Abnormal Sprint Termination
I did the official PSD I training course, and I'm doing sprints right now. Most of Scrum has been easy to pick up and difficult to master. However, one thought has passed my mind that I really can't figure out the correct answer to.
Suppose the theoretical situation where there is agreement that a two week sprint needs to be abnormally terminated four days in. The Goal is useless, the tasks have been superseded, etc, etc.
Scrum prescribes that a new Sprint starts immediately on the conclusion of the prior sprint. This will have one of two effects:
1. The new sprint is shorter than normal, set to be the length of the remaining days of the prior sprint (6 working days, in this example case), and therefore information like the velocity must be scaled appropriately, plus any metrics produced from the sprint must be considered inconsistent and therefore clearly marked when doing any sprint history analysis.
I personally believe 1 sounds more likely, however, there is a comment in the footnotes of version 5.2 of the PSD I official training that "All sprints have the same duration". Therefore, there is a second possibility in my mind:
2. The new sprint is a full-lengthed sprint. However, this increases general complexity as the day of sprint planning has now changed, possibly affecting stakeholder availability, and possibly affecting meeting room bookings, which is explicitly against Scrum's intention to always reduce complexity by keeping things consistent.
I would be very interested to hear what you have to say. I know that sprint terminations are very rare, and therefore unlikely to be mentioned copiously across literature.
Sprint terminations are indeed extremely rare. A fundamental shift in priority and direction of a company (i.e. - corporate merger, competitive pressure) is needed in order to render a Sprint Goal meaningless.
In my mind, you are right about predictability and consistency. This is just my preference, but if a sprint was abnormally terminated, I would keep the schedule the same, and begin the next sprint on the normally-scheduled day, same as if the sprint was not terminated.
Since there is now a "gap" of a few days between the abnormal end of the previous sprint and the start of the next sprint, that time can be spent re-assessing the situation, grooming, and preparing for the next sprint. I would not count the terminated sprint in any metrics.
Use the remainder of the Sprint Timebox to leverage and release as much value from the work done as possible. A viable goal may be reframed along with a backlog sufficient for the remainder of the timebox. This may be done regardless of how early or late cancellation occurs. Sprint cadence ought not to be affected.