Xerox PARC built the graphical user interface, the mouse, and ethernet — and then watched Apple and Microsoft commercialize all of it. The engineers weren't incompetent; they were trapped by what organizational theorist James March called the exploitation-exploration tension: organizations that get good at executing a known thing systematically defund the search for the next thing. Most product teams live entirely on the exploitation side. The roadmap is a queue of improvements to what already works, and velocity metrics reward shipping, not questioning whether you're building in the right direction at all. March's research showed that the bias toward exploitation isn't laziness — it's rational in the short term and lethal in the long term, because exploration has delayed, uncertain payoffs while exploitation has immediate, legible ones. The practical implication is structural, not motivational: if 'explore the adjacent unknown' competes for resources against 'ship the next feature,' it loses every sprint. The only teams that escape this are the ones that ring-fence exploration as a non-negotiable budget line — time, headcount, or both — before the roadmap is built, not after.

What percentage of last quarter's actual working time — not stated values, but real time — went to genuinely exploratory work versus improving what already exists?

Drawing from Organizational Learning Theory / Marchian Exploration-Exploitation Framework — James G. March (Exploration and Exploitation in Organizational Learning, 1991)