This keeps it modular, explainable, and tunable.
Where:
- (
$A_i$ ) = Count or intensity of activity type i - (
$W_i$ ) = Weight for that activity
| Activity ( |
Description |
|---|---|
| ( |
Tasks completed (Jira) |
| ( |
Code commits |
| ( |
PRs merged |
| ( |
Meetings led |
| ( |
Design docs / proposals |
| ( |
Mentoring / reviews done |
Where:
- (R) = Peer review score (0–1 or 1–5 normalized)
- (S) = Stability / success rate (bug-free releases, acceptance rate)
- (
$\alpha, \beta$ ) = Tunable impact factors
📌 Ensures quality > quantity
Where:
- (K) = Knowledge contribution index (helping teammates, answering Slack threads, docs, unblocking others)
Where:
- (U) = Peer kudos / endorsements
- Log prevents popularity bias
- (
$\delta$ ) controls recognition influence
Where:
- (
$D$ ) = Delays caused - (
$F$ ) = Rework or rejected submissions - (
$R_w$ ) = Repeated unresolved blockers - (
$\lambda$ ) = penalty weights
🚨 Penalizes friction, not silence
Where:
- (I) = Business impact (revenue, customer impact, risk reduction)
- Role-agnostic (engineer, manager, designer)
- Resistant to gaming
- AI-friendly (features from Slack, GitHub, Jira)
- Explainable to HR & leadership
- Encourages collaboration over noise