Each platform's value to a user isn't fixed — it scales with how many other users are already there. A fixed pool of 1,000 users continuously re-evaluates which platform is worth more and migrates toward it.
value_i = quality_i × share_i ^ strength
d(share_A)/dt = sensitivity × share_A × share_B × (value_A − value_B) + noise
- Starting share — the initial split; start it near 50/50 to watch small random fluctuations decide the winner.
- Network-effect strength — how superlinearly value scales with user count. Higher means the leader's advantage compounds faster.
- Migration sensitivity — how quickly users act on a value gap.
- Quality A / B — each platform's underlying feature quality, independent of user count.
Because value depends on both quality and share, a big enough quality edge for the smaller platform can still out-value a larger rival — but only if it's large relative to the network-effect strength. Otherwise the early leader's compounding advantage wins.