The Platform Is the Easy Part
The platform is the easy part. The harder work is translating a governing agenda across the country’s competing moral vocabularies without losing the policy core.
Note: This essay was based on research papers exploring the ideal liberal and conservative political platforms for 2026. Jump to the underlying research papers.
Politics usually starts from the wrong place. A party begins with its people, its history, its donors, and its old enemies, and it works toward policy only once all of that has been settled. The platform arrives late, after the slogans and the sorting are already done. By the time a voter hears it, the thing no longer sounds like a plan for making rent payable or institutions sturdier. It sounds like a signal of who belongs in the room and who should be resented from outside it.
I wanted to know what happens when you reverse that order. Start from the problem, judge each idea by evidence and by the limits of administration, and let identity arrive last. So I set out to build the strongest liberal platform I could, then the strongest conservative one, and to hold both to the same standard rather than to whichever tribe they flattered. The point was never to find a soft center or to prove that the two sides secretly agree. They do not. The point was to see what each tradition produces when it honors its real commitments and then has to defend them against arithmetic, capacity, and law.
A surprising amount of ordinary political language did not survive that test, and watching it fail is most of what the exercise was for. Two problems surfaced along the way, one about how a platform is designed and one about how it is heard, and they turned out to be the same problem seen from two sides.
Where the evidence came from
The work rested on a mix of sources, and they were not given equal authority. Governors and long-serving elected officials showed where a coalition was heading, which is different from where its activists say it is heading. Think tanks and policy economists were useful when they explained a mechanism and far less useful when they were doing advocacy with footnotes. Government evaluations and quasi-experimental studies carried the most weight on questions of effect, because they are the closest thing available to a controlled answer about whether a policy did what it promised. Legal scholarship set the outer boundary of what a given level of government may do at all. And newer political figures were worth reading mainly for the pressures they named, the things an older coalition had trained itself to stop noticing.
None of those sources counted as final on its own. An idea with elegant theory and no delivery record was treated as a candidate for a pilot, not a plank. That is why the agenda that survived leans on housing supply, Medicaid, the Child Tax Credit and Earned Income Tax Credit, apprenticeships, and tutoring, where both the evidence and the administrative path are relatively mature. It also explains why a few ideas sit at arm's length as experiments rather than promises, each for a different reason. Carbon pricing has a deep economic literature behind it but carries high political transition risk. Hospital rate-setting has real implementation evidence in Maryland but depends on a level of administrative competence most states have not built. The auditing of artificial-intelligence systems is genuinely immature, because the thing being audited keeps changing under the auditor. Evidence had to survive contact with implementation before it earned a place.
A plank has to answer for itself
The test I used is not complicated, which is part of why it works. Before a policy earns a place on a platform, it has to answer five plain questions. What problem does it solve. How does the mechanism work. Who benefits, in terms a person could recognize in their own life. What guardrail keeps it from being abused or going wrong. And which institution is supposed to carry it out, with what money and what legal authority
Many ideas fail before they become policy.
Most modern politics dies somewhere in those five questions. "Build more housing" collapses the moment you ask where, by what authority, with what infrastructure, and with what protection for the people already living there. "Secure the border" falls apart when it has to account for courts with finite calendars, detention with finite beds, and employers whose demand for labor does not disappear because a speech asked it to. "Tax the rich" rarely survives the question of what is collected, what behavior changes in response, what the money funds, and what stops the wealth from simply moving. "Defend democracy" tends to evaporate when you press it past mood and into paper ballots, accurate rolls, transparent procedures, and rules against self-dealing.
None of this is a partisan observation. The same questions cut through grievance on the right and abstraction on the left with equal ease. What remains after they have done their work is a much smaller set of ideas, but a far more durable one. Those surviving ideas share a quality that is easy to miss. They are specific enough to be administered, and they name the body that would have to administer them.
The deeper move, and the one most platforms skip, is to take an idea that survives those five questions and push it one layer further. A surviving idea still needs the precise recommendation rather than the direction, the agency that holds the pen, the cost and the way it is paid for, the constitutional risk it runs, and the order in which its parts come online. Housing is the cleanest place to watch that deepening happen, so it is worth slowing down there before going anywhere else.
Housing, all the way down
Housing sits underneath almost every other problem a family has. A household paying half its income in rent has little left for child care, medicine, savings, or the move to a town with better work. A region that blocks homes near its jobs converts opportunity into a bidding war and then wonders why its workers commute ninety minutes each way. Scarcity in housing behaves like a tax on ordinary life, and it is collected most heavily from the people with the least room to absorb it.
The first instinct on the left is rent control, and the evidence on it is genuinely mixed rather than simply negative. In San Francisco, the economists Rebecca Diamond, Tim McQuade, and Franklin Qian found that tenants who held rent-controlled units gained real protection from displacement, which is not a small thing when leaving an apartment can mean leaving a job, a school, and a network of people who help raise your kids. The same study found that landlords responded by converting and redeveloping units out of the rental market, which tightened supply and pushed the pressure onto everyone who did not already hold a protected lease.[^1] Austin tells the other half of the story in plainer terms. A surge of permitted construction was followed by falling rents across the metro.[^2] Minneapolis adds a quieter version of the same lesson, where an early move to legalize more housing tracked with more moderate rent growth than comparable cities saw.[^3]
The honest reading is that tenant protection helps the people it covers and does nothing to add a single bedroom to the housing stock. Protection without supply is rationing with better paperwork. That conclusion is uncomfortable for parts of the left, and the platform has to state it anyway, because the alternative is to keep promising relief through a mechanism that cannot produce it.
So the surviving housing plank is supply, and this is where the work gets harder. It is not enough to say "legalize more homes." The precise recommendation is to make a defined set of housing types legal to build by right, meaning a project that meets objective written rules can proceed without a discretionary political approval at the end. Accessory dwelling units, the backyard cottages and garage apartments that older neighborhoods always had, should be permitted on residential lots without owner-occupancy mandates, with ministerial review on a clock measured in weeks rather than years, and without the parking requirement that quietly kills them. Duplexes, triplexes, and small apartment buildings, the missing middle that zoning made illegal across most of the country, should be allowed near jobs, transit, and commercial corridors. Minimum-parking requirements, which can add tens of thousands of dollars in concrete to the cost of a single home regardless of whether anyone living there owns a car, should be removed near transit and pared back elsewhere.
Each of those recommendations has an owner. Most of them belong to states, not to Washington, because zoning is delegated to states and exercised by cities. California showed what happens when a state preempts local rules on accessory units, fee caps, and parking, and watched a tolerated form turn into tens of thousands of permits a year.[^4] Colorado and a handful of other states have moved along the same path. The federal role is real but narrower, and it works mostly through money, financing, and the conditions attached to federal programs rather than through direct mandate. Washington can attach housing-supply conditions to transportation and infrastructure grants, fund the planning capacity that small jurisdictions lack, and underwrite the construction finance that the private market will not extend to nonprofit and public builders.
Subsidy is the part of the housing plank that costs real money, and it has its own discipline. When a city lowers a project's costs through fee waivers, loans, or tax rebates, the public should be able to read exactly what it bought. San Antonio publishes its incentive agreements at the project level, with the private investment required, the number of units, how many are affordable and at what share of area median income, the value of the rebate, and the enforcement terms.[^5] That kind of public ledger is the difference between leverage and a gift, because some of these agreements buy durable affordability and some of them merely subsidize buildings that would have risen anyway. A platform that hands out incentives without that ledger has stopped doing housing policy and started doing patronage with a press release.
The whole package carries legal risk that an honest plank should name rather than hide. State preemption of local zoning invites litigation and political revolt from cities that experience it as a loss of self-government, and the durable versions pair preemption with infrastructure money and a real appeals process so that the state is offering a bargain rather than issuing an edict. Done carelessly, preemption produces a backlash that repeals the reform within an election cycle. Done with that bargain in view, it survives.
That is what one layer deeper looks like. The same template runs cleanly across the rest of the platform. Take the mechanism and the evidence, which are the easy columns, and then force the harder four. State the precise recommendation rather than the direction. Name the agency. Attach the cost and the way it is paid. And describe the sequence and the legal exposure. The columns that platforms usually leave blank are exactly the ones that decide whether anything gets built.
Public safety as an operating model
Public safety responds to the same scrutiny, and the result cuts against the loudest voices on both sides. The weak version of the debate asks whether a politician is tough or soft. The stronger version asks what reduces serious harm, and the evidence there points somewhere more specific than either camp's instinct.
Serious violence is not spread evenly across a city. It concentrates around a small number of places, a small number of people, and a small set of recurring conditions. That fact changes the operating model from posture to logistics. Hot-spot policing, which concentrates attention on the blocks and hours where harm recurs, has a stronger evidence base than broad random patrol.[^6] Focused deterrence, which identifies the few people at high risk of serious violence and pairs a clear consequence with a genuine offer of help, has produced measurable reductions in shootings in several cities.[^7] Neither of these resembles indiscriminate enforcement, and their value depends on precision and legitimacy rather than on volume.
The unglamorous half of the model is capacity. Clearance rates govern deterrence, because crimes that go unsolved teach a neighborhood that cooperation is pointless and that the state cannot protect a witness. Detectives need enough time and forensic support to close serious cases. Witnesses need protection if the system expects them to testify. And a large share of what arrives as a public-safety problem is really untreated addiction, psychosis, or homelessness, which means a city needs somewhere to send a crisis that is not a jail cell and someone to answer the call who is not always a patrol officer. Permanent supportive housing, which combines long-term housing with services for people cycling through streets, emergency rooms, and jails, has strong evidence for housing stability even where its broader cost savings are harder to prove.[^8]
A municipal version of this exists in detail, built around a coordinating body that pulls mental-health response, violence interruption, and homelessness services into one agency so that someone clearly owns the step after the 911 call, the arrest, or the overdose. Stripped of its local particulars, the transferable architecture is straightforward. Fund the detectives and the forensics that close serious cases. Run focused deterrence on the small number of people driving violence. Stand up a civilian crisis response for calls that are medical rather than criminal. Build supportive housing for the chronically homeless population that otherwise rotates through the most expensive public systems a city operates. Measure homicides, nonfatal shootings, clearance rates, response times, overdose encounters, and unnecessary arrests, and let those numbers, rather than a slogan, tell you whether the model is working.
Where the platforms converge, and where they do not
It is worth setting the two platforms beside each other plainly, because the overlap and the divergence are both more specific than the prose has so far let on. The shared terrain is real, and so is the disagreement, and a reader trying to use any of this needs to see which is which rather than take my word that both exist.
| Policy area | Liberal plank | Conservative plank | Where they converge | Where they really divide |
|---|---|---|---|---|
| Housing | Supply plus tenant protection, nonprofit and public capacity, vouchers | Supply through property rights, zoning reform, a local-control bargain | Build more homes, cut permitting friction, target subsidies to results | The role of public and nonprofit building, tenant regulation, how hard a state should preempt cities |
| Public safety | Focused enforcement with treatment, accountability, supportive housing | Public order, serious violence first, treatment capacity, due process | Concentrate on serious violence, build crisis capacity, reject theater | Police accountability, sentencing severity, prosecutorial discretion |
| Immigration | Lawful pathways, due process, labor protection, humane order | Border control, lawful hiring, faster adjudication, employer penalties | Processing capacity, employer enforcement, legal channels sized to demand | Enforcement intensity, humanitarian obligation, whether to legalize the settled undocumented |
| Technology and privacy | Civil rights, consumer protection, labor transition, AI accountability | Family autonomy, free expression, child safety, limits on surveillance | Privacy law, data-broker limits, accountable AI, no broad identity gate | Speech regulation, platform liability, labor rules for automation |
| Energy | Cleaner grid, transmission, storage, adaptation, carbon price with a rebate | Abundant reliable power, permitting, nuclear and gas, grid security | Build generation, transmission, and the permitting capacity to approve it | Carbon pricing, the speed of the fossil transition, how climate is framed |
| Work and skills | EITC, child care, apprenticeships, labor standards | Work incentives, apprenticeships, benefit cliffs, small business | Paid routes into useful work without a mountain of debt | Unions and worker power, wage floors, how subsidy is designed |
This is not a unity platform, and nothing in the table dissolves the disagreements in the last column, which rest on genuinely different beliefs about authority, markets, and the purpose of the state. What the table offers is a map of where cooperation is structurally plausible and where differentiation is honest. A politician looking for common ground should start in the convergence column, where both platforms collide with the same delivery constraints and where a deal is held together by physics and administration rather than by sentiment. A politician looking to differentiate should do it in the last column, where the moral disagreement is real, rather than manufacturing conflict over machinery that both sides need built. The failure mode of current politics is the reverse. It invents fights over the shared infrastructure and goes quiet on the disagreements that voters deserve to adjudicate.
The translation problem
That map is where the first problem hands off to the second. Designing a plank that survives the five questions is necessary, and it is not enough, because good policy fails routinely before voters ever understand it.
The clearest warning came from an experiment that Planet Money ran in 2012. The show asked economists from across the political spectrum to assemble a platform out of ideas they could broadly agree on. The resulting candidate supported things with real expert backing, including trimming popular tax deductions, shifting toward taxing consumption, putting a price on pollution, and legalizing marijuana. On paper the platform was defensible.[^11]
In front of voters it was radioactive. People heard the loss first and the gain never. A mortgage deduction disappearing is concrete and immediate. The diffuse, delayed benefit of a less distorted tax code is something you have to be talked into, and most people are not sitting in a seminar waiting to be talked into it. Later episodes brought in consultants and focus groups to package the platform, which sharpened the lesson instead of solving it.
The durable finding underneath that experiment is that the size of a change governs how much disruption a voter feels before they can weigh the argument for it. The bigger the reform, the more the near-term pain arrives ahead of the long-term case. A person asked to tolerate disruption for a future payoff will first ask what happens to their bill, their job, their town, and their kid's school. That makes transition design a part of policy design rather than an afterthought. A platform that wants people to accept short-term cost has to build the bridge before it starts the sale, which usually means a visible rebate, a phase-in, or a credit that lands in the same window as the disruption. Translation cannot rescue a policy whose transition was never designed.
When policies do reach voters, they arrive wrapped in coalition code. Housing supply becomes a developer giveaway to one audience and social engineering to another. Immigration competence becomes amnesty on the right and cruelty on the left. Public safety becomes defund or lock them all up. Carbon pricing becomes a clever new way to raise the grocery bill. The substance never changes, but the private vocabulary of whichever coalition is speaking determines how the policy is heard, and most of the time it is heard as a threat.
So I asked whether a strong policy could be explained to very different audiences without changing its substance at all. The answer, for policies that had survived the five questions, was usually yes. A sound policy can be defended honestly through several different values. A bad one usually cannot, and the moment persuasion starts to require a hidden cost or a quiet change in the policy itself, the communication problem has exposed a policy problem.
Housing shows how the same plank reaches four different doors. A voter on the center-left does not need a lecture on property rights. They need to hear that more homes mean less displacement and that targeted protection can hold people in place while supply catches up. A voter further left needs the power analysis, that scarcity is itself a gift to incumbent owners and exclusionary towns, and that public land and nonprofit builders belong in the toolkit alongside private construction. A voter on the center-right hears the family and ownership logic first, that a homeowner should be allowed to add a safe apartment for an aging parent or a local worker and that a permit should not take longer than the build. And a voter on the far right, suspicious of anything resembling metropolitan planning, can hear the same plank as local continuity, because a town that bans starter homes exports its own children and imports long commutes. One policy, four honest framings, no lie required.
Immigration travels the same way once the plank itself is serious, and on the liberal side it usually needs to be made serious first. A credible immigration plank is a working system rather than a posture in either direction. It combines border-processing capacity, asylum adjudication fast enough that people are not left in limbo for years, an entry-exit system that knows who overstayed, employment verification built with privacy protection and a real way to correct errors, penalties aimed at employers who exploit illegal labor, and legal channels sized to the work the country needs. That single design reaches the center-left as lawfulness and humane order, the left as protection for workers against exploitation that hurts migrants and citizens alike, the center-right as seriousness about a controlled border and lawful hiring, and the far right as sovereignty, the plain idea that a country decides who enters and can see that its rules are enforced. The emphasis shifts by audience. The architecture does not.
Energy is the example with the most current weight, because it now sits underneath household bills, manufacturing, artificial intelligence, and national security at the same time. The plank is to build the physical system a cleaner and more productive economy needs, which means transmission lines, reliable generation, storage, grid modernization, faster interconnection, and the permitting capacity to approve any of it before the decade is out. A carbon price belongs in that plank only when it is paired with a visible rebate or family tax credit, because a price signal that arrives as an unexplained bill will be repealed before it works. A voter on the center-left hears the emissions and reliability case, that a cleaner grid which cannot keep the lights on will lose public confidence no matter how clean it is. A voter further left hears community benefit and public value, that public money should buy measurable emissions cuts and local jobs rather than become a subsidy stream for firms with weak accountability. A voter on the center-right hears infrastructure and affordability, that power should stay reliable and bills tolerable, and that new capacity should be built before anything old is shut off. And a voter on the far right hears household independence and grid security, that a family needs power it can count on and a country needs a grid that is not a standing vulnerability. The reason climate messaging so often fails is that it begins with planetary obligation instead of the grid people live on, where they notice outages, bills, wildfire smoke, and the project that took nine years to permit. The architecture does not change across those four doors. Only the value through which it is explained does.
The point of those translations is honesty rather than manipulation. A platform written entirely in one coalition's dialect will be rejected by everyone outside it before the policy is even heard. Voters can sense factional vocabulary. They know when a speech is built out of internal code, when a villain has been named but no mechanism has, and when compassion has no operating model or toughness has no implementation path behind it.
Messaging is a policy audit
The examples point to a rule that runs in the opposite direction from how campaigns usually treat communication. Messaging works less like paint applied to a finished platform and more like a stress test the platform has to pass before it is ready.
A housing policy that cannot be explained to renters, homeowners, builders, and the people afraid of being pushed out almost certainly has an unresolved design problem underneath the words. An immigration plan that cannot account for enforcement, labor demand, due process, employer penalties, and local service pressure in one coherent breath is a slogan wearing a policy's clothes. A health plan that cannot explain how people keep their coverage during the transition is not finished. An energy plan that cannot speak to power, water, land, cost, and reliability at once is not yet an energy plan. The five questions that build a plank are the same five questions that test whether it can be said out loud without falling apart.
Those questions also surface coalitions that the usual map hides. A voter on the far left and a conservative can meet on breaking up concentrated private power. A center-right voter and a liberal can meet on apprenticeships, on the idea of useful work without a mountain of debt. A far-right voter and a liberal can meet on auditable elections, on showing the work rather than asking for trust. A center-left voter and a conservative can meet on privacy, and privacy becomes concrete the moment you stop saying the word and describe the thing. A purchased location trail is a record of where a phone slept and where it spent its days, which means it can mark a clinic visit, a recurring stop at a church, a night at a domestic-violence shelter, a protest, or the gate of a military base, and that record is bought and sold by brokers most people have never heard of and cannot inspect. A conservative can object to that as an assault on family autonomy and a liberal as a consumer and civil-rights harm, and both arrive at the same rule. None of these alignments dissolves the deep disagreements, which are real and rest on genuinely different claims about authority, family, markets, and risk. They simply show that more policy can travel across the spectrum than the parties believe, and that most of it fails for reasons of language rather than reasons of merit.
Paying for it in public
A platform that asks to be taken seriously has to be honest about money, and this is the discipline that even careful platforms most often skip. The same five questions that interrogate a policy should interrogate the budget around it. What is collected, what is spent, what is funded out of revenue, and what is left on the deficit. Fiscal realism is not a conservative tic but the thing that keeps a social commitment durable instead of leaving it exposed to the first downturn that comes looking for something to cut.
What follows is an illustration rather than a score, and the distinction is worth stating plainly. Real numbers require the Congressional Budget Office and the Joint Committee on Taxation, and they depend on assumptions about behavior that move the results by large amounts. These are orders of magnitude drawn from existing public estimates, meant to show that the major pieces of the platform can be set against one another honestly rather than to pretend a precision I do not have.
| Plank | Direction | Rough annual magnitude | Basis and main uncertainty |
|---|---|---|---|
| Carbon price with dividend or credit recycling | Revenue | On the order of $100 billion a year | A $25-per-ton price rising over time was scored near a trillion dollars across a decade. The net depends on how much is returned to households.[^12] |
| Drug-price negotiation | Savings | Tens of billions a year, building over time | Federal savings from Medicare negotiation were estimated near $100 billion across a decade, growing as more drugs come into scope.[^13] |
| Permanent expanded Child Tax Credit | Cost | Roughly $100 billion a year | The 2021 fully refundable credit ran near that figure for the year. Long-run modeling suggests large offsetting gains in earnings and health that are real but slow.[^14] |
| EITC expansion | Cost | Tens of billions a year | Scales with the size of the expansion and how far it reaches childless workers.[^16] |
| Medicaid expansion completion and ACA subsidy preservation | Cost | Tens of billions a year | Driven by how many remaining states expand and whether enhanced subsidies are made permanent.[^17] |
| Housing supply reform | Near zero direct cost | Mostly regulatory | Preemption, by-right rules, and parking reform cost little. The subsidy and supportive-housing pieces are where the spending sits.[^18] |
The lesson of the table is not in the individual figures, which will move once anyone scores them properly, but in the shape they make together. A carbon price and drug-price negotiation can plausibly carry a meaningful share of the family-support agenda, housing supply is cheap because it is mostly a matter of changing rules rather than writing checks, and the genuinely expensive commitments are health coverage and the cash credits. A platform that shows this arithmetic in public, including the parts it cannot fully fund, earns a kind of trust that no amount of message discipline can manufacture.
A federal floor and a state engine
There is a structural fact about this agenda that determines where it can be enacted, and most platform writing ignores it. A surprising share of the strongest planks are not federal levers at all.
Zoning preemption, by-right development, parking reform, and permit shot-clocks are state powers exercised over cities. Medicaid is administered by states. Hospital rate-setting of the kind Maryland pioneered is a state experiment.[^15] Election administration, policing standards, the school day where tutoring would live, and apprenticeship systems are all closer to a governor's desk than to a president's. The agenda keeps reaching instinctively for governors, the Whitmers and Newsoms and Polises, precisely because that is where its best ideas are executable. A liberalism of abundance and competence is, in practice, a governing program for a statehouse.
That suggests a cleaner way to organize the whole thing. Washington sets the floor and supplies the money. It funds the planning capacity that small jurisdictions lack, attaches housing conditions to infrastructure dollars, preserves the health subsidies, runs the national privacy and AI-governance rules that only a federal government can, and scores its own commitments honestly. The states run the engine. They preempt exclusionary zoning, administer the coverage expansions, pilot the payment models, set the policing standards, and build the apprenticeship pipelines. Reading the platform as a federal floor with a state engine resolves the altitude confusion that otherwise makes it feel like a presidential white paper trapped in a mayor's body. It also tells a candidate exactly which office to run for if they want to do most of it.
What survives, and what it asks of us
The agenda that survived this process is leaner than any party's wish list and sturdier than most of them. Build more homes by making them legal to build, mostly through the states, and protect tenants without pretending protection adds supply. Make public safety a matter of detectives, focused deterrence, and crisis capacity rather than posture. Reconnect people to the health coverage they already qualify for before promising to rebuild the whole system. Turn work and parenthood into stability through credits, child care, tutoring, and skilled trades. Govern technology through privacy and accountability instead of panic or deference. Build energy and grid capacity as the substrate everything else now runs on. And score all of it in public, including the parts that do not yet balance.
The parties will keep fighting over the purpose of the state, and they should. A well-defined public program is not the same as a broad guarantee, and tax restraint and redistribution do not share a moral center. Those fights are honest. What is not honest is the habit of hiding a policy's mechanism, its cost, and its owner behind a slogan, and then blaming the voters when the slogan fails to move them.
The best public argument does not open by announcing a theory. It opens with the thing in front of people. A downtown that empties out. A city that prices out the workers it depends on. A hospital that closes. A permit that takes longer than the construction. A family that loses coverage because a letter arrived at the wrong address. From there the work is to name the mechanism, the beneficiary, the guardrail, and the institution, and to do it in language a person can hear without first being told which side they are on. The platform, it turns out, is the easy part. Saying it plainly, and meaning it down to the last institutional detail, is the rest of the job.
Research Papers
I asked ChatGPT Pro to perform Deep Research into the ideal political platform. Then I asked it to figure out how to message the planks to anyone on the political spectrum. These are the results. Curious about the details of my approach and why it matters? Check out my field note on the subject.
