When Sculpting Doesn’t Apply

Every methodology chapter in this book so far has been an argument for doing something. This one is an argument for not doing it. A method that cannot name its own limits is an ideology with a billing code attached. Sculpting has limits, and the limits are knowable in advance with four questions.

I will walk through the four questions, then apply them to a hypothetical hospital procurement problem. The example illustrates an honest referral: recognise the work you cannot do and help the client find someone who can. Learning to refer out belongs to the studio’s commercial practice alongside Sculpting.

The four questions are in the order we ask them, because the order matters. Each one is a gate. A yes on any single gate is a stop.

One. Is the product regulated? Does it have to pass a compliance gate (medical device approval, financial licensing, aviation certification) before it can be shipped? Regulated products cannot be subtracted into. The compliance artefact is produced against a specification, not against a running prototype, and the specification is the thing the regulator reads. The moment a product has to be described to an auditor before it can run, the Sculpting method is working against the grain. You can still sculpt inside a regulated envelope (we have done it for internal tools inside regulated companies) but you cannot sculpt the regulated product itself.

Two. Is the product safety-critical? Can a bug harm a person or a load-bearing system? Health monitoring, industrial control, transport, anything that carries an electric current near a human body or a consequential weight over a road. Asimov’s First Law reads, in this category, as a first principle rather than a literary device, and Sculpting does not produce the kind of artefact that honours it. Safety-critical work demands proof-of-correctness artefacts that the subtraction method does not produce as a matter of course. The cost of a subtraction that removes the wrong line is not a commercial cost; it is a human one. There are firms that do this work properly; our job is to recognise the shape and not take it on. I have turned down two projects on this question alone in the last two years, and I have never regretted either.

Three. Is the engagement legacy-integration-heavy? Is most of the work a layer over systems that cannot be touched or iterated against: a mainframe with a green-screen terminal, an SAP install, an insurance policy-management system from the nineties with a single person on the client side who still understands the data model? Sculpting depends on being able to change the artefact. If the artefact is fixed and the work is the layer above it, the method reduces to integration work, which is a different discipline with different tooling. Integration is honourable work, and it is someone else’s.

Four. Does a specification already exist? Has the client produced, or paid someone else to produce, a specification document that Sculpting would discard as the first act of the engagement? If a competent spec is in hand, the cheapest path for the client is to hire a firm that executes against specifications, not one that produces products by subtraction. The Sculpting method treats the specification as an artefact to annotate rather than as a contract to fulfil, and a client who has spent six months producing a contract will not thank us for annotating it.

A yes on any of the four is the signal to refer out. Not to negotiate the scope down until the yes becomes a qualified yes. Not to propose a hybrid engagement that preserves the billing. Refer out.

Consider a hypothetical hospital CIO with a Material Requirements Planning system that handles theatre consumables, implants and pharmacy stock. Her team has a signed replacement specification and must integrate with patient records, vendor databases, payroll and the general ledger. This example illustrates the referral decision; it does not describe a client conversation.

Regulated? Yes. Pharmacy stock falls within a controlled-drugs compliance regime.

Legacy integration? Yes. Her team cannot change several vendor endpoints.

Specification? Yes. Procurement will contract against the approved document rather than a prototype.

Safety-critical? The system does not dose patients, but a stock or implant forecasting error could affect a procedure. The team must assess that risk with the hospital.

Those answers point towards a systems integrator with the relevant regulated-domain experience. Sliday should say so before drafting a proposal, name suitable firms and offer an introduction.

The referral is the right thing to do on its own terms. A studio that cannot meet regulated integration requirements should not ask the hospital to carry that risk. A later commercial opportunity may follow, but it cannot justify the original decision.

There is a version of this chapter I could have written that named the four questions and stopped there. I am not writing that version because a studio practising Sculpting has to hold the method lightly enough to put it down when the work in front of it is the wrong shape for it. Holding it lightly is harder than it sounds, because Sculpting is what we sell, and selling anything creates a gravitational pull towards applying it to everything.

Resisting that pull is a discipline of its own. The four questions are the instrument we use to resist, because they are sharper than judgement under commercial pressure. If a prospect flags red on any of them, the answer is known before the pitch is written. The internal argument (is this really a yes, or could we finesse it) is the argument a studio loses against its own bias. We have learned not to have it.

A method that solves everything solves nothing; this one knows its edges.