Cagan argues hardware and manufacturing companies do MORE product discovery than software companies, not less, because a shipped mistake in hardware can't be patched post-release the way software can — the cost of being wrong is far higher and mostly irreversible. Tesla is cited as designing orders-of-magnitude more custom parts than traditional automakers while still innovating faster, evidence that heavy discovery and rapid innovation aren't in tension.
This complicates any assumption that Discovery (Product Discovery Process) is primarily a software-industry discipline, and sharpens Rapid Multi-Fidelity Prototyping: prototyping fidelity and volume should scale with cost-of-being-wrong, not shrink because the medium is physical.
Apply: don't treat 'this is hardware, we can't iterate as fast' as an excuse for less discovery — the irreversibility argument points the other way, toward more upfront validation.
Из тем: The Product Operating Model