danielpcox/belief

Prototype of a searchable, browsable, editable accounting of your own reasoning

1. npm install

2. npm run watch

3. open index.html in your browser.

belief's Issues

Propositions should have English description

Implement save/load to CSV

“I'm thinking we make the save format a csv that excel can import, to kill two birds with one stone: save/load and import/export.”

Cancelling in the 'How many times more likely is the target if the source turns out to be true?' prompt results in NaN connection and descendant probability values

Implement basic backing data structure

Implement Pan/Zoom

The paid JsPlumb toolkit has this feature. We may need to reimplement.

Implement Causal Network

Something intelligent here

Priors should be editable

In case of typo

True/False proposition end hooks should scale with probability

Assuming each proposition has a true and false hook, it may be instructive to the user to scale the hooks by their respective probability, if available.

Statements should not be allowed to connect circularly

Spec Out Data Model

What constitutes a proposition?
What kinds of cards will we need?

Add ordered listing of displayed propositions

On start, create a sequence of 3 connected propositions

Clicking on proposition label/title/text should not initiate dragging

Connector line should illustrate carried probability value and direction

Should probably indicate via red color (or perhaps dashed line?) if probability is <50%.

Should probably indicate direction via arrows.

Propositions should compute a probability and present for display and output to dependent propositions

Propositions should accept many probability inputs

Map out “she’s a witch” syllogism

Define MVP feature set

Once we’ve figured out a data model, we should define scope of work for the MVP, so we know what we’re getting into.

Allow grouping of logical units

I can imagine a lot of cases where a user may want to group a set of logical steps into a single block. Maybe for re-use, maybe for a more intuitive diagram.

Could we summarize sets of propositions somehow? Have them link to other maps with a clear input and output to encapsulate logic?

Click of a card connection should create a new proposition card bisecting the connection

Double-clicking a proposition card should not create a new proposition

Alt-drag of a Proposition Card Should Duplicate Card

And card should follow mouse until mouseup.

Update content on new connection

“The next thing I want to do is figure out how to update content associated with each box when certain connections are made, so we can actually run calculations on the network.”

If cards have interactive internal elements, dragging affordances need to be clearer vs. internal interactions

Proposition cards should display their relative significance

I’m picturing a single core proposition that has a relatively large impact on many descendent/child/posterior propositions, and using scaling as a way to illustrate that to the user, and perhaps motivate greater scrutiny and pursuit of evidence for the most weighty props.

For example, “Birds squawk at x decibels” may be true or false, but far more of my life and behavior depends on this: “People other than myself exist.”

Limits may be necessary to prevent some from growing too large or small.
Typographic changes, color changes, opacity, etc., may be better methods than scaling.

Add ability to delete an unwanted proposition

Implement Visual Value manipulation

Something intelligent here

Implement mutually-exclusive proposition groups

To allow for better speculation and exploration of ideas and their consequences...

Propositions should have float to designate prior