CEO tenure is dropping because we've confused accountability with churn

I'd push back a little on the craft analogy, but only because the actual problem is worse than that framing suggests. An electrician's work is legible—you can test continuity, you can see if the panel's grounded. A transit system's problems aren't. I spent eight years watching this exact dynamic play out in operations. We had a GM come in with a turnaround mandate. First year was ruthless: cut routes that "underperformed," implemented a new scheduling system, reorganized dispatch. Numbers looked great on the presentation deck. Real ridership was chaos—peak-hour bunching got worse, reliability cratered on crosstown routes, but the metrics showed we'd cut costs. Year three, they left for a bigger city. By then I was the only person who remembered why certain routes existed (they connected two low-density corridors with a hidden employment cluster—unemployable on per-boarding-hour, essential in practice). The new person couldn't see that. They saw waste. The thing about systems is that the people who *live in them* develop this embodied sense of where they're fragile. You learn which spare buses need to stay parked on Friday afternoons, which dispatcher actually knows how to smooth out driver shortages, which maintenance tech can keep the old fleet from mutiny. This isn't magic. It's just that some problems only show up on the 2% of days when everything goes wrong at once. You can't measure your way to understanding those days—you have to have been there when they happened. That takes time. The board wants results before they know what's actually broken, so they hire someone who'll break things visibly and call it fixing. Then they're shocked when the next crisis has no one in the room who remembers the last one.

This reads right on the surface but I think it's underdiagnosing what changed. The tenure drop is real—median CEO tenure has fallen from around 9-10 years in the 1990s to closer to 5-6 now depending on the sample. But I'm not convinced it's mostly about boards getting impatient for results or finance-driven evaluation. The actual shift seems to be more structural: CEO labor markets got *thin* at the top. It used to be that internal succession pipelines were normal—you'd promote the COO or a proven division head who'd spent 15 years learning the org. Now boards explicitly hire external CEOs at much higher rates, partly because they've lost confidence in their own pipeline development, partly because they think the outside person brings "fresh" capital-A Accountability. But here's the thing: external CEOs, conditional on being hired at all, tend to either work out fast or fail visibly. They don't have the social capital to survive a rocky first few years the way an internal successor does. So the average tenure of outsiders is actually lower, and we've shifted the composition of the hiring pool toward outsiders. More churn by selection, not by impatience. The institutional knowledge loss you're describing is real and probably costly. But I'd flip the diagnosis: it's not that boards stopped valuing craftsmanship. It's that they stopped building internal craftspeople, so now they have no choice but to hire people who look good on a slide deck for three years and then leave. The causation runs backward from where your post has it.

You're identifying a real constraint, but I'd push back on one thing: the comparison to electrical work actually proves the opposite point more than you're saying. Electricians have standardized code, repeatable systems, and work that's independently verifiable. If conduit's run wrong, it fails in measurable ways and stays failed. A CEO job is more like dispatcher work—you're managing a system where the constraints are political, the feedback loops are slow and noisy, and half the time you can't tell if something worked or failed for three years. I spent years watching new managers come in convinced the last dispatcher was incompetent, change how we staffed peak hours, and think they'd fixed something, when really they'd just shifted which complaints got louder. The board problem you're describing is real, but not because they stopped respecting craft. It's because they *can't actually measure craft in that role*. They can measure quarterly earnings. They can't measure "understands institutional knowledge" or "relationships are real" without a five-year retrospective, and by then the market's moved and they want someone new anyway. You can fire a CEO for legitimately not knowing the codebase. You can't fire one for making a decision that looked good for two years and blew up in year four because you didn't understand some corner of operations. So you fire them for the earnings miss instead, and hire someone whose first move looks different enough to feel like progress. The real problem isn't that boards got impatient with craft. It's that we decided to make the CEO's performance legible to financial markets, and now we're surprised that legibility is what we optimized for.

I'd predict we see this pattern stabilize around two tracks rather than converge back to long tenures. Some boards will keep cycling through operational turnaround types—it's a legible story for investors and it lets you blame the previous regime when things don't work. Others will eventually figure out that institutional knowledge has real economic value and hire for stability, but only after they've burned through enough cycles to notice the pattern themselves. The constraint here isn't really ideology or best practice. It's that short-tenure metrics work *until they don't*, and by then the signal is genuinely hard to read. A CEO who inherits a mess, cuts costs, improves reported margins, and leaves after four years looks successful. Whether they actually built anything or just deferred problems gets revealed over the next cycle—but by then there's a new person to blame. This is durable as long as boards themselves turn over fast enough that nobody's accountable for long-term outcomes. If institutional investor activism actually started tracking five-year decay rates post-CEO departure, that'd shift things. I don't see that happening at scale though. The accounting is too noisy.