an argument, where "someone smart and unbiased" goes through all the arguments, and decomposes them into some "proper" representation, essentially comprehensively aggregating all the distinct points made in the arguments into a master summary, ideally with some sort of a "properly representational structure" between all the points. Perhaps then, in a phase 3, there could be voting &/or further discussion on this summary, the results of which would be passed back through phase 2 to expand and refine the model, also noting where genuine disagreement remains (specifically where in the complex, multidimensional model), and this process is repeated indefinitely - and when some new event occurs in reality, then that could be run through this very same process, updating the multiple related models based on the results of the process. Obviously this is an incredibly ambitious idea, and is fraught with difficulties (gosh, why is reality so darn complicated!!), but even a half-assed version of it might provide some value - and, perhaps along the way, the true nature of reality might slowly start to dawn on some people, which plausibly might improve the performance of the participants in the overall process, recursively over time.
  • Nowado 6y ago
    I started with your first idea, that is (arguably, self-) referenceability of (conversations about) reality. I made a site intended to have ((('statement', 'response'), response), ... response) structure. Bunch of problems, including user input interaction. After some time that's how I explain what I'm building: https://news.ycombinator.com/item?id=25788274 https://news.ycombinator.com/item?id=25788274 if there's some way to contact you, I'd be super happy to send you an introductory video and some alpha version. For the original problem, I think the flatter the ontology, the better. I find short videos to be great explanation tools for example - 3Blue1Brown is absolutely insane. But it can be just linked in text after someone asks specific question. I think what is being attempted here is kind of doing philosophy and philosophy is done by giving responses (often: 'you are having wrong argument' responses). I've seen approach of making argument follow some traditional form (like enforcing using predefined logic to structure points), but I would like to have it as an emergent application. Extension from some subcommunity. It's just a very costly solution - it works perfectly fine when you need to explain management methodology in corporate setting, for example. Lots of resources, expensive mistakes. And I've seen some attempts at providing just that, for example (warning, in Polish) https://mysljaknaukowiec.pl/dyskusje https://mysljaknaukowiec.pl/dyskusje In a way, that is a methodology debate. Making a tool for checking 'what is' is one approach, making smaller tools that let people fail/iterate upon ideas faster is another.
    • mistermann 6y ago
      > I started with your first idea, that is (arguably, self-) referenceability of (conversations about) reality. I made a site intended to have ((('statement', 'response'), response), ... response) structure. Bunch of problems, including user input interaction. Within this process, did you have a step that decomposed each comment to extract the constituent entities and behaviors, and then append them to an overall model (which subsequent comments would refer to)? This is not an easy task at all, and expecting average users to be able to do it is likely expecting too much. However, if this is what's needed to return sanity to the world &/or vastly improve our collective understanding of reality, then so be it. > After some time that's how I explain what I'm building That's a good idea. I was thinking about something similar to that, except rather than picking items from a personal list, users could pull an idea(s) from a common "tree of knowledge and ideas" or something like that, and then frame their comment around or in the context of those (often complex) ideas*. > For the original problem, I think the flatter the ontology, the better. I find short videos to be great explanation tools for example - 3Blue1Brown is absolutely insane. Wouldn't it be nice to have a new, rich multimedia-based language for communicating complex ideas in forums, rather than being stuck with this crap, "low signal/high noise and ambiguity". I added contact info to my profile, I would love to see what you are working on - I think this space is well overdue for some innovation, and maybe I have some ideas that might fir into what you're working on.