Thriving in a Stochastic World ** Don Reinertsen
Deterministic and stochastic mindsets differ significantly.
Waterfall is poorly suited for a stochastic environment
Growth of Uncertainty
Shortened development time improves the quality of requirements
...because they haven't had time to change
Cruise missile arrives where you want it to
...because the target doesn't change
Moves at mach 4, target moves at mach 2
Target is uncertain.
Go really fast and stay pointed at the target
- Plan more
- Obviously wrong. Lack of planning is not the problem.
- Avoid variability
- Select low-variability and avoid high-variability * Less effort to implement but less positive outcomes * Example: improving grades by stopping maths/science classes and doing more phys ed classes
- Implementing a buffer
- Essentially Redefining a bad outcome as a good outcome
- Example: making a "B" grade the
- But this doesn't change
- Schedule buffers convert possible delay into certain delay
- Select high-variability but mask variability
- Adding robustness always comes at cost
* Chaos + turbulence is unavoiadble
* Example: sailing, wind is unavoidable
- You can pull in more sail to avoid
- But you will go slower and lose races * Playing it safe will mean competitors can get an edge
- Nassim Taleb - Anti-fragile
- Most of us think fragility means:
- Uncertainty = Worse Performance
- Certainty = Better Performance
- Opposite of fragile is actually a system that performs better in the presence of uncertainty
- Rooted in financial markets: options
- In options: higher variability (chaos)
- In manufacturing: lower variability
- So uncertainity combined with an asymmetric payoff function is profitable
- How to generate asymmetric payoffs in prod development
- Reduce economic damage from bad outcomes
- Increase economic gains from good incomes
- Fast feedback is required for both
- Incremental product development: small step forward with MVP
- Get a new piece of information * Proceed or Pivot * Shutdown unproductive options sooner * Add resource to productive options sooner * This is a payoff asymmetry decision!
- Idea of "buying information in small batches" is through all new tech (DevOps etc as well)
- This is the Value of Feedback
- Like betting on a horse race in which the horses are already running * Take money off horses that our out of the race * Add money to those who look promising
- Payload Management
- What is the indication of success?
- Waterfall: Meeting cost, performance, schedule targets set at the beginning of the project
- Agike: Making good economic choices "Is everything we did more valuable then everything we chose not to do?"
- Did we fully exploit the gifts of economically valuable information we were given as we went?
- Old style: try to ignore the new information because it breaks your success criteria (sticking to the plan)
- How to achieve this
- People need to understand system level economic consequences of their acitons
- Push understanding of economic consequences to a low level in the org
- Feed information down so that people can make good economic decisions
- Commanders who try to obtain certainity will fail. War is inherently uncertain.
- Uncertainity is a place where we must not only survive but thrive
- Situational awareness
- Encourage quick aassessment
- Decide with partial information (if you have 80% of the required information and haven't acted you are squandering advantage)
- Communicate intent, at least two levels deep:
- Not 'be on the hill at 10am'
- 'Be on the hill in 10 am because X is going to happen and you need to do Y so we cna achieve Z'
- Be able to shift focus quickly
- Developing Products in Half the Time
- Managing the Design Factiry
- The Principles of Product Development FLOW