the second hundred · metaphor 189
How can one rich year of a person seem to contain all of them — so that once you truly know the fragment, the rest is not invented but forced, with nothing left free to make up?
We meet a stranger and, after one intense season together, feel we know the shape of their whole life — before and after, the parts we never saw. Sometimes that feeling is a projection. But sometimes it is uncanny and correct: the fragment really did determine the whole, and any other ending would have had to break something to be different. This page is about the kind of object where that is literally, provably true.
In mathematics there is a class of functions — the analytic ones — so rigid that their values on a tiny patch fix their values everywhere. Not approximately; exactly. Two of them that agree on the smallest sliver must agree across the entire line. Below, you can hand the instrument only a small fragment of such a function and watch it reconstruct the whole hidden curve — and watch a naive reading of the same fragment fail, so you can see what the rigidity is really doing.
The rigidity
Most functions are loose: knowing one on a small interval tells you almost nothing about the rest. You can bend the continuation any way you like. Analytic functions are the opposite — the ones that equal their own Taylor series, built from all the derivatives at a single point. That infinite list of derivatives at one point is called a germ, and the astonishing fact is that the germ determines the entire function, out to the farthest reaches of its domain.
This is the identity theorem: if two analytic functions agree on any patch — even a tiny arc — they are the same function everywhere. There is no room to make one veer off later. The local does not merely hint at the global; it fixes it, uniquely, with zero remaining freedom. The naive power series only sees a disc of this — a finite radius where it converges — but the function itself extends far beyond that disc, and the germ already contains those distant values, folded up. Drawing them out is analytic continuation.
What to try
Drag the fragment down to 2 derivatives: the reconstruction is rough where the function is complicated. Raise it and watch the continuation snap onto the hidden curve until, for the rational functions, the continuation error reads a flat 0 — the fragment has forced the whole thing exactly. Meanwhile the naive reach never grows: no matter how many derivatives you feed the plain power series, it still blows up past its fixed radius. More knowledge doesn't extend the naive reading; it only sharpens the right one.
Press Try a different ending. The rival matches the true function perfectly on the patch, then bends away outside — a perfectly nice, smooth curve. The catch, printed in mono: it is not analytic. That is the whole point. You can invent a different ending — but only by giving up the rigidity, by letting the function have a seam. Among analytic functions there is exactly one that fits the fragment, and it is the truth. Swap functions with the presets and starve or feed each in turn.
The mapping
Analyticity is a metaphor for a life of one piece — where character runs all the way through, where the person is the same person under pressure, at leisure, young and old. For such a life, a rich fragment genuinely forecloses the alternatives: once you have seen enough of the pattern, the missing years are not open questions but consequences. The biographer feels this as a kind of gravity — the known season pulling the unknown ones into a single determined shape, leaving nothing to invent.
And it explains the specific eeriness of really knowing someone: the loss of their freedom to surprise you, in your eyes. To continue a person analytically is to grant them no seams — to insist the fragment you hold already contains all of them. It is intimacy and it is a trap, because the moment they are non-analytic — the moment there is a hidden room the fragment could never reach — your beautiful continuation is confidently, precisely wrong.
Read as life lessons
A thing that is analytic cannot contradict itself later. Its wholeness is exactly what removes the freedom to make things up. Integrity, in the old sense, is being continuable from any fragment.
A plain extrapolation trusts the fragment only within a radius, then breaks. Understanding a whole life needs more than local slope — it needs the deeper form the fragment implies but doesn't display.
You can always draw an alternative continuation — but only by introducing a discontinuity in the deep structure. Where the rules hold, there was never a choice; the surprise means the rules didn't.
In the wild
The Riemann zeta function, and countless amplitudes, are known first on a small region and then continued to give the only consistent values elsewhere — the physics has no other legal answer.
A band-limited signal is analytic: a finite window in principle determines it everywhere. It's why a fragment of a very smooth process can be extrapolated with startling confidence — and why sharp edges defeat it.
The instinct that one decisive chapter “explains” a life is analytic continuation as a habit of mind — sometimes profound, sometimes a projection onto a person who had a hidden room.
The honest model
The fragment is the first L Taylor coefficients of the chosen function at the origin — its derivatives at one point, and nothing else. The naive curve is the partial sum Σ cₖxᵏ, which provably converges only inside a disc of radius R set by the nearest singularity (here R = 1), and diverges outside no matter how many terms you take. The continuation is the Padé approximant built from those same coefficients — a ratio of two short polynomials matched to the germ, solved live by a small linear system.
Padé is a genuine analytic-continuation tool: for the rational curves it reconstructs the hidden function exactly from a handful of derivatives (the readout shows error 0), and for log(1+x) its error shrinks as you feed it more of the fragment, reaching far past the naive radius. The "different ending" rival is honest too: it equals the true function on the whole patch, then adds a smooth-but-non-analytic bump — a legal continuation for a general function, and an illegal one for an analytic one, which is the entire lesson made visible.
The mapping, exactly
| Mathematics | Life |
|---|---|
| an analytic function | A life or character that is genuinely all of one piece, consistent under every condition. |
| the germ at a point | One rich, fully-known fragment — a decisive year, a formative season. |
| the identity theorem | Agree on a sliver and you agree everywhere: no room to have secretly been someone else. |
| radius of convergence | The horizon of a naive reading — how far a surface impression can be trusted before it breaks. |
| analytic continuation | Drawing the whole out of the part, when the part truly determines it — with nothing left to invent. |
| a non-analytic bump | A hidden room, a genuine discontinuity — the one thing that lets the rest be different, and that the fragment could never see. |
Where the metaphor tears
The rigidity is a property of a special, smooth class of functions. People, histories, and institutions are riddled with seams — traumas, conversions, accidents, choices — that a fragment cannot foresee. Treating a person as analytic is exactly the error of the confident continuation: it produces a whole that is coherent, beautiful, and false the instant a hidden room appears. The theorem's certainty is not transferable; only its warning is.
The instrument reconstructs so cleanly because we know in advance the fragment came from a well-behaved function, and we picked the continuation tool to suit it. Real reconstruction faces noise, singularities near the patch, and no guarantee the source was analytic at all — where Padé and its cousins can lock confidently onto the wrong answer. The demo shows what rigidity would buy, not that your fragment has it.