Vision-Only Decision Framework
Start from physics—can the fundamental sensor do the job alone? If yes, delete the crutches
When deciding whether a system needs multiple redundant components (sensors, modules, subsystems), start from physics: can the fundamental component do the job alone? If physics says yes, then additional components are crutches that add complexity, cost, and supply chain risk. Make a bold decision to delete the crutches, enforce it, but remain open to evidence—if a superior technology emerges later, revisit the decision. The framework balances decisive action with intellectual honesty.
- Start from physics: can the fundamental component do the job alone?
- Treat additional components as crutches that add complexity, cost, and risk
- Be willing to make a bold, binary decision and accept short-term risk
- Remain open to evidence—if better technology emerges, revisit the decision
- Being stubborn about the vision does not mean being closed to new data
- Identify the fundamental capabilityDetermine whether the core component (camera, primary sensor, main algorithm) can theoretically do the job alone based on physics.Pro tipMusk's reasoning: humans drive with only eyes. Therefore cameras should be sufficient for driving. The physics supports it.WarningBe honest about whether the physics actually supports the simplified approach. Do not confuse what you want to be true with what is true.
- Evaluate the cost of redundancyCalculate the full cost of additional components: hardware cost, supply chain complexity, integration engineering, failure modes, and maintenance.Pro tipLiDAR systems cost thousands per car and created supply chain vulnerabilities. The redundancy cost was not just financial—it was strategic.
- Make the binary decisionIf physics supports the simplified approach and the redundancy costs are high, make a decisive call to delete the redundant components.Pro tipMusk paused for 40 seconds of complete silence before deciding. Take time to decide, but then commit fully.WarningIf your team strongly disagrees, hear them out. But distinguish between physics-based objections and comfort-based objections.
- Enforce the decision but allow quiet R&DImplement the decision organization-wide while allowing continued research into alternatives. Being decisive does not mean being permanently closed-minded.Pro tipMusk deleted radar but quietly allowed continued development of an advanced system. When it was good enough, he approved it for high-end models.
When the radar supplier could not deliver and an alternative system was failing, Musk faced three options: shut down production, make the alternative work, or delete radar entirely. After 40 seconds of silence, he said delete radar. Engineers resisted. He said if they would not remove it, he would find someone who would. But he allowed continued R&D on advanced radar.
Musk applied this most dramatically when deciding whether Tesla's Autopilot needed radar in addition to cameras. Humans drive using only visual information, so from a physics standpoint, cameras should be sufficient. When the radar supplier could not deliver and an alternative system was not working, Musk made the binary decision: delete radar entirely. He enforced it even when engineers resisted, but quietly allowed continued R&D on advanced radar for future use.