Technology Title
Bio-Inspired Materials for Energy Applications
Bio-Inspired Materials for Energy Applications
Project Title
Generational Cost of Ownership: building the model and closing the open questions
Generational Cost of Ownership: building the model and closing the open questions
Category
Finance
Finance
Authors
michael.wright@interceptinghorizons.com
michael.wright@interceptinghorizons.com
Short Description
A working group building and stress-testing a generational cost model and its recognition standard, and closing six named open questions ahead of the November 2026 session.
A working group building and stress-testing a generational cost model and its recognition standard, and closing six named open questions ahead of the November 2026 session.
Long Description
What this group is for. Three instruments exist in draft and none has been ratified by anyone but its authors. This group's job is to break them where they are weak, rebuild the parts that do not survive, and close six specific questions before 9 and 10 November 2026. What exists today. The BiosVerse-Compatibility Index (BCI) is a recognition standard. Seven questions asked of every project in the same order, scored on evidence rather than assertion, building up toward 100. Five are weighted and two are gates, so a strong column cannot paper over a failed one and a project can score 88 and still be ineligible. The anchored ladder, the weights and the two gate positions are all drafted and all contested. Legacy Projection Metrics (LPM) is the projection layer. Four cost layers kept deliberately separate, a generational horizon, real constant dollars, and no discount rate applied to physical entries. It exists as a working cost engine: fourteen linked sheets, a confidence tier and a source on every input, a three-way allocation splitting each line onto the operator, the community and the successor, and a live A-versus-B comparator. The Generational Cost of Ownership (GCOO) is what the two produce together, stated as a range rather than a point. Underneath sits a taxonomy of ten sectors by eight domains, eighty cells, sorting capability by the physical function a solution performs rather than by the organism it came from. What state it is actually in. Nobody outside its author has rebuilt the engine, run it against a real project, or checked whether two people starting from the same inputs land on the same answer. No inter-rater reliability study exists. Two of the seven questions carry weight without an auditable ladder. The rank scale is undecided. The community gate, which can veto a project outright, is the least specified input in the whole instrument. Several constants sit at scenario confidence and are labeled as such. We publish that list rather than defend against it. A standard that arrives finished arrives owned by whoever wrote it, and that is the one thing this cannot be. How the group works. Five packages, running in parallel. One, reconcile the engine. Two modelers independently rebuild the twenty-year chain from the input sheet without reading the model sheet, then compare. Any gap is either an error or an undocumented assumption, and both are findings. Two, replace a declared input. Several constants rest on judgment rather than evidence: carbon price, cooling capital per point of efficiency, grid-stranding rate, delay carrying cost, ecosystem-service value. Bring a real contract for one. Three, redline a ladder. One question per practitioner. Do the anchors survive contact with someone who holds the drawings, and is the evidence test producible? Then set the weight. Four, take a denominator for a sector that is not data centers. Five, build a second engine, independently, on the same standard, and report every place you had to make a judgment call we failed to document. Several engines that agree is evidence. Several that disagree is more valuable still, because the disagreement locates precisely which input the industry has to settle. Packages one and two are weeks. Package three is the working block. Four and five outlast the November session and are what turn a workbook into a base case. The reason for the deadline. The comparable standard in semiconductor manufacturing took eight years, and most of those years went into inventing the practice of standard-making as it went. We are not starting there. The precedent exists, a draft of every instrument exists, and circulating a redline to forty people in three countries is now free. The ambition is eight months. The constraint is how quickly the right people get around the same table.
What this group is for. Three instruments exist in draft and none has been ratified by anyone but its authors. This group's job is to break them where they are weak, rebuild the parts that do not survive, and close six specific questions before 9 and 10 November 2026. What exists today. The BiosVerse-Compatibility Index (BCI) is a recognition standard. Seven questions asked of every project in the same order, scored on evidence rather than assertion, building up toward 100. Five are weighted and two are gates, so a strong column cannot paper over a failed one and a project can score 88 and still be ineligible. The anchored ladder, the weights and the two gate positions are all drafted and all contested. Legacy Projection Metrics (LPM) is the projection layer. Four cost layers kept deliberately separate, a generational horizon, real constant dollars, and no discount rate applied to physical entries. It exists as a working cost engine: fourteen linked sheets, a confidence tier and a source on every input, a three-way allocation splitting each line onto the operator, the community and the successor, and a live A-versus-B comparator. The Generational Cost of Ownership (GCOO) is what the two produce together, stated as a range rather than a point. Underneath sits a taxonomy of ten sectors by eight domains, eighty cells, sorting capability by the physical function a solution performs rather than by the organism it came from. What state it is actually in. Nobody outside its author has rebuilt the engine, run it against a real project, or checked whether two people starting from the same inputs land on the same answer. No inter-rater reliability study exists. Two of the seven questions carry weight without an auditable ladder. The rank scale is undecided. The community gate, which can veto a project outright, is the least specified input in the whole instrument. Several constants sit at scenario confidence and are labeled as such. We publish that list rather than defend against it. A standard that arrives finished arrives owned by whoever wrote it, and that is the one thing this cannot be. How the group works. Five packages, running in parallel. One, reconcile the engine. Two modelers independently rebuild the twenty-year chain from the input sheet without reading the model sheet, then compare. Any gap is either an error or an undocumented assumption, and both are findings. Two, replace a declared input. Several constants rest on judgment rather than evidence: carbon price, cooling capital per point of efficiency, grid-stranding rate, delay carrying cost, ecosystem-service value. Bring a real contract for one. Three, redline a ladder. One question per practitioner. Do the anchors survive contact with someone who holds the drawings, and is the evidence test producible? Then set the weight. Four, take a denominator for a sector that is not data centers. Five, build a second engine, independently, on the same standard, and report every place you had to make a judgment call we failed to document. Several engines that agree is evidence. Several that disagree is more valuable still, because the disagreement locates precisely which input the industry has to settle. Packages one and two are weeks. Package three is the working block. Four and five outlast the November session and are what turn a workbook into a base case. The reason for the deadline. The comparable standard in semiconductor manufacturing took eight years, and most of those years went into inventing the practice of standard-making as it went. We are not starting there. The precedent exists, a draft of every instrument exists, and circulating a redline to forty people in three countries is now free. The ambition is eight months. The constraint is how quickly the right people get around the same table.
Potential Applications
Read as testing scope rather than as a market case. These are the domains the instrument has to survive before anyone should trust it.
Data centers, worked. The only domain with a complete case: a hundred megawatts run in full, twenty years, physical units before dollars, with the denominator settled as the metered megawatt-hour of IT load delivered and its weakness printed rather than buried. This is the reference case every other domain will be checked against, and it is also the domain where the capital is moving fastest and the community conflict is sharpest.
Process plants, agriculture, buildings, ports: unclaimed. The unit of production is settled for data centers and for nothing else. The denominator decides everything downstream of it, and picking it wrong makes the most careful arithmetic above it worthless. Each of these needs a room of that sector's own operators choosing the unit already sitting in their own contracts. Not a committee, and not us.
Project finance. The consequence the instrument is wired to. A lender underwriting a generational difference currently has no shared way to measure it. The test is whether a bond desk or a credit committee can use the output as it stands, and the honest expectation is that the first one who tries will find something we missed.
Permitting and community agreements. Projects die at the gate rather than on net present value. The community gate is both the most powerful input and the least specified, so this is the domain that most needs practitioners and least has them.
Industrial symbiosis and waste-heat reuse. A hundred megawatts of computing is also a hundred megawatts of heating, and a district network wants heat for five months a year while a greenhouse wants it for twelve. The instrument has to be able to price a generation of guaranteed warm water, which nobody currently can.
Public procurement. The standard should plug into readiness tiers that public buyers already use rather than asking anyone to learn a private vocabulary first. Whether it actually does is untested.
Supplier discovery. The taxonomy indexes capability by physical function crossed with buying sector. Whether eighty cells is the right resolution, and whether practitioners can place their own work in the right cell without help, is an open empirical question and a fast one to test.
Read as testing scope rather than as a market case. These are the domains the instrument has to survive before anyone should trust it.
Data centers, worked. The only domain with a complete case: a hundred megawatts run in full, twenty years, physical units before dollars, with the denominator settled as the metered megawatt-hour of IT load delivered and its weakness printed rather than buried. This is the reference case every other domain will be checked against, and it is also the domain where the capital is moving fastest and the community conflict is sharpest.
Process plants, agriculture, buildings, ports: unclaimed. The unit of production is settled for data centers and for nothing else. The denominator decides everything downstream of it, and picking it wrong makes the most careful arithmetic above it worthless. Each of these needs a room of that sector's own operators choosing the unit already sitting in their own contracts. Not a committee, and not us.
Project finance. The consequence the instrument is wired to. A lender underwriting a generational difference currently has no shared way to measure it. The test is whether a bond desk or a credit committee can use the output as it stands, and the honest expectation is that the first one who tries will find something we missed.
Permitting and community agreements. Projects die at the gate rather than on net present value. The community gate is both the most powerful input and the least specified, so this is the domain that most needs practitioners and least has them.
Industrial symbiosis and waste-heat reuse. A hundred megawatts of computing is also a hundred megawatts of heating, and a district network wants heat for five months a year while a greenhouse wants it for twelve. The instrument has to be able to price a generation of guaranteed warm water, which nobody currently can.
Public procurement. The standard should plug into readiness tiers that public buyers already use rather than asking anyone to learn a private vocabulary first. Whether it actually does is untested.
Supplier discovery. The taxonomy indexes capability by physical function crossed with buying sector. Whether eighty cells is the right resolution, and whether practitioners can place their own work in the right cell without help, is an open empirical question and a fast one to test.
Open Questions
The work program. Six questions, ordered by importance rather than by ease, each with the discipline best placed to answer it.
1. What does the community gate actually read? First in importance and hardest. One of the seven questions can veto a project outright regardless of engineering quality, and nobody has decided what community standing is, who measures it, on what evidence, or against what rungs. A survey can be bought. A hearing can be packed. A benefit agreement can be signed with ceremony by people who speak for nobody. Wanted: planning law, participatory governance, community benefit agreement practice, and people who have sat on the receiving side of a permitting fight.
2. Can questions three and six be made auditable? Nourishing the local biosphere, and operating like an ecosystem rather than a machine. Both carry weight and neither has a ladder that would survive a hostile audit. Every candidate proxy we have found is the kind that turned a generation of sustainability scoring into theater. Either someone builds an anchored, auditable ladder for both, or their weight comes down until one exists. Wanted: ecologists, ecological survey practitioners, eDNA and bioacoustic monitoring.
3. Coarse scale or fine? Each question is scored against five named anchors. Whether the line between them runs 0 to 4 or 0 to 10 changes how finely two designs separate: a worked comparison ties two projects at 71 on the coarse scale and separates them to 77 and 63.5 on the fine one. A governance decision, not an arithmetic one, and it needs deciding before anyone scores a real project.
4. What is the unit of production outside data centers? Cost per what. Settled for data centers, open everywhere else. Wanted: operators from process industries, agriculture, buildings and ports who can say what unit their own contracts already run on.
5. Would two trained assessors reach the same score? No inter-rater reliability study has been run. We do not believe two assessors scoring the same site today would land within a point of each other. Designing and running that study is a discrete, publishable piece of work and it is unclaimed.
6. Can an independent team rebuild the engine and get a comparable answer? Nobody outside its author has tried. Several engines that agree is evidence; several that disagree is better, because the disagreement locates the input the industry has to settle.
A seventh, technical and well defined. The proposal states that the output is a range, never a point, with its width set by the confidence tiers of the inputs. The current engine produces point figures and nothing in it computes a width. Closing that gap is a bounded modeling problem and a good first contribution.
The work program. Six questions, ordered by importance rather than by ease, each with the discipline best placed to answer it.
1. What does the community gate actually read? First in importance and hardest. One of the seven questions can veto a project outright regardless of engineering quality, and nobody has decided what community standing is, who measures it, on what evidence, or against what rungs. A survey can be bought. A hearing can be packed. A benefit agreement can be signed with ceremony by people who speak for nobody. Wanted: planning law, participatory governance, community benefit agreement practice, and people who have sat on the receiving side of a permitting fight.
2. Can questions three and six be made auditable? Nourishing the local biosphere, and operating like an ecosystem rather than a machine. Both carry weight and neither has a ladder that would survive a hostile audit. Every candidate proxy we have found is the kind that turned a generation of sustainability scoring into theater. Either someone builds an anchored, auditable ladder for both, or their weight comes down until one exists. Wanted: ecologists, ecological survey practitioners, eDNA and bioacoustic monitoring.
3. Coarse scale or fine? Each question is scored against five named anchors. Whether the line between them runs 0 to 4 or 0 to 10 changes how finely two designs separate: a worked comparison ties two projects at 71 on the coarse scale and separates them to 77 and 63.5 on the fine one. A governance decision, not an arithmetic one, and it needs deciding before anyone scores a real project.
4. What is the unit of production outside data centers? Cost per what. Settled for data centers, open everywhere else. Wanted: operators from process industries, agriculture, buildings and ports who can say what unit their own contracts already run on.
5. Would two trained assessors reach the same score? No inter-rater reliability study has been run. We do not believe two assessors scoring the same site today would land within a point of each other. Designing and running that study is a discrete, publishable piece of work and it is unclaimed.
6. Can an independent team rebuild the engine and get a comparable answer? Nobody outside its author has tried. Several engines that agree is evidence; several that disagree is better, because the disagreement locates the input the industry has to settle.
A seventh, technical and well defined. The proposal states that the output is a range, never a point, with its width set by the confidence tiers of the inputs. The current engine produces point figures and nothing in it computes a width. Closing that gap is a bounded modeling problem and a good first contribution.
AI Assistant
Summarize published methods for establishing inter-rater reliability in qualitative assessment rubrics used for infrastructure or environmental certification, and for measuring the enforceability of community benefit agreements. Prioritize methods with reported reliability statistics over frameworks and position papers.
Summarize published methods for establishing inter-rater reliability in qualitative assessment rubrics used for infrastructure or environmental certification, and for measuring the enforceability of community benefit agreements. Prioritize methods with reported reliability statistics over frameworks and position papers.
Keywords
Proposal
Proposal
Email
michael.wright@interceptinghorizons.com
michael.wright@interceptinghorizons.com