The Fingerpost Is Not the Road
The fingerpost points; it does not walk
Some of my books come by way of my local bookshop, Camilla’s in Eastbourne. But the one I’ve just finished reading came a quite different route. I found it, at the end of a long walk through the Christmas Markets, in The Munich Readery.
And now I’ve just finished reading it. It is called An Instance of the Fingerpost by Ian Pears and, rather than simply racing to the next book, I’ve found myself staring at the cover for a while longer than usual.
There’s something about closing a novel that has so meticulously rearranged your sense of certainty that makes you reluctant to move on too quickly. An Instance of the Fingerpost isn’t just a historical mystery; it’s an anatomy lesson in perspective. Four narrators. Four coherent, persuasive, intelligent accounts of the same set of events. Each one rational. Each one confident. Each one incomplete and only surfaced with growing hindsight.
And that, I suspect, is the point.
The novel is ostensibly about murder, politics, religion, science, and restoration England. But beneath the wigs and intrigue it is really about observation and evidence, about what we think we see, what we choose to notice, and what our existing beliefs allow us to accept.
Each narrator is, in their own mind, rigorous. They gather facts., they apply logic, and they construct arguments. They suck you into their world view, comfortably or uncomfortably depending on your experience of each narrative in turn. Each is constrained by their worldview by ambition, loyalty, prejudice, theology, natural philosophy. They are not liars, exactly. They are interpreters. And interpretation, the book reminds us, is rarely neutral.
As someone who spends a good deal of time thinking about systems — technical and human — I found this quietly unsettling. We often talk about “the evidence” in software engineering. We speak of metrics, observability, telemetry, dashboards. We invoke data as if it were immune to framing.
But data does not speak.
People do.
And people bring priors.
One of the most striking undercurrents in the novel is the early tension between emerging scientific empiricism and older structures of belief. The intellectual atmosphere of the time, with figures like Robert Boyle hovering in the background, is charged with a new confidence in observation. And yet observation itself is shown to be fragile. What one man calls experiment, another calls heresy. What one calls proof, another calls manipulation.
Which makes this line from the novel land with particular force:
“It is often the case that men succeed in persuading themselves through the exercise of reason that what they know to be true is not so, merely because they cannot understand it.”
That sentence could have been written yesterday.
We like to imagine reason as a clarifying light. But the author here suggests something more troubling: that reason can become a defensive mechanism. A way of protecting ourselves from truths that unsettle us. A way of rationalising away anomalies because they do not fit our model.
In software, in organisations, in life — we see this constantly. Signals emerge that something is wrong: friction in a workflow, dissatisfaction in a team, strange production behaviour. Yet rather than revising our mental model, we often construct ever more elaborate explanations for why the system is fine and the anomaly is mistaken.
We reason ourselves out of discomfort.
An Instance of the Fingerpost is, among other things, a warning about epistemic humility. About the limits of perspective. About the danger of mistaking coherence for truth. Each narrator believes he is moving closer to certainty. The reader realises that certainty is receding.
For me, having just closed the book, it felt less like a solved mystery and more like a mirror held up to how we construct reality. Evidence is never raw. Observation is never neutral. Perspective is never absent.
And reason — powerful, necessary, beautiful reason — can sometimes be the very thing that blinds us.
Which, perhaps, is why the fingerpost remains such an apt symbol.
It points.
But it does not walk.
And so this enchiridion entry was born…
Mistaking the Fingerpost for the Road
There is something deeply satisfying about a signpost.
It stands there, wooden arm extended with calm authority, announcing: This way to Oxford. That way to London. No anxiety. No hesitation. No mud on its boots. It does not tremble in uncertainty. It does not second-guess itself. It does not confess that it has never been to Oxford…
It merely points. And we, coherence-hungry creatures that we are, adore it.
Just as we adore the slide with the framework diagram. Adore the tidy architecture principles laminated into corporate folklore. Adore the crisp README that suggests mastery rather than confusion. Adore the AI-generated summary that feels complete before anything has actually been tested. Adore the well-crafted but nonsense statement on LinkedIn that, almost invariably, begins “This Changes EVERYTHING!”
Spoiler alert: It probably doesn’t.
The things is that the signpost soothes us. It reduces ambiguity to direction. It implies that someone, somewhere, has already done the walking. But here is the small, dangerous truth that Francis Bacon understood in the early seventeenth century: a fingerpost at a crossroads points the way — it does not travel the road.
And the road is where reality lives.
The signpost never meets resistance. It never discovers that the bridge is washed out. It never finds that the path forks again into five smaller, less photogenic paths. It never experiences doubt.
In software engineering, particularly in this age of AI-augmented coherence, we are surrounded by fingerposts. Metaphors. Principles. “Best practices.” Governance models. Even our beloved habitat thinking.
Each one points. None of them actually walks.
The problem is not that we have signposts. The problem is that we have begun to treat them as destinations.
We have mistaken coherence for contact, articulation for understanding and explanation for experimentation.
And now, as AI agents produce persuasive, syntactically correct, architecturally plausible artefacts at breathtaking speed, the temptation to stop at the fingerpost grows stronger. The document is generated. The summary is convincing. The design looks symmetrical. The code LGTM.
Surely that is enough?
It is not enough.
Because the road is contextual, and the mud is still out there.
Bacon’s philosophy was a rebellion against scholasticism. Against the belief that verbal elegance or inherited authority constituted knowledge.
His warning translates perfectly into software:
A principle is not a practice.
A diagram is not a deployment.
A metaphor is not a mental model.
A README is not a running system.
The fingerpost is essential. Without it, we wander blindly. But if we stop at it, we become decorators of crossroads alone. In habitat thinking, this matters profoundly.
A cultural principle that says, “We value experimentation,” is a fingerpost. The habitat only plays its part when experimentation is safe, funded, expected, and recoverable.
Symmathesy, the co-learning within a system, does not happen because we described it eloquently. It happens because people interact with the terrain, encounter friction, adjust, and adapt together.
The teaching is in the walking.
Some practices to consider
Treat principles as hypotheses — Every maxim is provisional until tested in your context.
Instrument the road, not the sign — Measure outcomes, feedback loops, recovery time; not just compliance with stated guidelines.
Prefer lived friction over theoretical elegance — If a workflow looks beautiful but feels brittle in practice, trust the friction.
Encourage experimental traversal — Short loops. Safe-to-fail trials.
Separate articulation from validation — A well-written proposal is not evidence.
Some things to avoid
Worshipping frameworks.
Mistaking AI coherence for correctness.
Equating policy existence with behavioural change.
Confusing documentation volume with system understanding.
Treating metaphors as operating models.
The most dangerous moment is when everyone nods.
Some Signals You Are Investing at the Fingerpost
Conversations revolve around definitions rather than experiments.
Disagreements are settled by citation rather than measurement.
AI outputs are approved because they “look right.”
Retrospectives discuss principles without examining outcomes.
Governance enforces form but ignores consequence.
A little example
A team adopts “Platform as Habitat” as a guiding metaphor. They create beautiful diagrams showing flow, learning loops, and developer affordances. They write an elegant internal manifesto.
Yet developers still bypass the platform.
Why?
Because onboarding takes two days. Because local testing is slow. Because feedback loops are noise and un-actionable.
The fingerpost was inspirational. The road was muddy.
Until someone walks the road, i.e. measures friction, removes obstacles, iterates on affordances, nothing changes.
Walk the Walk, Explore the Lived Experience
Don’t invest just at the fingerpost. Ask yourself:
Where are we admiring the sign?
Where have we confused explanation with experience?
Where has coherence replaced contact?
The antidote is simple, though not easy: Walk.
Walk into production telemetry. Walk into developer complaints. Walk into real usage patterns. Walk into failure.
The fingerpost is necessary. Without it, we are lost. But it does not walk for us.
And in software, the mud is where the truth waits.
Further Reading
If the fingerpost is not the road, then these works are invitations to walk.
As you read them, ask:
Where is the fingerpost in this work?
Where does the author insist that you walk?
Where have you reasoned yourself away from something you knew, quietly, to be true?
Because the problem is rarely a lack of signposts. It is in our reluctance to leave them behind.
Enjoy!
On Francis Bacon and the Birth of Empiricism
Francis Bacon — Novum Organum
The foundational argument for disciplined empirical inquiry. Bacon’s critique of “idols” (systematic distortions of thinking) is particularly relevant when considering how reason can mislead.
On Perspective, Evidence, and Competing Truths
An Instance of the Fingerpost — Iain Pears
A masterclass in epistemic humility. Four narrators. Four rational worlds. A powerful reminder that coherence is not the same as truth.
Thinking, Fast and Slow — Daniel Kahneman
A modern exploration of cognitive bias, motivated reasoning, and how easily we persuade ourselves of convenient falsehoods.
The Structure of Scientific Revolutions — Thomas S. Kuhn
How paradigms shape what scientists see and do not see. A reminder that observation is theory-laden and that progress is not linear or genius-driven.
On Observation, Measurement, and Systems
The Signal and the Noise — Nate Silver
A practical meditation on separating evidence from narrative in complex systems.
Thinking in Systems — Donella H. Meadows
Why the behaviour of systems often contradicts our intuitive reasoning — and how feedback loops reveal what signposts conceal.
Observability Engineering — Charity Majors
A modern engineering articulation of the same core insight: you cannot reason your way to system truth without instrumenting and interrogating reality.
On Mental Models and Error
The Black Swan — Nassim Nicholas Taleb
How fragile our explanatory models are when confronted with rare events — and how retrospectively we rationalise what we never predicted.
How to Measure Anything — Douglas W. Hubbard
A disciplined challenge to the idea that “we can’t measure that,” urging us off the fingerpost and onto the terrain.