Last week I met with Brendan Baker to exchange notes on venture capital observations — and catch up with him on joining Rackhouse as Partner. Brendan asked me why I think most GPs fail – quite an open-ended question, and I took it as “why do most GPs fail to raise more than 1 fund.” … [“fail” is quite a broad spectrum and depends on your personal values].
I immediately defaulted to Integrity: (1) Say what you think and (2) do as you say. Integrity is ethics-free. It has nothing to do with absolute “good” or “bad”. It merely is self-consistent behavior and actions according to the principles or values or desires or aspirations you articulated.
Brendan paused for a second and said: “I think it’s trust”.
We went off into another direction about LPs after that, but something kept nagging me: I wasn’t able to articulate the relationship between Trust and Integrity, but (as to be expected) Brendan is right that Trust is encapsulating a relationship aspect that Integrity alone is missing: You can act with high integrity all by yourself, without anyone watching. Trust includes a response that forms an implicit social contract. We feel betrayed when someone breaks our trust – they broke an implicit social contract. But it doesn’t end there.
Trust Issues
And therein lies my problem, what kept nagging me: I make a difference between personal trust and professional trust, which perhaps is different from most people. I have the impression that people often use Trust as in “I trust that they will do nothing to harm me!”; or “How could he do that to me, I trusted him!”. When I trust someone professionally, it means that I thought I had enough evidence to know how a person operates and that will continue to operate that way in the future (= Integrity over time). Trust is a prediction, not a feeling. That also means that they might act according to their ethics in a way that could be negative for me personally, both in the short term or the long term, but they acted fully self-consistent. That’s why I can trust and work with people across a wide political spectrum. I’d rather not help my own demise, of course, but I also might have committed myself to a mission or contract, and I will honor that commitment (or excuse myself).
Trust is broken when someone acts differently from what we predicted. That’s on us. We might just suck at predicting. Perhaps there is an information asymmetry or a narrative break: We have the same facts but come to different conclusions. A break of trust is a lack of my understanding of your motivations or narrative.
Trust in Venture Capital
Trust in VC is not a static belief—it’s a multi-temporal signal that accelerates decision velocity without sacrificing judgment. It is built in moments of uncertainty, tested in capital asymmetry, and revealed in post-term sheet behavior.
- Relational – Longitudinal consistency of intent and follow-through (with founders, co-investors, LPs).
- Epistemic – Confidence in the other’s judgment under uncertainty and absence of information.
- Procedural – Transparency, clarity, and fairness in processes (term sheets, board governance, conflict resolution).
There is a fourth dimension that’s special to venture capital acting under uncertainty, and under pressure, and under vast choice: As VCs, we chose with whom we work. I select startups and LPs who share my most important values and do not fatally violate my ethos. E.g. we can co-exist and work together without killing each other and both achieve our goals.
4. Moral – Values alignment, especially under pressure (e.g., protecting founders during layoffs, cap table cleanup).
Trust Framework: T3i Score
T3i is a synthetic metric combining:
- T1: Tactical Trust – Has this person done what they said? (Execution fidelity)
- T2: Temporal Trust – How have they behaved over time? (Consistency under pressure)
- T3: Transfer Trust – Do others independently vouch for them? (Relational references)
- i: Integrity Signal – Have they made hard choices for the right reason, perhaps even taken reputational or professional risks? (Skin in the game when not required)
Each scored 1–5. Weighted average, but i-factor is a multiplicative integrity gate (i.e., it’s binary: if violated, total trust collapses). Example (1-5 scale):
- Tactical: 4
- Temporal: 3
- Transfer: 5
- Integrity: 0 (breach)
→ Trust = 0 despite strong performance. No integrity, no trust.
T3i + Predictive Layer
Trust is the calibration error between what you expect and what actually happens. A good trust model isn’t about hoping people are good — it’s about forecasting what they’ll do under specific stressors, constraints, and incentives.
[Thinkstorm Thesis]
Overlay a confidence score for each prediction:
| Component | Score | Prediction Accuracy (0–1) |
|---|---|---|
| T1 | 4 | 0.75 (3/4 forecasts accurate) |
| T2 | 3 | 0.50 (2/4 cycles behaved as expected) |
| T3 | 5 | 0.80 (reference predictions held under stress) |
Aggregate trust score weighted by prediction accuracy:
Where:
- T1, T2, T3 are the original scores
- A1, A2, A3 are the respective Prediction Accuracies
Apply the i-gate at the end:
If i = fail → Trust Score = 0
Instead of asking “Can I trust them?”, I’m constantly refining: “How well do I understand what they will do next?”, and more importantly: “What signals improve or degrade my forecasting skill?”