Labeling of data in multiple components distributions

[nicobaga]

Current behavior

When predicting a continuous question using multiple distribution components it can be complex to fine-tune the cumulative prediction. One problem is the lack of information regarding individual components.

Expected/desired behavior

Here are a few suggestions on how to improve the system:

• Allow forecasters to label individual components: when working with multiple components, especially on questions with longer time frames, it is easy to forget what each represents. A small textbox that allows writing a small comment to identify the accompanying distribution might solve the issue.
• Display the value of weight assigned to each component: visualizing what weight between 0 and 1 is assigned to each component would allow forecasters to better fine-tune their predictions.
• Allow on/off toggling of individual components: I use data in the "My Prediction" section to adjust for individual components. To do so, I have to move the weight of all the other components to 0 and then manually move it back up later. Since individual components and weights are not labeled, this is time-consuming and error-prone. The possibility of turning components on/off without changing their values and weights would help.
• Display (at least) the center value of each component: same rationale as above. When working with multiple components it is too easy to lose track of what each represents.

[SylvainChevalier]

@aseckin thoughts on using the new binary slider designs? The numbers in the big sliders are in the question's scale.

[lsabor]

I love this, the problem I can see is that the slider doesn't actually correspond to any quartiles, so people will be extra confused thinking that the value on the center slider is their median while the chart below disagrees...

[lsabor]

Also, there would be a spacing issue for date questions.