Is Value Even the Right North Star?
The closing instalment, questioning the premise of the site that publishes it.
The Question
Everything on this hub points at value: proved through attribution, defended in money. But a north star is a navigation choice. Other stars are available: safety, rights, fairness, capability, sovereignty.
Is value the right thing to steer by, or just the most fundable?
Optimist
The Optimist's Case
Value earns its place by what it cuts through. Enterprise AI is squeezed between hype and fear. Value is the only frame that disciplines both at once.
To hype: show me the outcome. To fear: show me the exposure, costed. Both conversations become adult when a defensible number enters the room.
Interpretation Value is the frame that keeps everything else funded. Safety teams, governance, responsible AI: they do not survive on virtue. They survive on budget. Budget flows to what demonstrates worth.
The iceberg argument: value is the visible tip that pays for the submerged base. A pure safety north star becomes a compliance function defunded in the first hard quarter. A value north star, honestly measured with risk netted off, is a programme that defends guardrails to the CFO. The guardrails are inside the number.
Value is the only star that travels. It is one language for finance, engineering, risk, and the board.
Sceptic
The Sceptic's Case
First move: name what kind of thing value is. It is a lagging, gameable, narrow proxy.
Lagging: it arrives quarters after the decisions. You are steering by the wake. Gameable: every value metric has been gamed somewhere by someone. Narrow: money only counts what crosses the ledger.
The deeper objection is moral. Some things matter independent of whether they pay. A discriminatory system is wrong even if it is profitable. A safety duty does not lapse because attribution coverage needed the spend elsewhere.
Interpretation Make value the north star and you implicitly make everything else negotiable against it. The business case for ethics is the only ethics that survives.
Evidence The pattern is visible: sustainability programmes framed as cost savings get cut when energy prices fall. What is justified by value gets unjustified by value.
Forward bet: as AI grows more capable, the gap between valuable and wise widens. An enterprise steering purely by measured value sails competently into the gap.
Synthesis
House viewWhat Would Settle It
This settles less than the other debates. It is partly a values question that evidence cannot close.
The empirical edge is watchable: do value-led AI governance organisations underinvest in the base compared to peers? Do they show worse incident records, bias outcomes, or compliance failures?
InterpretationIf value-led firms systematically starve foundations, the sceptic's everything-negotiable claim is confirmed. If the best-funded, most durable safety functions live inside value-led programmes, the optimist's funding argument wins.
Watch where the strong bases are. Watch what star those organisations steer by.
If We Get It Right / If We Get It Wrong
Value as star, rightly bounded: Programmes prove their worth, fund their own safety, and speak one language across the enterprise.
Value as star, unbounded: Slow conversion of every duty to a business case. What cannot be justified in money gets defunded.
Another star, wisely chosen: Perhaps a safer estate. Struggles for budget, scale, and influence.
No star at all: The current default. Hype and fear take turns at the wheel.
The Author's Honest Position
The resolution is layered, not balanced. Value is the right operating north star, sitting on top of a non-negotiable base.
Safety, rights, lawful conduct: these are constraints, not competing objectives. They bound the space inside which value is maximised. No attribution number buys them back.
Interpretation The iceberg properly read: the base is not weighed against the tip. The base is what the tip stands on.
Within those bounds, I want value, with all its flaws, as the star. It is the only one that disciplines hype, funds the base, and travels across the enterprise.
The sceptic is right: value is lagging, gameable, and narrow. That is why the discipline nets risk off every claim, demands counterfactuals, and publishes coverage honestly.
The sceptic is right about one more thing: a hub named for valuemaxxing has an interest in this conclusion. That is exactly why this series exists. That is why the strongest case against the house appears on the house site.
If the layering ever fails in practice, if the base starts losing arguments to the tip, the sceptic is right and the framework should be revised. That is a falsifiable commitment, meant as one.