How does cadence and team-lead assignment work?

[jeffreywescott]

Context

@shereefb and I stumbled across this issue while thinking through some software stuff. The problem is that team leads need to be available to later cohorts on the same "interval" / cadence.

This is incredibly complex. My recommendation is to punt on "dynamic cycles" for v0 (maybe forever?), and instead come up with one "master schedule" that is at least 40 weeks long that ensures that team leads will be available when needed.

Criteria

Define the success criteria for this issue. What must an acceptable solution include?

Principles razor

In order for an issue to be marked as passing, the proposed solution must be aligned with our principles. Check off the principles that this solution is consistent with.

• [ ] Cultivate a learning community.
• [ ] Prioritize scale over quality
• [ ] Align incentives
• [ ] Build a team game
• [ ] Trust people's intentions and potential
• [ ] Enable self-organization & self-determination
• [ ] Map feedback loops
• [ ] Optimize for agility
• [ ] Simulation not preparation
• [ ] Sustain the tension between individual and collective gain

[tannerwelsh]

Related to #126

[tannerwelsh]

Proposal

Assumptions:

• Assume pods of 10-15 (learners demarcated A-O)
• Assume phase = 5 weeks, cycle = 1 week

Phases 3-8 (and beyond) look like this:

• Cycle 1: A, F, (K) team lead, rest team members
• Cycle 2: B, G, (L) team lead, rest team members
• Cycle 3: C, H, (M) team lead, rest team members
• Cycle 4: D, I, (N) team lead, rest team members
• Cycle 5: E, J, (O) team lead, rest team members

Phases 1-2 look like this:

• Cycle 1: all are team members
• Cycle 2: all are team members
• Cycle 3: half work in teams, half go solo
• Cycle 4: half work in teams, half go solo
• Cycle 5: all are team members

Rationale

• Keep things simple. After phase 2, every phase has the same rhythm.
• Allow time for solo work in first 2 phases to avoid "cruising" on team members' skills

[jeffreywescott]

Brain-dump from my meeting with @shereefb around this.

Constraints

• 3-5 person teams (including lead)
• Cycles must all be divisible by some “sprint” unit (e.g., 1 week, 4 hours)
• Cycles are powers of 2 “sprints” (e.g., 1, 2, 4)
• Lead / led ratio target: 25%

Lead Choosing Algorithm

• At end of cycle, leads are selected from the cohort(s) ahead who are:
  • UNASSIGNED
  • Have the lowest ratio of time leading to time led (LMR)
  • Have the worst(?) relative contribution delta (RCΔ)
  • Enough leads to maintain ~ ¼ ratio of leader to member

Team Assignment Algorithm

• Start assigning teams from newest cohorts first (e.g., week 1 not week 39)
• Target 4-5 players per team (including lead) (see algorithm)
• Choose N leads using Lead Choosing Algorithm (above) where N = # of teams
• Each player gets a certain # of votes: # of teams
  • Can spend multiple votes on a single goal
  • Can change votes assigned to each goal and which goals voted on
• Voting ends at HH:MM:SS.NNN
  • Select all goals that have received votes from at least 2 people
  • Within that, see which goals got the most votes
  • Choose maximum of N goals, where N = # of teams
  • Aim to assign people to the thing they spent the most votes on

Parking lot

• What happens if we don’t have enough leads?
• Take into consideration my (LMR - Lead to Member ratio) when you’re thinking about assigning me half a lead
• Assisted Independent study? Drill and assist?