Today’s workbench tool is obliquity, John Kay’s argument that complex goals are often best achieved indirectly. The digest frames the idea through companies that became valuable by pursuing engineering excellence, scientific ambition, or genuine problem-solving rather than profit as a naked target. Profit arrived as a consequence of conditions, not as the only compass.
The idea is uncomfortable because direct pursuit feels disciplined. A team wants revenue, so it sets a revenue target. A product wants retention, so it optimizes retention. A professional wants a title, so he maps every move toward the title. Some goals are simple enough for that. Complex systems are not. Push one visible metric too hard and the surrounding conditions can degrade.
In product work, the pattern is familiar. A company that optimizes only engagement may learn to irritate, addict, or polarize users. The chart improves while trust decays. A support organization that optimizes only handle time may become faster and less helpful. An engineering team that optimizes only tickets closed may ship more fragments and own less of the system.
Obliquity does not mean wandering without goals. It means respecting causal distance. If the true goal is a trusted product, the near work may be reliability, taste, customer empathy, and careful defaults. If the true goal is long-term profit, the near work may be technical excellence, brand credibility, and a culture that tells the truth early. If the true goal is promotion, the near work may be becoming useful in ways the organization cannot easily replace.
The AI capital cycle makes this especially relevant. Many teams are now optimizing directly for “AI strategy” because boards, investors, and competitors demand one. That can produce theater: demos without workflow integration, agents without permissions design, metrics without customer value. The oblique path may be less glamorous: find the painful internal process, understand the exception cases, improve the data, narrow the tool surface, and make one workflow materially better.
The same applies to personal productivity. A person who optimizes every hour for output can destroy the conditions that make good output possible: sleep, reflection, curiosity, relationships, and the slack needed for synthesis. The direct pursuit of productivity can make the worker less productive in the only sense that matters.
A useful test is to ask whether the metric is a compass or a substitute for thought. Revenue can be a compass. Customer trust is harder to measure but easier to destroy. Velocity can be a compass. Architecture health is harder to count but painful to lose. Follower growth can be a compass. Reputation is slower, quieter, and more valuable.
Managers can put obliquity to work by naming both the target and the conditions. Do not merely say “increase retention.” Say “increase retention by making the product more reliably useful for the right customers.” Do not merely say “ship faster.” Say “ship faster by reducing rework, clarifying ownership, and protecting quality.” The preposition matters. It prevents the shortcut from eating the goal.
The frontier ledger’s rule is simple enough: when the trail is complex, a straight line may be the slowest route.