Best teams I’ve worked on laughed a lot. Not polite chuckles in standups — actual laughs. The kind where someone unmutes just to wheeze.

I noticed a pattern: the laughs compound fast early on. A new team that cracks jokes in week one builds trust faster than one that stays formal for a month. But after a point, more laughs don’t make the team better. You plateau. The trust is already there.

That’s a saturation curve.

The model

Borrowed from enzyme kinetics (Michaelis-Menten), applied to teams:

Q(L) = Q_base + ΔQ_max * (L / (K + L))

Four terms:

• Q_base — Baseline technical skill of the team. What you ship with zero rapport.
• ΔQ_max — Maximum boost from positive team chemistry. The ceiling above baseline.
• L — Number of genuine laughs (shared moments of levity).
• K — Half-saturation constant. The number of laughs needed to hit 50% of the max culture boost.

When L = K, you’re at half the potential boost. When L >> K, the boost asymptotes to ΔQ_max.

Why saturation, not linear

If laughter scaled linearly with quality, stand-up comedians would ship the best code. They don’t.

Linear:      Q = Q_base + c * L         (unbounded — nonsensical)
Saturation:  Q = Q_base + ΔQ_max * L/(K+L)  (bounded — matches reality)

The saturation model captures two things:

1. Early laughs matter most. Going from 0 to K laughs gets you halfway. Going from K to 2K only gets you another 17%.
2. There’s a ceiling. No amount of laughter substitutes for skill. Q_base is doing the heavy lifting. ΔQ_max is the multiplier, not the foundation.

Plugging in numbers

Say a team has baseline skill of 70 (out of 100), max chemistry boost of 20, and K = 10:

L=0   → Q = 70 + 20*(0/10)   = 70.0
L=5   → Q = 70 + 20*(5/15)   = 76.7
L=10  → Q = 70 + 20*(10/20)  = 80.0   ← half the boost
L=20  → Q = 70 + 20*(20/30)  = 83.3
L=50  → Q = 70 + 20*(50/60)  = 86.7
L=100 → Q = 70 + 20*(100/110)= 88.2   ← diminishing hard

The first 10 laughs buy you 10 points. The next 90 buy you 8. That’s the shape of trust.

The K problem

K is the interesting knob. Low K means the team bonds fast — a few good laughs and you’re there. High K means the team takes longer to gel.

Remote teams have higher K. You need more shared moments to build the same trust because you’re missing ambient laughter — the overheard joke, the hallway bit. Slack threads don’t carry tone well. Video calls compress social bandwidth.

Co-located teams have lower K. Laughter is contagious in physical space. One person laughing triggers others. The same number of funny moments generates more L through amplification.

What this doesn’t model

• Forced fun. Mandatory team-building events increase L on paper but not the kind that builds trust. The model assumes genuine laughs.
• Toxic humor. Laughing at someone isn’t the same as laughing with them. Negative ΔQ_max is a thing.
• Personality variation. Some people bond over quiet intensity, not laughter. K → ∞ for them, and the boost comes from a different variable entirely.

The takeaway

You can’t engineer laughter. But you can stop killing it. Skip the “let’s keep this professional” defaults. Let the weird joke land. Let the tangent happen.

The first few laughs do more work than the next hundred.

Quality(team) ∝ number(laughs) — but only up to a point. After that, ship the code.

😂