From 58ee7b6caee06506332e30cee86a2c769a28b3a1 Mon Sep 17 00:00:00 2001 From: Joseph Flinn <58369717+joseph-flinn@users.noreply.github.com> Date: Mon, 5 Feb 2024 10:31:55 -0800 Subject: [PATCH] [post] Question Your Assumptions: initial commit (#92) --- .../0026-e2m-st-new-devops-perspective.md | 2 +- .../0027-st-question-your-assumptions.md | 91 +++++++++++++++++++ frontend/src/lib/assets/data.json | 5 +- 3 files changed, 94 insertions(+), 4 deletions(-) create mode 100644 data/posts/0027-st-question-your-assumptions.md diff --git a/data/posts/0026-e2m-st-new-devops-perspective.md b/data/posts/0026-e2m-st-new-devops-perspective.md index 516d90d..17a4ef4 100644 --- a/data/posts/0026-e2m-st-new-devops-perspective.md +++ b/data/posts/0026-e2m-st-new-devops-perspective.md @@ -17,7 +17,7 @@ it means. With this in mind, what does DevOps mean to you?" Unsurprisingly, each Since reading Accelerate [[1](https://itrevolution.com/product/accelerate/)] a few years ago, my working for any discussion around DevOps has been: _"DevOps is a collection of engineering practices and methodologies to deliver more -value, faster, in a more stable way to the end user."_ I have not been satisfied with this definition ore perspective +value, faster, in a more stable way to the end user."_ I have not been satisfied with this definition or perspective and recently I have found it lacking. A quick follow up question asking about the specific engineering practices and methodologies can quickly lead to a opinion heavy debate over Trunk Based Development vs GitFlow, Continuous Delivery vs Continuous Deployment, and how much testing is needed, where it's needed and what needs to be tested. diff --git a/data/posts/0027-st-question-your-assumptions.md b/data/posts/0027-st-question-your-assumptions.md new file mode 100644 index 0000000..7424abf --- /dev/null +++ b/data/posts/0027-st-question-your-assumptions.md @@ -0,0 +1,91 @@ +!! title: Always Question Your Assumptions +!! slug: st-question-your-assumptions +!! published: 2024-02-05 +!! description: Double-loop learning and questioning assumptions. + +--- + +Training dogs holds a very special place in my heart. There is very little that brings me more satisfaction and joy than +to work with and train an intelligent canine. I have found that consistency, almost to an extreme, is the key of +developing and maintaining desired behavior. + +The traditional learning loop that is proposed and studied extensively in psychology for learning for both animals and +humans is single-loop learning. An agent is placed into a world, receives feedback information, makes a decision, +experiences the feedback from the decision, receives more feedback to make another decision. And so on and so forth _ad +infinitum_. + +> Single-loop learning does not result in deep change to our mental models--our understanding of the casual structure of +> the system, along with the boundary of the model (which variables are included and which are excluded) and the time +> horizon we consider relevant--our framing or articulation of a problem (Sterman, 2000) + +The famous Skinner Box experiments shows this well. A rat is placed into a box that has at least one lever connected to +some feedback. Depending on the research question, the lever could deliver different types of feedback: positive +reinforcement, positive punishment, negative reinforcement, and negative punishment. The overall idea of the Skinner Box +is to test different operant conditioning research questions in isolated and controlled environments (Nickerson, 2024). +After a few times in the box, rats show the behavior of running over and immediately pressing the lever. From a +single-loop learning perspective, the box is the world and throughout the experiment stays constant. The rat is the +agent. The agent presses a lever and receives feedback and quickly builds a mental model of how the world works. The +world is held constant, feedback is delivered, decisions are made and behavior is formed and executed with continued +experimentation. + +> We are taught from an early age that every event has a cause, which in turn is an effect of some still earlier cause: +> "Inventory is too high because sales unexpectedly fell. Sales fell because the competitors lowered their price. The +> competitors lowered their prices because ..." Such event-level explanations can be extended indefinitely, in an +> unbroken Aristotelian chain of cause and effects, until we arrive at some First Cause, or more likely, lose interest +> along they way (Sterman, 2000). + +In a different Skinner Box experiment, Skinner experimented with food delivery regular interval food delivery with +pigeons and found that the pigeons associated whatever behavior they were doing right before the food delivery with the +appearance of food (Skinner, 1948). A cause-effect relationship was formed between behavior and reward even though there +was no relation between the cause and effect. Anthropomorphically, the pigeons assumed that it was from the event of +their behavior that caused the event of the food appearance and formed a superstition. A mental model was formed of the +world that it was their behavior that caused the food to appear. However, when the regular interval of the food delivery +was stopped, it was observed that the pigeon's learned "superstitious" behavior deteriorated. When the regular food +delivery started back up, a completely new behavior was "learned" at the same rate as before. + +Single-loop learning assumes that the world is constant and not changing. The Skinner Box doesn't change...until it +does. In the pigeon experiment, the world doesn't stay constant as it did the experiment with the rats. The period of +the appearance of food changes and so does the observed behavior of the pigeon's mental model. From the anthropomorphic +perspective of the pigeon, the pigeon receives positive reinforcement feedback for a certain behavior when the food +appears--even though there is no relation--so it decides to keep doing that behavior. The feedback from the system +continues. When the feedback stops--the regular intervals of food delivery stops--the behavior continues but +deteriorates over time as the perceived feedback stops. The feedback that indicates that it is not the behavior that +causes the food updates the pigeon's mental model in relation. When the regular interval food delivery is resumed, +a new cause-effect relationship starts, even though the delivery is still not related to the behavior. In the studied +cases, a new cause-effect "superstitious" behavior was observed (Skinner, 1948). + +In double-loop learning, the feedback doesn't just affect the decisions in the world. Feedback also updates the mental +models, decision rules, decision making strategies, and the system itself (Sterman, 2000). The resulting behavior from +the system is different than before because the system is different than before. + +> Systems Thinking is double-loop learning: information feedback about the real world no only alters our decisions +> within the context of the exiting frames and decision rules, but also feeds back to alter our mental models. As our +> mental models change, we change the structure of our systems, creating different decision rules and strategies. The +> same information processed by different decision rules and strategies now yield a different decision (Sterman, 2000). + +There is are lists of things that impede learning at each point in double-loop learning: the real world is very complex +and difficult to reason about with accuracy, selective perception, biases, ambiguity, misperceptions of feedback, +unscientific reasoning (even by scientists), judgemental biases, defensive routines, inconsistent decision making, and +performance as goal are just some. These all have something in common: a strong underlying bias that currently held +assumptions are more accurate that something else, even in the face of disconfirming evidence of the held assumptions. + +Identifying and questioning all assumptions is paramount to learning in complex systems. As we try to work towards a +specific system goal, have we defined the boundaries well enough? Is there a variable that we have intentionally or +unintentionally not included in the model that has a large effect on the behavior of the larger system? What is the +stated goal of the system that we are studying? Does the behavior of the system move towards that goal, or is there a +goal that it is moving towards in action that is different than the stated goal? + +When we start questioning all of our assumptions, we allow flexibility in our mental models to conform to the observed +world instead of demanding the world conform to our mental models. Our perspectives shifts. It becomes easier to relate +to other humans because our mental models of the world remain fluid. + +> Each participant in a conversation employs a different mental model to interpret the subject. Fundamental assumptions +> differ but are never brought into the open (Sterman, 2000). + +--- + +## Resources + +1. [Charlotte Nickerson; _Skinner Box: What is an Operant Conditioning Chamber?_; 2024](https://www.simplypsychology.org/what-is-a-skinner-box.html) +2. [B. F. Skinner; _'Superstition' in Pigeons_; 1948](https://psychclassics.yorku.ca/Skinner/Pigeon/) + diff --git a/frontend/src/lib/assets/data.json b/frontend/src/lib/assets/data.json index 27f9ed0..8cecf6c 100644 --- a/frontend/src/lib/assets/data.json +++ b/frontend/src/lib/assets/data.json @@ -9,14 +9,13 @@ "skillsData": { "leadership": [ "curiosity first", + "systems thinking", "psychological saftey", "organizational alignment", - "strategic thinking", - "systems thinking" + "strategic thinking" ], "poeple management": [ "psychological saftey", - "output oriented", "high performance: purpose + expectations + feedback", "effective feedback with the PSBIQ model", "situational leadership: delagate, support, coach, or direct"