Apply HN: Wikipedia for how, and how well, we know what we think we know | Hacker News Reader

Apply HN: Wikipedia for how, and how well, we know what we think we know

3 pointsjnicholasp6 comments

I want to combine Wikipedia, argument mapping, and prediction markets to make a collaborative website for productive large scale argumentation, that will collect and organize all the world's inferential knowledge into a single vast graph, visually show the logical relationships between ideas as connections between nodes on the graph, allow for concepts to be iteratively clarified and improved on, and quantify the degree of support (in both the rational and democratic senses) that every individual idea has. The exponential curve of advancing understanding that we are on is driven by the increasing spread of more reliable knowledge to more participants working from a more complete database: the printing press + the scientific method + the university + the library has given us the modern world. Spreading and examining knowledge more reliably and widely will enable more minds to contribute more usefully, and will bend the curve even more sharply upward. We can do this by creating a collaborative graph-building sandbox in which users can create and place nodes (each containing a single idea), and connect them to other nodes with virtual elastic threads to show the logical connections between them: a node upstream of node X is an idea that X depends on, and a downstream node is an idea that depends on X. Users can vote on the connections they think are appropriate for X by drawing threads of their own: connections that most users endorse will pull closer together. Nodes containing the same idea stack to occupy the same place in the graph, and user ratings of the strength and clarity of each version of the idea float the best and clearest statement to the top of the stack. Users can't edit other users' statements, but voting on which ones are best allows us to improve on weak versions of decent ideas without making us settle for Wikipedia's committee style production of a single article and the edit wars that go with it.

Good idea. I'm working on something related (knowledge graphs) if you want to talk about it. Do you think it would require users to understand formal logic?

Do you have an idea how to turn this into dollars? Donations like Wikipedia?

The catch with up-voting is that it doesn't produce the same results as expert curation. I assume that's why Wikipedia relies on editors instead of voting.

This site wants to be Wikipedia for inferential knowledge - where Wikipedia is a general encyclopedia and catalogs knowledge of all kinds (people, places, things, events, ideas, etc.), this site deals only with ideas that are <i>arguments</i> or <i>claims</i>, and that depend for their plausibility and truth on chains of inference stretching back ultimately to some fundamental axiom. This includes essentially all of philosophy, science, religion, and politics, since all of these make debatable claims about what reality is and how it is likely to respond in given circumstances, and since those claims are all based on some chain of prior reasoning every link of which are themselves debatable.

We need a Wikipedia-like central resource that explicitly graphs all of these chains of reasoning, reduces them to their individual links, and lays bare their connections and branchings, so that we can 1) understand the underpinnings of our own ideas, 2) understand the underpinnings of <i>other peoples's</i> ideas (i.e., stop assuming that other people are stupid strawmen), 3) show definitively that genuinely stupid ideas are genuinely stupid, by systematically meeting their bad arguments with clearly stronger counter-arguments, 4) stop having the same arguments constantly all over the internet/tv/congress/world, by having it all out in one place where the argument can be logically organized, permanently recorded, and exhaustively fleshed out, and 5) finally be able to walk through, clearly and completely, exactly what gives us reason to think that something is true.

So apparently I should look at my comments after I post them, so I can edit the bad formatting.

Also, is there a HN formatting guide somewhere that will tell me how to do italics, bold, make links, etc.?

Disclaimer: I'm not really eligible for a YC Fellowship at this point, as I'm not a hacker and don't have a technical co-founder, and therefore can't create the proposed site. This post is mostly for generating discussion, and for the outside chance of finding someone who sees the potential and wants to come on board with me to make it happen.

Your paragraph of text is making my eyes glaze over.

Wikipedia is popular because it is easy to use and shows up in search results. However it's not clear that your thing can or will be the same. I need a concrete example of how your thing will help me in order to be able to give further advice on the idea.

You're right about the formatting, oops. There are supposed to be three paragraphs there, but I'm not familiar with HN's formatting, and I didn't think about it after I posted it.

I would say that Wikipedia became popular, and eventually earned itself a high place in search results, because it is easy to use, yes, but mostly because it provided a place/format for useful information that couldn't be accessed anywhere else as easily. It doesn't mean much to say, "Oh, my site will be easy to use too!" - that's just an unsubstantiated claim at this point. But I don't think that's a hard target to hit.

On the other point (providing a place for useful information that can't be easily gotten elsewhere), once you have a clear idea of what the site does, it's immediately apparent that the information it provides is extremely useful, and that there's nowhere else to get it.

I'll try to give a couple more concrete examples: in politics, how often does it appear to be the case that people understand why their opponents believe the things they do? How many people can give a genuinely strong argument for the opposing view, vs. how many people are unable to give a plausible argument for the other side and tend to assume that their opponents think what they do simply because they're stupid/selfish/malicious/etc.? Our political process is highly polarized, and there is almost no reasonable, informed political debate, certainly not anywhere that is prominent and accessible, and this is extremely detrimental to the project of trying to run a well-functioning democracy. Wikipedia made it not possible for reasonable people to be trivially wrong about matters of fact, by making it simple to find out what is right. We need to do the same thing for arguments, by making it simple to look up what is strong and coherent.

In science, suppose you're a biochemistry grad student trying to get a handle on everything we think we know about the biochemistry of dementia. Currently this means reading tons of uncorrelated papers scattered across dozens of journals, and attempting to assemble in your own head a logical structure that ties all of them together, sussing out and keeping track of thousands of relevant bits of data about who knew what when, who references who, how strong each paper's individual arguments are, which points undermine which others, which ones strengthen which other ones (and how much), what are the strongest known approaches, where the open questions are, etc. Every expert in a given topic has in their head their own ad hoc version of this logical structure, and all of them vary from each other. We will gain immensely by combining all of those implicit maps into one explicit map, that will be more complete than any individual's, improvable, allow quantified collective judgments on uncertain points (a la prediction markets), and which will be vastly simpler to acquire for new would-be experts.

Do those examples make the site's function any clearer for you?

Reply on news.ycombinator.com