📦⚙️🫀product process organization - projection, (G, E)

[hyunjimoon]

estimation P()📦⚙️🫀product process organization - projection, (G, E)
#12 $P_A(\theta|y), P(\theta, y), \tilde{y}$
can we make an analog from each model to organization: $\left.p(\theta, y), p_{\mathrm{A}}(\theta \mid y)\right) \quad\left(p(\theta, y), p_{\mathrm{A}}(\theta \mid \tilde{y}), \tilde{y}\right) \quad(\quad p(\theta, y), \tilde{y}$

PA modeling
If model structure is to follow real-world structures underlying the problem, qualitative data is needed for building it
Insights can be generated before starting to use numerical data
Save lots of time on collection, preprocessing and calibration
Potentially reallocate to client interaction, robustness testing and scenario experimentation
Less cognitive load for participants

PAD modeling
Uncertainties in parameter values can only be settled with empirical data
Many theoretical and practical implications depend on these parameters, e.g. any cost-benefit requires reliable estimates of both costs and benefits, and estimates of uncertainty around both
No learning about the data, or from the data directly
No contribution to model quality from tests against data
Hard to verify that asserted reference modes or decision structures match reality
Less face validity of historical runs
Difficulty understanding the gaps between a priori parameter values and most plausible values, given the model
No objective basis for parameter values or confidence bounds
Hard to understand the joint uncertainty of a parameter set

https://trello.com/c/hl12zijg

[hyunjimoon]

algorithms each people have is (culture, mbti? etc)
people (system1, 2: capability spectrum: opt, vi, mcmc / pathfinder: warmup),

exploration and exploitation in org (complex, high uncertainty space) learning

funnel: scale problem (mapp this to charlie's loop) - loosening the correlation? (sigma and )

during exploration (warmup), we learn metric (does each chain have different metric?)