[aidenn0]

Lack of a plan is what results in bullwhip effect; the bullwhip effect is caused by each silo iterating on its local knowledge...

> in contrast, "workflow" is akin to kanban in this context. you start with constraints (in this case time and money) and then design the system to those constraints. mary, the speaker, mentions that they had 4 different, decoupled workflows, which helped them avoid those pesky cascading delays. workflows are process oriented (repeatable events), so steel construction, for example, was thought of as an separate repeatable (if varying) process (swimlanes, in kanban parlance) as they went up in height. kanban also focuses on realtime learning and adjustments as well as just-in-time inventory systems (important to steel being delivered on time, like using 2 different suppliers to make sure there were no delays).

That sounds like a plan to me...

[clairity]

no, the bullwhip effect is caused by uncertainty across coupled tasks along the critical path, no matter the level of detail of your plan. it's a statistical phenomenon that you can mitigate with other measures, like critical chain. the empire state building example shows that they reduced coupling by creating 4 separate, independent workflows, mitigating the bullwhip effect by roughly a factor of 4.

[aidenn0]

Yes, the bullwhip effect is caused by uncertainty across coupled tasks. A plan that doesn't reduce and manage uncertainty is bad. Not having a plan means there is uncertainty even when all of the information is there to determine the answer. The empire state building had a plan for obtaining steel from foundries at the velocity they needed.

If one defines plan to mean "We schedule everything up front and then stick our fingers in our ears and say 'la la la la' when reality conflicts with our imagined schedule" then of course plans are bad. Similarly if one were to define agile as "a system that is incapable of delivering any feature that takes more than 2-4 weeks to develop" then agile would be bad too.

[clairity]

yes, another restatement of the distinction between 'plan' and 'workflow' as used in the talk.