Solutions Must Survive The Three Filters
A transition pathway has to survive the STEM, beat all of the alternatives after the full chain is counted, and pass through the institutions and people required for adoption.
A lot of transition analysis gives too much credit to technologies that can be made to work and not enough scrutiny to whether they matter. That distinction shows up everywhere: in carbon capture hubs, synthetic fuel claims, small modular reactor schedules, cement decarbonization pathways, aviation fuel projections, ammonia shipping forecasts, grid storage proposals, critical-mineral alarm bells, and hydrogen strategies. A process can work in a lab, a pilot, a funded demonstration, or a polished diagram and still fail as a useful pathway for climate, capital, or policy.
One of my maintained assumptions is simple: claimed solutions have to pass three filters before they belong in a serious long-range projection. They have to survive the STEM and the internal cost stack. They have to beat all of the alternatives after the full chain is counted. They have to pass through the institutions, firms, workers, customers, and voters required for adoption. Failing any one of those filters does not make the technology fake. It means it should not be treated as a major future pathway until the failure is resolved.
The first filter is technical, but not in the soft sense of asking whether something can be made to work once. Almost anything can be made to happen once with enough money, attention, engineering talent, and tolerance for inconvenience. The harder test is whether the proposed solution hangs together as science, engineering, operating system, and real-world cost stack at the scale being claimed. Before a pathway competes with alternatives, it has to survive its own denominator.
Cement is a useful example because every serious decarbonization pathway has a different internal stack. Clinker substitution has to survive supplementary cementitious material availability, cement standards, performance, logistics, and construction practice. Electrified heat has to survive kiln temperature, heat transfer, retrofit complexity, electricity supply, utilization, and capital cost. Carbon capture has to survive capture energy, kiln integration, compression, transport, storage, monitoring, liability, and cost per durable tonne. Alternative binders have to survive chemistry, feedstock volume, durability, standards, and contractor acceptance. The fact that cement is hard to decarbonize does not make every cement pathway plausible.
Steel is similar. A route does not pass the first filter because it produces iron once. It has to survive demand, stock, scrap availability, ore quality, new iron requirements, furnace configuration, electricity supply, reducing-agent supply where needed, capital cost, utilization, and product quality. Scrap-electric arc furnaces, electrochemical ironmaking, direct reduced iron, biomethane, biocarbon, and hydrogen routes all have different internal stacks. Before they compete with one another, each one has to hang together on its own.
This is the receipts layer. It is where tonnes, joules, kilowatt-hours, temperatures, flow rates, concentrations, pressures, degradation rates, capacity factors, asset lifetimes, balance-of-plant costs, and cost per delivered unit enter the room. The same discipline applies to ocean carbon capture, gravity storage, subsurface energy storage, direct air capture, shipping fuels, and aviation fuels. The mechanism may be real and the pilot may run; the full operating system can still be too energy-intensive, too material-intensive, too fragile, too hard to monitor, too slow to scale, or too expensive.
The first filter therefore asks whether the proposed solution is internally coherent and costed honestly. Carbon dioxide captured at a stack is not carbon dioxide permanently stored. Synthetic fuel at a reactor outlet is not certified fuel delivered to an airport. Ammonia at a production plant is not safe marine fuel available at the right port, on the right route, with trained crew, emergency response, insurance, and regulation. A molecule produced at one end of an energy chain is not automatically useful energy, useful heat, industrial output, transport service, or reliable reserve at the other end. The diagram is not the asset.
The second filter is economic competition. Once a proposed pathway has survived its own science, engineering, operating system, and cost stack, it has to beat the alternatives. The comparison is delivered service, not elegance, novelty, or addressable market size. A technology that is internally coherent but more expensive, more complex, less convenient, slower to build, or more fragile than a simpler competitor does not become a central transition pathway because it made it through the first filter.
This is where many plausible pathways get smaller. Passenger transport does not need to preserve fuel distribution when battery-electric drivetrains are simpler and more efficient for most use cases. Most building heat does not need a new clean fuel chain when heat pumps deliver much more useful heat from the same electricity. Synthetic liquids can run internal combustion engines, but electric drivetrains remove the need for the liquid fuel in most road transport. Carbon capture can reduce emissions from some concentrated streams, but it has to compete with process change, electrification, substitution, and demand reduction. Low-carbon marine fuels can move ships, but they compete against batteries, shore power, wind assistance, speed changes, hull efficiency, logistics, port constraints, safety, and changes in cargo itself.
The stronger pathways tend to have shorter causal chains and fewer things that have to go right. Efficient electric motors, heat pumps, batteries for daily transport and many grid-balancing duties, low-loss wires, mature rail electrification, methane detection and repair, and scrap-electric arc furnaces all solve defined service problems without requiring long chains of new conversions to preserve an old fuel pattern. None of them is effortless, and all of them need deployment systems. Their advantage is that the technical and economic cases usually start closer to the delivered service.
The third filter is social and institutional adoption. That is broader than public acceptance and harder than public relations. A pathway has to pass through households, firms, workers, regulators, courts, insurers, utilities, ports, municipalities, emergency responders, procurement departments, and voters. It also has to deliver something people actually value. Technologies scale faster when they are cleaner and better, not merely cleaner and more virtuous.
Heat pumps show why adoption is not just an institutional question. A good heat pump provides heating, cooling, filtration, quiet operation, and often lower total household energy cost, especially as buildings improve and electricity gets cleaner. Installer shortages, retrofit complexity, tariff design, and landlord-tenant problems still matter, but the adoption case is much stronger when the cleaner product also improves the service.
Electric vehicles have a similar adoption advantage. They are not just emissions-reduction devices. They accelerate quickly, drive smoothly, need less routine maintenance, can be charged at home or depot for many use cases, and integrate naturally with a cleaner grid. Charging infrastructure, upfront cost, apartment access, road trips, cold weather, and heavy-duty duty cycles still matter. But the pathway does not require most drivers to accept a worse vehicle. In many segments it offers a better one.
The opposite case is a pathway that depends mainly on abstinence, inconvenience, or moral discipline. Some demand reduction is real and necessary, but any solution that requires everyone to wear a hairshirt is a weak central case. Getting billions of people to stop eating meat is harder than reducing livestock emissions, shifting feed, cutting methane, improving manure management, changing breeding, reducing waste, and expanding the share of lower-impact foods where consumers already accept them. Diets can and will change, especially with income, health concerns, price, culture, and better substitutes, but a strategy that depends on universal renunciation is not the same as a deployable transition pathway.
Institutional adoption still matters. Transmission is one of the best answers and still runs into land-use conflict, interconnection queues, cost allocation fights, fragmented authority, permitting delays, and public distrust. Those are not reasons to give up on transmission. They are reasons to treat transmission deployment as a real institutional task rather than a line on a model.
Ammonia shipping faces a different adoption test. The molecule has advantages on paper. It contains no carbon atom and is easier to store than hydrogen. But using it as marine fuel means bunkering systems, port safety cases, vessel design, engine-room procedures, crew training, emergency response, regulation, insurance, and tolerance for toxic fuel handling near busy ports. That does not make ammonia impossible. It means an ammonia forecast that stops at energy density and production cost has stopped too early.
The filters matter because early evidence is routinely overread. Grants, memoranda of understanding, pilots, demonstrations, patents, targets, and friendly policy language all have meaning, but only at the right stage of evidence. A grant may buy learning without proving demand. A pilot may prove an operating point without proving repeat procurement. A carbon capture hub may have public funding and an industrial map while still lacking pipeline rights, storage certainty, liability clarity, and durable customers. An airline may have SAF commitments while the available supply remains small, expensive, and contested by feedstock limits. A low-carbon fuel strategy may have production targets while still lacking cheap delivered fuel, high-utilization demand, infrastructure, permits, and customers willing to pay. The error is counting these as scaled transition supply before the filters have been passed.
The three filters are not a bias against new technologies. They are a bias against unearned certainty. Solar, wind, batteries, heat pumps, electric vehicles, methane detection, and grid technologies became important as technical performance, delivered economics, and adoption systems improved together. Manufacturing scaled, costs fell, supply chains learned, regulators adapted, customers bought again, and installers and operators gained experience. Passing the filters looks like ordinary institutions and ordinary customers choosing the pathway repeatedly because it has become the better way to get the job done.
The filters are also dynamic. A pathway that fails today can pass later. Costs can fall, materials can change, supply chains can diversify, standards can mature, permitting can improve, and specific niches can become valuable enough to justify complexity. But the reverse is also true. A pathway that looks plausible at pilot scale can fail when utilization falls, safety cases harden, feedstocks tighten, insurance costs rise, customers dislike the result, or a simpler competitor improves faster. Long-range projections should not freeze today’s technology rankings, but they should not assume away today’s unresolved filters either.
For 2100 work, this changes the denominator. A model should not ask only what could theoretically supply future demand. It should ask what survives enough of the real system to deserve a material share. Aviation liquids remain because long-haul aircraft still require dense fuels, but the scale is residual after efficiency, regional electrification, price sensitivity, feedstocks, and route changes are counted. Steel still needs primary iron, but the denominator comes after demand, stock, scrap, ore quality, electricity, route choice, electrochemical options, biomethane, biocarbon, and hydrogen are compared. Grid storage still needs reliability, but rare-duration problems do not automatically become a standing market for a preferred molecule. Critical minerals matter, but durable assets, recycling, chemistry shifts, and material efficiency change the long-term extraction story.
The evidence to watch is practical: delivered cost, utilization, repeat procurement, actual permits, interconnection, storage rights, pipeline routes, insurance, safety incidents, workforce requirements, customer satisfaction, and repeat purchases. These reveal whether the pathway is leaving the protected world of pilots and becoming ordinary infrastructure or an ordinary product. The best evidence is not that something can be made to work once. The best evidence is that the relevant institutions and customers start choosing it repeatedly because it is the best available way to get the job done.
That is the worldview assumption: solutions must survive the three filters. Technologies that pass the STEM test, beat the alternatives after the full chain is counted, and make it through adoption reality deserve capital, policy attention, and a place in serious projections. Technologies that have not passed them may still deserve research, pilots, or niche use. They do not deserve to be treated as central transition pathways merely because someone needs them to be.
I do not claim to be right. I claim to be less wrong than most. In this case, being less wrong means answering three questions before believing a transition claim: does it hang together as science, engineering, operating system, and cost stack; does it beat the alternatives after the full chain is counted; and will the real world choose and deploy it repeatedly?
Subscribe to TFIE Strategy Briefing for the professional due-diligence layer behind transition claims: science and cost-stack tests, comparator cases, adoption filters, scorecards, workbooks, and update triggers.