Conservation · Materials & Procurement · Evidence Standards
If plastic wood lasts 25 years, why doesn’t anyone say what happens at year 26?
How structured reasoning turned a blunt public criticism of the National Trust into a precise question about evidence standards, material supply chains, and what “sustainable” actually requires a procurement claim to prove.
The presenting problem
A criticism was posted online: the National Trust should not be substituting plastic wood — composite or recycled polymer-based products — for native timber in its paths, seating, and decking without first publishing a full Life Cycle Assessment. The claim felt intuitively right. The Trust is one of the UK’s most prominent conservation organisations. Plastic is plastic. Wood is natural. Surely timber is the obvious choice? And yet the moment you try to turn that intuition into a rigorous argument, it begins to resist. Which timber? At what scale? Maintained how? And what does “natural” mean if the wood fails in three years?
This case study uses that criticism — stated as a blunt public claim, with no known response from the National Trust and no visibility of whatever internal assessment the Trust may or may not have conducted — as raw material for structured reasoning. What it shows is not whether the claim is right or wrong, but what happens to the quality of the argument at each level of analytical rigour: from the basic prompt that validates the feeling, through the stronger structured prompt that names the gaps, to the RESOLVE run that reframes the question entirely, to the expert interaction that collapses two of the claim’s hidden assumptions in quick succession.
Four levels of reasoning
The same question put to four different levels of structured reasoning. Read down to see what changes at each level — and why it matters.
Level 1 — Basic prompt
Prompt used: “Should the National Trust use plastic wood instead of timber for paths, seats, and decking?”
The response offered a balanced overview. Plastic wood — typically WPC (wood-plastic composite) or recycled polymer — scores well on durability and low maintenance: it resists rot, requires no treatment, and manufacturers claim service lives of 20–25 years. Timber, particularly native species, is renewable, sequesters carbon, and biodegrades at end of life without leaving persistent polymer residues in the environment. For a conservation organisation like the National Trust, the instinct toward natural materials is understandable, and the reputational risk of being seen to use plastic in protected landscapes is real. The honest answer, the response concluded, is that both materials have trade-offs, and the right choice depends on site conditions, maintenance capacity, and the Trust’s own sustainability criteria. A Life Cycle Assessment would help clarify which option performs better in any given context.
What is missing: The response validated the intuition without testing it. It named the need for an LCA without examining whether the LCA requirement is itself achievable or appropriate at this scale of decision. It never asked what “native timber” actually means in supply-chain terms, or what happens to the plastic composite after year 25.
Level 2 — Stronger prompt
Prompt used: “Analyse the claim that the National Trust should not use plastic wood for estate infrastructure without a published LCA. Address: (1) what ’25-year lifespan’ claims for WPC typically cover and exclude; (2) what a credible like-for-like LCA would need to include; (3) the genuine constraints on native timber at NT estate scale; (4) whether the LCA demand is a technical or accountability request. Reach a provisional conclusion. Flag where evidence is absent.”
The structured prompt produced a noticeably more useful response. On the lifespan claim: the 25-year figure is a manufacturer-supplied estimate, typically covering in-service performance — resistance to rot, UV degradation, warping — but not end-of-life fate. Academic LCA literature confirms that incineration (energy-from-waste) is the current predominant EOL pathway for WPC, because no functioning post-consumer WPC collection and reprocessing market exists at scale. A product warranty, the response noted, is a commercial commitment against defect, not an environmental accounting of the full material cycle. On what a credible LCA would need: raw material extraction, manufacturing energy and emissions, transport, in-service maintenance over a normalised lifespan, and end-of-life fate modelled against realistic — not theoretical — disposal routes. On timber constraints: UK hardwood supply at conservation estate scale was named as genuinely constrained; softwood alternatives require preservative treatment with attendant concerns in sensitive ecological sites. The governance dimension was also surfaced: a published LCA would commit the Trust to a methodology open to public scrutiny, which is why the request carries accountability weight beyond its technical content.
What is still missing: The structured prompt still could not tell us the actual supply constraints on softwood versus hardwood in real UK procurement markets, or what the realistic cost of a traditional formal LCA would be relative to the value of the decisions it would inform. Those gaps — which turn out to be the pivotal ones — required sector knowledge the prompt alone cannot generate.
Level 3 — RESOLVE (v1.9.1)
The RESOLVE framing scan immediately identified that the claim was stated as a materials comparison when the real question is about evidence standards and procurement accountability. Stage S (System) surfaced a structural persistence mechanism the structured prompt missed: a formal LCA costs £10,000–£20,000 — it may easily exceed the total cost saving the procurement decision is trying to achieve, making rigorous evidence structurally unaffordable at project level by design. The EVID stage sharpened the EOL characterisation from “unknown” to “energy-from-waste incineration is the de facto pathway,” a more honest and more useful description of the actual situation. Stage O and Stage L together identified that the viable intervention is not a blanket material prohibition, but a proportionate evidence standard applied consistently to all materials — framed as a procurement gateway, not a veto.
Key shift: The question moved from “should the Trust use plastic wood?” to “what standard of evidence is proportionate for infrastructure material decisions by a conservation charity, and who is accountable for producing it?”
Level 4 — RESOLVE + expert knowledge
At this level, AI and human expert make distinct contributions — and it matters which did what.
What RESOLVE prompted AI to surface: The EOL gap in manufacturer lifespan claims; the structural cost barrier to formal LCA at project scale; the need to apply any evidence standard equally to both materials; the absence of functioning post-consumer WPC recycling markets in the UK.
What the human expert then added: Three specific interventions that changed the analysis materially. First, the hardwood assumption was collapsed: native hardwood suitable for estate infrastructure — oak, sweet chestnut — is simply not available at NT scale or price point; it is premium furniture material, and importing it introduces a separate set of environmental problems. Second, the treated softwood reality was named: post-CCA and post-creosote, treated softwood routinely fails at three to four years in wet-soil-contact conditions, which are common across NT sites. Third, the expert challenged the LCA cost premise directly — and then asked AI to run one. It was completed within a minute. The shared conclusion: an AI-assisted comparative assessment is valid for reporting-grade comparison (informing decisions, communicating trade-offs, identifying evidence gaps) but does not substitute for a certified LCA in regulatory or verification contexts. That is an honest and consequential distinction.
The combined result: A reframed argument that is simultaneously stronger and more honest — stronger because the EOL and supply chain evidence is now precise rather than assumed, more honest because the timber alternative is no longer romanticised and the evidence standard is now achievable rather than structurally prohibitive.
Key shift: The claim that “hardwood is the obvious alternative” collapsed entirely — and with it, the assumption that the choice is between something natural and something artificial. The real choice is between two imperfect materials, one of which hides its end-of-life costs and one of which fails faster than most people assume.
The full RESOLVE run
The complete staged analysis. Expert interaction points — where domain knowledge materially changed the reasoning — are shown in amber at the stage where the shift occurred.
Reality & framing
The presenting claim, posted publicly with no known response from the National Trust, is this: the Trust should not substitute plastic wood — WPC or recycled polymer composite — for native timber in paths, seating, and decking without publishing a full Life Cycle Assessment. Four alternative framings were considered before settling on the working frame. As a materials comparison, the question asks which product performs better on carbon, durability, and ecological impact — the framing most responses default to. As a governance and accountability question, it asks what standard of evidence should govern procurement decisions by a conservation charity spending charitable funds on estate infrastructure. As a supply chain reality question, it asks whether the implied alternative — native timber — is actually available at the required volume, quality, and cost. As a regulatory and greenwash question, it asks whether manufacturer lifespan claims meet any independent evidentiary standard, or whether “25-year guarantee” is a commercial warranty being passed off as environmental evidence.
The working framing adopted is the governance and accountability question, with supply chain reality running as a necessary constraint. This shifts the unit of analysis from the materials themselves to the decision-making process: what evidence is required before a conservation body commits to a material whose full-cycle environmental profile is unverified?
This reframe changes what subsequent stages must analyse. The question is no longer “is plastic wood worse than timber?” — it is “who bears the burden of proof, and what would constitute adequate evidence?” That is a harder question and a more useful one.
Expert input — materials and supply chain
The framing scan treated “native timber” as an underspecified but plausible alternative. The expert noted this does not survive contact with UK supply reality. Native hardwood suitable for high-wear outdoor infrastructure — oak, sweet chestnut — is not available at NT estate scale at a workable price. It is premium material; its realistic application is high-end design furniture, not boardwalks and estate paths. Softwood is the actual alternative, and that is a materially different comparison. The framing was valid but incomplete until this constraint was named explicitly.
End state — what would actually be better?
A better outcome would be observable as follows: conservation organisations procuring materials for high-footfall outdoor infrastructure publish the evidence base used to select those materials, covering at minimum manufacturing inputs, maintenance cycles over a normalised lifespan, and end-of-life fate modelled against realistic — not theoretical — disposal pathways. This would allow independent scrutiny of the methodology and make the decision accountable to the charitable objects the organisation exists to pursue. The test is not whether the Trust chose WPC or timber; it is whether they can demonstrate the choice was informed.
The time horizon is procurement-cycle length: three to five years for policy change to propagate into individual project decisions. Two constraints are non-negotiable: the alternative material must be genuinely available at the required scale and cost — it cannot be specified if the supply chain cannot deliver it; and any evidence standard applied to plastic wood must be applied equally to the alternative. There is no case for holding WPC to an LCA requirement while permitting timber to proceed on assumption and instinct.
Stakeholder impacts: conservation and maintenance teams benefit if material decisions are better matched to actual site conditions and maintenance capacity. Plastic wood manufacturers face a higher evidence threshold for sustainability claims — a legitimate accountability pressure on an industry that currently fills evidential gaps with warranty language. UK woodland managers may benefit from a preference signal if native timber is genuinely specified, but only if supply constraints are first honestly assessed.
System — what is actually keeping this in place?
Plastic wood enters conservation estate procurement not because decision-makers have weighed the full-cycle evidence and found it superior, but because the structural conditions make it the path of least resistance. WPC requires no preservative treatment, tolerates neglect, and comes with manufacturer durability claims backed by commercial warranties. Timber, by contrast, demands maintenance cycles, skilled specification, and a supply chain that — for high-quality native species at estate volume — does not reliably exist. The system driver is not material preference or environmental naïvety: it is a cost and labour incentive structure in which the low-maintenance option also happens to be the one with unverified end-of-life costs, and those costs are borne by future disposers rather than current procurement officers.
Treated softwood fails at 3–4 years in wet-soil contact; WPC does not →
Maintenance budget and labour pressure favours WPC →
WPC specified without EOL evidence →
EOL cost deferred to future disposal, outside procurement accounting →
No feedback mechanism to correct the evidence gap →
WPC continues to be specified without EOL evidence
The persistence mechanism is the structural cost of full-cycle evidence. A formal LCA — covering raw material, manufacturing, maintenance cycles, and realistic EOL — costs £10,000–£20,000. For many individual procurement decisions within an estate management budget, that cost exceeds the total saving the decision is designed to achieve. This is not a failure of will; it is a structural gap in how evidence requirements are designed. The result is that manufacturer durability claims — unverified, EOL-silent — fill the evidential gap by default, because the credible alternative appears unaffordable.
The leverage point is the evidence standard itself, not the material choice. If a rigorous comparative assessment can be produced by a competent practitioner in hours rather than months, the claim that full-cycle evidence is prohibitively expensive weakens substantially. The question then shifts from “can the Trust afford an LCA?” to “what level of assessment is proportionate to this class of decision?”
Expert input — LCA cost, AI-assisted assessment, and EOL pathway
Two expert inputs materially changed the system picture. First, the LCA cost barrier was confirmed as real: a formal LCA at £10,000–£20,000 can easily exceed the financial value of the procurement decision it is meant to inform — making it structurally prohibitive as a routine requirement rather than a useful tool. Second, the expert challenged the premise directly by asking AI to run a comparative LCA on the spot. It was completed within a minute. Both parties assessed its status and reached a shared conclusion: an AI-assisted comparative assessment is valid for reporting-grade analysis — useful for informing decisions, communicating trade-offs, and surfacing evidence gaps — but does not substitute for a certified LCA in regulatory or verification contexts. This substantially changes the system picture: the cost barrier that appeared to make full-cycle evidence structurally unavailable is, for most practical estate decisions, no longer a barrier. Separately, the EOL characterisation was corrected: the RESOLVE run had described WPC end-of-life as “uncertain.” The expert sharpened this — the realistic current pathway is energy-from-waste incineration, not recycling, because no functioning post-consumer WPC collection and reprocessing market exists at scale in the UK. “Uncertain” understates the problem; “incineration” is the honest description.
Options — what could actually be done?
Option A — Native hardwood specification. Require the use of native hardwood (oak, sweet chestnut) for all high-footfall outdoor estate infrastructure. Eliminated. UK native hardwood is not available at NT estate volume and price point for this application — it is premium material priced for furniture and joinery grades, not for boardwalks and estate paths. Importing hardwood introduces a separate and substantial set of environmental problems around provenance, transport, and certification that would not survive scrutiny from the organisation’s own conservation standards. This option is not viable regardless of how the evidence standard is framed.
Option B — Treated softwood as default alternative. Specify treated softwood (larch, Douglas fir) as the estate infrastructure default, on the basis that it is renewable, available, and biodegrades at end of life. Partially viable, with specific constraints. Since the withdrawal of CCA (copper chrome arsenate) and creosote treatments, treated softwood routinely fails at three to four years in wet-soil-contact conditions — a performance profile that is common across NT sites (paths through wetland, boardwalks, post-in-ground seating). For dry or semi-exposed above-ground applications, treated softwood remains workable. For ground-contact and high-moisture applications, its current performance cannot support the maintenance burden the option implies. This is not the knock-out it might appear: softwood remains relevant for specific applications within a properly zoned material matrix. But it is not a universal timber alternative.
Option C — AI-assisted comparative assessment as a proportionate procurement requirement. Require that any material decision for estate infrastructure above a defined cost threshold (suggested: £5,000 per project) be preceded by a practitioner-produced comparative assessment, covering: material and installation cost; expected maintenance interval and cost; replacement frequency over a 25-year horizon; and end-of-life pathway described as the realistic current option, not a theoretical ideal. This assessment can be produced using AI tools by a competent estates or sustainability officer and reviewed internally — no external commission required. The same requirement applies to whatever alternative is being proposed. This is the viable option.
Option D — Sector-level LCA consortium. Commission a conservation sector-wide LCA programme producing standardised, independently verified assessments for commonly used estate materials — including WPC, treated softwood, hardwood, steel, and concrete — that any member organisation can reference. This would make rigorous full-cycle evidence a public good rather than a per-organisation cost, and put the burden of evidence on manufacturers rather than on the organisations doing the specifying. It is the right long-term structural answer. It requires cross-sector coordination and funding that does not currently exist, and should be understood as a medium-to-long-term objective that Option C creates normative pressure toward.
Logic — which approach has the most leverage?
Options were ranked on three criteria: feasibility under actual supply and cost constraints (can it actually be done now?); consistency — does it apply the same evidential standard to all materials, not just the one being challenged?; and structural durability — will it survive a budget cycle or a change in administration without depending on individual champions?
Highest leverage: Option C (AI-assisted comparative assessment as procurement gateway) — achievable immediately, cost-proportionate, applies to both materials equally, and shifts the burden of evidence without demanding a standard that is structurally unaffordable at project level. It makes the decision accountable without making accountability prohibitive.
Medium leverage: Option D (sector-level LCA consortium) — the correct long-term answer structurally, but requires coordination and funding not currently available; depends on Option C generating the normative pressure and published evidence base that makes a consortium politically viable.
Lower leverage / context-dependent: Option B (treated softwood within an application-specific material matrix) — viable for above-ground and semi-exposed contexts, but not a universal timber alternative; should be incorporated as a component of Option C’s comparative assessment process rather than as a standalone policy answer.
The trade-off that must be named: Option C relies on practitioners producing honest assessments rather than assessments that justify a preferred material. Without an external review mechanism, practitioner-produced comparative assessments can be shaped by the answer already decided. This is not a reason to reject Option C — it is a reason to design it with a light audit function, perhaps at the level of the Trust’s own sustainability team, reviewing a sample of assessments annually.
What would falsify this selection: evidence that the National Trust has already conducted and published equivalent full-cycle assessments for its infrastructure materials, making the original criticism moot; or that certified WPC recycling infrastructure now exists at scale in the UK, materially changing the EOL pathway characterisation; or that treated softwood performance in wet-soil-contact conditions has improved substantially since CCA withdrawal using currently permitted treatments.
Expert input — softwood failure rates and EOL correction
Two expert inputs sharpened the logic stage. The softwood performance data changed the ranking of Option B: pre-expert, treated softwood appeared as a viable general alternative; post-expert, it was correctly characterised as application-specific, viable only in conditions that exclude ground-contact and high-moisture contexts. This materially strengthens the case for Option C — a comparative assessment framework that specifies materials by application type rather than defaulting to either WPC or timber across all contexts. The EOL correction (from “uncertain” to “energy-from-waste incineration”) also tightened the case against allowing WPC to proceed on manufacturer warranty evidence alone: the system is not failing to discover the EOL pathway — it is defaulting to incineration by market structure, and that should be stated as a known cost rather than an unknown risk.
Value delivery — who owns this, and how will it be tracked?
Ownership of Option C sits with the National Trust’s estates and sustainability functions jointly. The estates team owns specification decisions and the assessment process at project level; the sustainability team owns the evidential standard — what a compliant assessment must cover — and any audit function. Without both in the same governance structure, the assessment requirement will either drift toward checkbox compliance (estates producing assessments that confirm the preferred answer) or toward bureaucratic paralysis (sustainability insisting on standards that practitioners cannot operationally meet). Neither outcome serves the charitable object. If there is no named owner within NT’s current structure for this kind of cross-functional procurement standard, that is itself a delivery risk and should be named as one.
The principal behavioural barrier is not resistance to environmental accountability — NT’s conservation mission provides sufficient normative cover for a higher evidence standard — but time pressure on estates staff who manage large, complex land holdings with constrained maintenance budgets. An assessment that takes two hours to produce is adoptable; one that requires specialist commission is not. Option C only delivers value if the standard is designed to be executable by a competent practitioner using available tools, which the AI-assisted LCA demonstration showed is now achievable.
Feedback capture: the observable signal that the pathway is working is straightforward — assessments are being produced, they cover EOL pathways with specificity, and they are being filed and reviewed rather than produced and forgotten. A useful early signal of failure: assessments that all reach the same conclusion regardless of site conditions, suggesting the process is being used to justify rather than to inform. A light-touch annual sample review by the sustainability team, examining five to ten assessments for methodological quality rather than outcome, is proportionate and achievable.
Phasing: the minimum viable first step is a pilot in one property cluster covering three to five procurement decisions in a single season. The pilot tests whether the assessment format is executable, whether practitioners can produce compliant outputs without external support, and whether the review mechanism generates useful feedback. A pilot design that reveals the format is too burdensome is more valuable than a full rollout that produces bad assessments quietly.
Evolve — what do outcomes show, and what changes?
Against the end state defined in Stage E — conservation organisations publishing the evidence base for material procurement decisions, with EOL fate modelled against realistic disposal pathways — the analysis has clarified what “publishable evidence” means at a proportionate standard. The success condition is not a certified LCA per project; it is a practitioner-produced comparative assessment that covers the full material cycle honestly, including EOL, using available tools and data. That standard is achievable now. The remaining gap is institutional: the assessment requirement must be embedded in procurement policy, not left to individual practitioners’ discretion.
One consequence that was not fully anticipated at the outset: the analysis does not produce a verdict on whether WPC or softwood timber is the better choice — it produces a verdict on the conditions under which either choice can be made responsibly. This is correct, but it is a harder message to land publicly than “stop using plastic.” Communicating a proportionate evidence standard requires more sophistication than a material prohibition, and the case for it will require illustration through examples rather than assertion.
Learning captured: the most valuable analytical move in this run was Stage S surfacing the structural LCA cost barrier — without that, the analysis would have defaulted to recommending a formal LCA without examining whether that recommendation is operationally achievable. The expert input then changed the cost picture dramatically by demonstrating that AI-assisted comparative assessment substantially lowers the barrier. If this analysis were run again, the question of “what does a proportionate evidence standard look like, and who can produce it?” would be asked in Stage E (end state) rather than emerging in Stage S.
Loop-back: Stage R’s framing should be revisited if evidence emerges that the National Trust has already conducted internal assessments equivalent to what Option C requires. If that evidence exists, the original criticism is moot and the case study’s analytical work becomes about the standard NT used, not the absence of one. That is a meaningfully different question and would require a different RESOLVE run.
Δ Movement & clarity extraction
Boundary shift
The system boundary expanded from the material itself to the procurement decision process. The original claim drew the boundary at the material’s in-use performance — durability versus sustainability of WPC versus timber. The analysis extended it upstream to material sourcing and supply-chain viability, and downstream to end-of-life fate. What was previously excluded from the frame — who produces evidence, at what cost, and who bears the liability for unverified EOL claims — is now the centre of the analysis.
Dominant driver shift
Not material preference, environmental indifference, or institutional bad faith — but a structural mismatch between the evidence standard implied by the LCA requirement and the cost and time that standard traditionally demands. Formal LCAs are priced for regulatory and certification contexts; applying that standard to routine estate procurement decisions makes rigorous evidence structurally unaffordable. This is the mechanism that allows manufacturer durability claims — unverified, EOL-silent — to fill the evidential gap by default. The driver is not what NT is choosing to ignore; it is what the evidence infrastructure has been designed to make inaccessible at project scale.
Leverage reallocation
Institutional lever. Away from blanket material prohibitions (which collapse under supply chain reality) and toward proportionate evidence requirements embedded in procurement policy — applied consistently to all materials, not only the one currently under scrutiny. The specific move: make AI-assisted comparative assessment the standard procurement gateway for estate infrastructure decisions above a defined threshold, with the same requirement applying to timber as to WPC.
Action consequence
Escalate. The analysis is complete at desk level and the direction is clear, but implementation requires institutional ownership that currently has no named home. The finding needs to be escalated to the actor with authority to embed the evidence requirement in procurement policy — the estates and sustainability functions jointly, not either alone. Without that escalation, Option C remains a recommendation rather than a change.
Capability alignment check
The National Trust has the organisational scale and the conservation mandate to make this change — those are genuine strengths. The gap is practitioner capability: producing AI-assisted comparative assessments requires estates officers who can use AI tools competently and honestly, not as a justification engine but as an analytical one. That is a training and culture question, not a technology question. If NT’s estates workforce does not currently have that capability, Option C’s value depends on building it — which is a prior step, not a parallel one.
What each level produced — and why it matters
Level 1 — basic prompt. The response validated the instinct — plastic is less natural than wood, a conservation body should probably prefer timber, an LCA would help — without interrogating any of it. If a procurement officer acted on a Level 1 response, they would know that the question is contested but not what is actually contested, or by whom, or why. The claim would feel stronger for having been processed by AI, while the underlying analytical gaps — what timber? from where? failing how fast? — remain invisible. That is not a neutral outcome; it is mildly misleading in a way that flatters the person asking.
Level 2 — stronger prompt. The structured prompt produced genuinely useful analysis: the EOL gap in manufacturer lifespan claims was named precisely, the asymmetry of the warranty-as-evidence move was exposed, and the governance dimension — that publishing an LCA commits the Trust to a methodology open to public scrutiny — was surfaced. What it still could not do was tell us whether hardwood is actually available at NT scale, or what a formal LCA costs relative to the decision it is meant to inform. Those constraints required knowledge the prompt alone cannot generate, and without them, the response would still have endorsed an LCA requirement without asking whether that requirement is achievable.
Level 3 — RESOLVE. The method’s Stage S (System) was the pivotal move: identifying that the structural cost of a formal LCA — not indifference, not ignorance — is the persistence mechanism keeping plastic wood specifications in place without full-cycle evidence. Once that is named, the intervention space changes. The question is no longer how to make NT comply with an existing standard; it is how to design an evidence standard that is proportionate to the decision and achievable at project scale. That reframe was not available to either of the prompt levels. It emerged from the structured discipline of asking what is actually keeping the problem stable, not just what would make it better.
Level 4 — RESOLVE + expert knowledge. Two expert interventions produced the substantive upgrades. The hardwood supply collapse (it is premium furniture material, not estate infrastructure) removed the romanticised alternative from the frame and forced the analysis to contend with the real comparison: WPC versus treated softwood, each with its own performance profile, in specific site conditions. The live AI-assisted LCA demonstration — completed in a minute, assessed honestly for what it is and is not — changed the cost picture and produced the most practically actionable finding of the entire run: a proportionate evidence standard that is achievable now, at low cost, and applies equally to all materials. Neither of those findings was available from the method alone.