I’ve been building a browser-based general-purpose Influence Engine tool for mapping beliefs - or any scalar values - and seeing how changes propagate through a network. It’s meant for situations where you want to make your reasoning structure explicit, whether for legal investigation, debate analysis, or structured forecasting, but without...
Suppose you want to help non-experts fill out the probabilities for a simple Bayesian network. For each link, you need them to specify the chance of a child node being true depending on the parent node’s state. What’s the most accurate and psychologically easy way to ask for these numbers?...
I'm a professional investigator. My standard of evidence is "more likely than not." I think it would be helpful, especially for collaborative purposes, to have a visual reasoning tool inspired by Bayesian thinking. Users could map out individual beliefs, assign probability and confidence levels, and link them with directional influence...
I’ve been building a browser-based general-purpose Influence Engine tool for mapping beliefs - or any scalar values - and seeing how changes propagate through a network. It’s meant for situations where you want to make your reasoning structure explicit, whether for legal investigation, debate analysis, or structured forecasting, but without needing to manually run Bayesian equations every time you tweak something.
The tool has two modes (and allows for active toggle):
Suppose you want to help non-experts fill out the probabilities for a simple Bayesian network. For each link, you need them to specify the chance of a child node being true depending on the parent node’s state.
What’s the most accurate and psychologically easy way to ask for these numbers?
Is it better to ask directly for two conditionals (e.g., “How likely is X if Y is true? How likely is X if Y is false?”)?
Or to ask for a “baseline” probability for X, and then how knowing Y is true or false would shift that estimate (i.e., a base rate and two “deltas”)?
Or is there a smart way to fold all that into... (read more)
Adele, that's awesome. I'll definitely use it next time I need Bayes-level precision.
I've been working on this quite a bit. It's only Bayes-lite though: it blends beliefs using weighted averages in log-odds space and outputs the robustness and probability of the recalculated beliefs. That should be fine for my purposes and for layman use.
I've got the rawest of functional prototypes. But the hard part is done - the math checks out. It even propagates complex loops without breaking. It will take me a while to develop the UI though. Here's a screen grab. Not too pretty yet. I'll let you know when I really have something worth sharing.
Thanks, I'll keep digging. You've already helped just by suggesting better search terms (e.g. “mind-mapping software").
That said, I’d still be surprised if someone hasn’t already built something close to this.
Slight tangent: About 15 years ago, I had what I thought was a novel app idea: an audio recorder that constantly buffers the last 60 seconds of sound, so if someone said something important, you could hit a button and retroactively capture it. I looked it up, and sure enough, it existed: fully developed but mostly abandoned. Same thing happened when I imagined vending machines at public tennis courts that dispense fresh tennis balls. Already patented, blueprint and all. I decided we'd reached the IP singularity (15 years ago!): naming is almost equivalent to invoking.
So unless my idea is downright incoherent or useless, I have to assume it’s already out there somewhere buried under the right search terms.
I'm a professional investigator. My standard of evidence is "more likely than not." I think it would be helpful, especially for collaborative purposes, to have a visual reasoning tool inspired by Bayesian thinking. Users could map out individual beliefs, assign probability and confidence levels, and link them with directional influence weights to reflect how one belief affects another. As evidence changes, the map would update, making belief revision explicit and interactive. It would be a lightweight, intuitive Bayesian network that could be intuitively understood by a layperson. It would be shallow and imprecise compared to formal models, but grounded in the same principles. It occurred to me that such a tool would be somewhat universal. One could use it for apple picking just as easily as sex crimes investigations, which makes me think that this must already exist. But I can't find it. I would be so grateful if anyone could point me in the right direction.