Here's the question that ruins every city built from a drawing board: if you get everything right on paper, why does the place feel dead by the time the first family moves in?
That's the one. And I think we have a case study sitting about halfway between Tel Aviv and Jerusalem that answers it.
Hannah sent us this one. She's the one with the long-running dream of building a mid-sized walkable Israeli city, and she wants to get specific about the document underneath it. She's asking what a form-based code for that city would actually have to say. Not the vision statement, the code.
The boring document.
The document that decides whether the city looks like a city or like a rendering somebody forgot to turn off. She wants to develop a new Levantine architectural language, picking up the thread from before modernism without copying a historical style. Traditional proportions, local materials, local climate, the whole span of building history in this land.
That's the hard part right there. Picking up a thread is not the same as photocopying what was on it.
And then the second half, which is the real trap. She doesn't want the city to feel like one architect drew every building. Different architects, individual buildings, and somehow still coherent. Organic. Like it grew.
Like it evolved.
And the concrete questions follow from that. How specific do you get about height, cornice lines, street walls, proportions, window openings, materials, ornament, roofs, arcades, balconies, courtyards, and the space between the building and the street. Whether you can require real structural stone instead of concrete with a decorative hat on. And how you write enough rules that any competent architect produces something beautiful, without writing so many that the city turns into pastiche.
Four hundred years from now is the test she sets. Would people still want to live there.
So, Herman. Is there a tool that does that, or is this a fantasy?
There's a tool. It's called a form-based code, and the first thing to understand about it is what it regulates. It doesn't regulate what happens inside a building. It regulates the physical character. How buildings shape streets and public space. That's the whole inversion.
So not "is this a shop or an apartment," but "how does this building stand on the street."
Conventional zoning does the opposite. It separates uses, it pushes density down, it bakes in car dependency, and it applies its standards generically. The Chicago Metropolitan Agency for Planning guide calls it a one-size-fits-all approach. The same setback, the same parking ratio, whether you're on a boulevard or a cul-de-sac.
And a form-based code throws that out.
It replaces it with a regulating plan. A map that's much finer-grained than a zoning map. Instead of "residential zone R-two," you have a street-by-street instruction about what kind of frontage is expected. And the frontage types are where it gets interesting. Porch and Fence. Stoop. Shopfront. Forecourt. Arcade. Those are the five you'll see over and over.
The stoop. That little stair that sits between the sidewalk and the front door.
That stoop is doing real work. It's a transition zone. It tells you where public ends and private begins, and it gives the person standing on it somewhere to be. That's the level of detail a form-based code operates at.
Give me the boulevard version of that. What does a shopfront frontage actually require?
A shopfront frontage means the building comes right up to the sidewalk line. No setback. The ground floor is glazed, the entrance is on the street, and there's usually a minimum percentage of transparency so the street feels watched. You're not allowed to put a blank wall or a parking lot between the door and the pedestrian. The building has to participate in the life of the sidewalk.
And the arcade?
The arcade is the colonnade. The building line steps back, columns hold up the upper floors, and the pedestrian walks underneath in shade. In a hot climate that's not decoration, that's infrastructure. You're literally building a covered walkway into the private building and making it public.
So the frontage type is the code's way of saying what kind of room the street is.
Right. A street lined with stoops is a different room than a street lined with shopfronts, which is a different room than a street lined with forecourts. The code picks the room, street by street.
How old is this idea?
The first modern one was Seaside, Florida. Nineteen eighty-six. Duany Plater-Zyberk. And here's the part I love. They presented the whole urban code on a single poster.
One poster.
That's the entire legal instrument that produced the town. It's been expanded into what's called the SmartCode, which is the template most places start from, and the largest implementation is Miami's citywide code.
Miami being the largest is funny, given what that city looks like.
The code doesn't design the buildings. That's the thing. It sets the frame. What goes inside the frame is up to whoever's drawing.
So if I'm an architect in a form-based code district, what am I actually free to do?
Almost everything about the building itself. You pick the style. You pick the material palette, within whatever the code says about durability. You pick the roof form, unless the code regulates it. You pick the ornament, the window details, the color. What you don't pick is where the building sits relative to the street, how tall it can be, how much of the frontage it has to occupy, and how the ground floor meets the sidewalk. Those are fixed.
So the code fixes the relationship, and the architect owns the object.
That's exactly the division of labor. And it's why you can have a hundred different architects working on one street and still get a coherent street.
Okay, so here's my question, and I think it's Hannah's question. If a form-based code regulates all of this, it's a tool that produces a city with a shape. But does it have anything to say about how the buildings look?
Almost nothing.
Say that again.
It's the thing most people get wrong. Most form-based codes have little or nothing to say about architectural style. Architectural standards are an optional element. You can write a form-based code and never once mention a material, a roof pitch, or an ornament.
So the tool that Hannah wants to use for the look of the city doesn't actually do the look of the city.
Not by default. What they do regulate is articulation, window proportions, the rhythm of openings. They'll ban blank walls. They'll say something about signage. That's about it.
So it stops at the skin.
It stops at the shape of the skin. Not the skin.
Then the premise of the whole prompt is a problem. She's asking how to write a form-based code that shapes architecture, and the standard toolkit says don't.
Right. And that's why this episode has to go somewhere the standard literature doesn't. The answer isn't going to be a bigger code. It's going to be two documents instead of one. But we should get to Modi'in first, because Modi'in is what happens when you want the thing and reach for the wrong tool.
That's the case study you mentioned.
Moshe Safdie's master plan, nineteen eighty-nine. Construction from ninety-six. Eight thousand acres on a hilly site between Tel Aviv and Jerusalem. And Safdie was not a man who wanted a generic city. His own stated principles are good, Corn. Embrace the topography. Organize development on strong spinal axes. Reduce the impact of the car. Generally limit buildings to four storeys. Around fifty persons per acre. And require that every unit has a garden, a roof terrace, or a balcony.
Every unit gets outdoor space.
Every unit. And cladding guidelines applied to public, commercial, and high-rise buildings. So there's already an architectural layer in there. He wasn't a naïf about this.
So what did he say he was trying to avoid?
In twenty fourteen he said it directly. "A normal city should not look like a collection of elements that repeat themselves. It should look more like old Tel Aviv — Rothschild Boulevard — not like all of the familiar Ministry of Housing projects."
Rothschild Boulevard. He's naming the boulevard.
He's naming it because it works. It's got variety, it's got shade, it's got a scale that a person can hold in their head.
What makes Rothschild work, specifically? Because I want to be precise about what he's pointing at.
It's a boulevard with a central tree-lined median. The buildings on both sides are three to four storeys, so the sun reaches the street at the right angle for most of the day. The ground floors are commercial, so there's life at eye level. And the buildings are all different — different architects, different periods, different styles — but they hold the same building line, roughly the same height, and the same relationship to the sidewalk. So the street has a consistent frame and an inconsistent content.
Consistent frame, inconsistent content.
That's the formula. And it's exactly the formula Hannah is asking for, just with a Levantine vocabulary instead of an eclectic Tel Aviv one.
So a man stands up with a clear idea of what he does not want, writes it down, builds a city over thirty years, and what do the residents say?
A woman named Ruti Weisberg, in the Globes piece on the city's twentieth anniversary. She says it's a very well planned city. Then she says, "too planned." Her words. "There are no places that are late additions, or different areas to wander around, like in a big city... it is missing the spontaneity that is a part of every city, it lacks personality."
No late additions.
And then my favorite line in the whole piece. She jokes with friends that they want to go to the big city to see some dirt.
To see some dirt.
Because Modi'in has none. Everything is swept.
You know what she's describing. She's describing a place with no accidents in it.
That's exactly what she's describing.
Every corner was decided by somebody at a drafting table in a specific year. Nothing was left to the people who came after.
And that's the paradox, right? Safdie was explicitly anti-repetition. He designed against the Ministry of Housing projects. And the outcome is a city people describe as missing personality.
So he lost the argument with himself somewhere between the principle and the code.
Between the principle and the instrument. He reached for conventional setback, footprint, and height rules with cladding guidelines on top. That's what the code actually said. Massing and cladding. It had nothing to say about the relationship between a building and the street. Nothing about the rhythm of openings. Nothing about the architectural language that makes a street feel coherent rather than merely coordinated.
The cladding is the surface. The relationship is the city.
The cladding is the surface. Correct. And it's not just Modi'in. This is a national critique. Naama Melis, an architect, on Israeli construction: there's no dialogue between building and street. She uses a striking image. She says the street is treated as a theater stage and the buildings are its backdrop.
Is that not how it should work?
It's the opposite of how it should work. The buildings are supposed to make the street. The street is supposed to be the room, and the buildings are its walls. If you treat the buildings as scenery, you've given up on the room.
That's a good distinction. The room and the walls.
And Safdie himself, on the towers. Twenty twenty-one. "These glass buildings, which look like the ones in Dubai, have nothing to do with our climate, our culture, or the way people live." Then, "Stuffing people into glass buildings under the blazing hot sun, is farcical."
He's the same man, Herman. Same man, thirty years later, still arguing the same direction, and losing both times.
He's been consistent. The instrument keeps not being there.
There's a demolition angle here too, isn't there? Somebody was saying the heritage gets knocked down.
Dr. Zvi Elhayani describes what he calls a pathological frenzy of demolition and destruction of the architectural heritage, driven by market forces. So on one side you've got a city that feels too planned, and on the other you've got the country tearing down the older stock that actually had the qualities we're talking about.
Both directions, then. You can fail by over-planning the new, or by erasing the old.
And the part that should make Hannah's ears prick up. The tool that could have prevented Modi'in's problem is exactly the tool she wants to write. Safdie had conventional zoning plus cladding guidelines. A form-based code is built to regulate the relationship between buildings and the street. But it would have to go further than standard practice into architectural standards. Because most codes stop at the skin.
So the answer to Hannah's question is, yes, the tool exists, but you have to use it in a way nobody has used it.
We'd be writing a form-based code with the architectural layer fully switched on. Which almost nobody does.
Well, that's the interesting version of the episode. Let's do the thing.
So the first move is to figure out what the tradition actually is, because you can't continue a thread you haven't read.
And this is where the pastiche danger lives, right? Because "Levantine architectural language" can mean twelve different things depending on whether you're talking about a courtyard house in Damascus or a red-roofed house in Irbid.
There's a study from twenty sixteen on the traditional Arab house in the Eastern Mediterranean. Nine houses across three Israeli climatic zones, ranging from the eighteenth century into the early twentieth. And the finding is that climate was designed in from the start. The house moderates the impact of the outside temperature — inside the house in winter, and mainly during the hot hours of the day in summer.
So the house was air conditioning before air conditioning was possible.
And the elements that did that work are still relevant to a Mediterranean climate. That's the study's conclusion. Not quaint. Relevant.
What are the elements?
Courtyards. Thermal mass. Passive ventilation. Water features. The Damascene house study describes the geometry. It's inward-oriented. The outer walls are nearly solid. The facades open onto the inner courtyard.
So the house is a donut with a hole in the middle.
And that hole is the weather system. Cool air comes down into it at night, the mass of the walls holds the cool, and through the day it releases into the rooms. That's the mechanism.
Walk me through a day in one of these houses. Because I want the mechanism to be concrete.
Okay. It's July. The courtyard is open to the sky. At night the sky is a heat sink — the courtyard radiates its heat upward and the cool night air settles into the courtyard and into the rooms around it through the open doors and windows. The stone walls absorb that cool. By morning the courtyard is cool and the walls are cool. Then the sun comes up and the outside air heats fast. But the walls are thick stone, so the heat takes hours to penetrate. The courtyard stays shaded because the walls are tall and the courtyard is narrow. Around mid-afternoon the outside air is at its worst, but the interior is still riding on the cool it stored overnight. By evening the walls have finally warmed through, but the sun is down and the courtyard starts radiating again. The building is running a twenty-four hour thermal cycle, and it never needs a machine.
And the water feature?
The fountain in the middle of the courtyard. Evaporation. Moving air across water drops the temperature and raises the humidity, which in a dry climate is exactly what you want. It's a passive cooler that also happens to be beautiful.
And from the street?
From the street you see almost nothing. Bare wall, one door. The study says it well. "From the outside, the houses appear similar, giving the impression of unity, without showcasing any disparity between the rich and poor of society."
So a rich man's house and a poor man's house look the same from the street.
From the street. On the inside, completely different. That's a deliberate social architecture. You don't announce your wealth to the road.
There's something in that for the coherence problem. If everyone builds to the same street wall, the street looks unified whether you're rich or poor.
You've just stated the thesis of a very good form-based code, Corn.
What about materials? You mentioned Irbid.
Irbid is the same tradition, different country. The houses are one or two storeys. Built of stone, and the wall is more expansive than the openings. The windows and doors are surrounded by ornament. The roofs are pitched and covered with red tiles. Stone-bearing walls, groined vaults, walls with cornices as ornaments, oak and bamboo ceilings, clay-plastered interior surfaces for thermal protection.
Clay on the inside.
Clay on the inside for thermal protection. The plaster wasn't decoration. It was a phase-change material before we had the word for it.
So the courtyard house isn't just a pretty thing. It's a machine for managing heat.
It is a machine.
And that's the point I want to nail down. Because I think this is the crux of the whole episode. If the traditional house is a machine, then the parts of it you want to continue are the working parts, not the decorative ones.
And the literature says the same thing, in different words. There's a line from the twenty twenty-three paper on the Damascene house that I want to read properly, because it names the trap. "When talking about the relationship between architecture and identity, what comes to the minds of many is adherence or non-adherence to the traditional and historical architecture of the region. They measure the identity of architecture by the degree of its similarity to previous architecture in its familiar forms."
So identity gets measured by resemblance.
By resemblance. Which means the more your building looks like the old one, the more "authentic" it's judged to be. And the paper rejects that. The alternative is to extract principles, not forms. Climate response. Human scale. Privacy. Courtyard logic.
Which means the code should tell you how to behave, not how to look.
Yes. The code can require a courtyard with a minimum dimension and a minimum planting area, without ever saying what shape the arches are.
Because a courtyard that does its job is a courtyard that does its job, whether the arcade around it has a pointed arch or a square one.
And that's the escape from pastiche. You mandate the performance, not the motif.
Now let me push on this, because I think you're being a little generous. There's a well-documented failure mode for exactly this kind of rule. The CMAP guide describes it. New development that follows the letter of the law but only superficially fits with the existing context.
The paste decorative shutters on the facade example.
That's the one I'm thinking of. An outsized modern building, and the developer agrees to paste decorative shutters on it. The letter of the law is satisfied. The building still looks like a box wearing a costume.
That's the concrete-with-a-decorative-facade problem Hannah named, in a US form. And the CMAP guide is honest that a weak architectural standard produces exactly this. The rule existed, and the outcome is still bad.
So how do you write a rule that can't be satisfied by pasting?
You have to regulate something structural. The proportions. There's a code in Winooski, Vermont, that does something I find clever. It doesn't say what style adjacent buildings should be. It says adjacent buildings have to differ from each other by at least twenty percent in fenestration proportions. Height, width, or the height-to-width ratio of the windows.
So the windows can't all be the same shape.
The windows cannot all be the same shape. And it goes further. It requires variation in the bay rhythm — the pattern of how the facade is divided. ABA versus ABB versus BACB. And it requires a change in wall material. It explicitly says color changes are insufficient.
Are insufficient.
Insufficient. So you can't satisfy the code by painting the place a different shade of beige. The developer has to actually change something real.
That's the anti-pastiche instrument right there. It forces difference without specifying what the difference should look like.
It's a variance rule. It guarantees that the block doesn't become a wall of identical window grids, without taking the decision away from the architect.
And there's a similar logic in Kansas City, Kansas. Maximum building height is limited to one storey higher than the lowest adjacent primary structure.
So you can't build a tower next to a two-storey house.
You can't. You can build three storeys next to a two-storey, but not twelve. The building has to acknowledge its neighbor.
And there's a third one that answers a different problem. Bothell, Washington requires a well-formed base and a well-formed top.
Define those.
The base defines the pedestrian-scale street room. The shopfront level, the human height. And the top — a cornice, a cap — defines the skyline. The building has to have a beginning and an end, at the right heights.
So the code is saying: your building has a floor zone and a roof zone, and each does a job.
Both of them are about human perception. The base is what you see when you walk past. The top is what you see when you look up the street.
That's a rule that produces beauty without ever naming beauty. It just names the eye.
So let me try to write Hannah's code out loud, and see how bad it is. Starting with the structural questions. Height. You could borrow the Kansas City rule. Limited to one storey higher than the lowest adjacent primary structure. That's the general rule, with taller allowances on the spinal boulevards.
Cornice lines. Bothell wants a required top.
Required base and required top, and a continuous cornice line along any block where the buildings are within one storey of each other. So the street gets a horizon.
Street walls. What's the rule?
A minimum frontage occupancy. Say seventy percent of the lot width has to have a building on the front setback line. That's the rule that prevents the parking-lot-in-front problem. It forces the building to hold the street.
Proportions. The thing you keep coming back to.
Window openings between one-to-one and one-to-two point five in height-to-width ratio. That's the traditional range. Not the strip window, not the square picture window. And a minimum solid-to-void ratio on the street facade, say sixty-forty solid, so the wall dominates the openings, the way the Irbid houses do.
Materials. Can you require stone?
You can require a masonry wall of a specified minimum thickness. You can require that the structural wall be the visible wall. That's the anti-hat rule. You cannot satisfy it with a concrete frame and a stone veneer, because the veneer isn't structural. The code can say the load-bearing element has to be the finish element.
Does that survive legal challenge? Because developers will fight that.
It survives if you can tie it to a legitimate public purpose. And you can. Thermal mass, durability, the character of the district. Courts have upheld material requirements when they're tied to something the municipality is allowed to care about. It's harder in some jurisdictions than others, and it's harder if you're vague. If you say "stone," you have to define stone. If you say "load-bearing masonry of at least thirty centimeters," that's a measurable standard and it's much more defensible.
Ornament. Can you require it?
You can require that openings be surrounded by a surround. A frame. Something that marks the edge of the opening. That's an old rule and it's an easy one to justify, because it's about legibility, not style. You can't require a specific motif, but you can require that the opening has a defined edge.
Roofs.
You can cap the roof. You can require a pitched roof above a certain depth, or a parapet with a defined cornice. What you can't do is require a specific tile from a specific region, because that's a product requirement and it gets shaky. But you can require a roof form and a roof color range.
Arcades and balconies.
Arcades where the frontage type calls for them, with a minimum column spacing and a minimum clear height. Balconies above the ground floor, minimum depth, maximum percentage of the facade, so they don't turn into a wall of glass.
Courtyards.
A minimum courtyard dimension, a minimum planted area, and a requirement that at least one habitable room opens onto it. That's the performance rule. It makes the courtyard do its job without saying what it looks like.
The space between the building and the street.
That's the frontage type. It's already covered. The code says which of the five frontages applies on which street, and the frontage type determines the setback, the ground floor treatment, and the transition zone.
We've got a code. It's specific. It's structural. It doesn't name a style. And here's the question I actually care about. Does it produce the thing Hannah wants, which is a city that looks like it grew?
Here's where I have to be honest. No.
Say more.
The code we just wrote is a set of constraints. Constraints produce consistency. They don't produce growth. Growth is what happens when a city is built over time by many hands, each responding to what came before. A code applied all at once to a greenfield site produces a consistent city, not a grown one. Modi'in is the proof. Safdie had constraints. He got consistency. He didn't get growth.
The code is necessary but not sufficient.
The code is the frame. The growth comes from time and from the number of decisions. If you want a city that looks like it grew, you have to let it grow. You have to build it in phases, over decades, with different architects and different developers, and you have to leave room for the late additions. The code has to be permissive enough that someone can add something in year thirty that nobody anticipated in year one.
That's the piece Ruti Weisberg was pointing at. The late additions.
That's the thing a code can't mandate. You can't require spontaneity. You can only fail to prevent it.
The two-document answer. You said the answer isn't a bigger code, it's two documents.
Two documents. The first is the form-based code, and it does what we just described. It regulates the frame. The relationship between buildings and streets. The frontage types, the heights, the street walls, the base and top, the proportions, the material requirements, the courtyard performance rules. That's the legal instrument. It's enforceable. It's specific. It's the thing that makes the city cohere.
And the second document?
The second document is not a code. It's a pattern book. A design guide. It's advisory, not mandatory. It shows the architectural language. It shows how a window surround is made in this place, how a balcony is detailed, how a roof edge is finished, how a courtyard is planted. It's a catalogue of the working parts of the tradition, translated into contemporary construction.
The code says "you must have a courtyard of at least this size." The pattern book says "here's how a courtyard in this climate is actually built."
The code is the law. The pattern book is the craft. And the reason you split them is that the code has to be enforceable and therefore has to be narrow, and the pattern book has to be rich and therefore can't be enforceable. If you put the pattern book into the code, you get pastiche, because you've made the motif mandatory. If you leave it out entirely, you get Modi'in, because the architects have no shared language.
The code is the grammar. The pattern book is the vocabulary.
The code is the grammar. The pattern book is the vocabulary. And the architects write the sentences.
I like that. Let me push on the pattern book, though, because I think there's a trap there too. If the pattern book shows a specific window surround, aren't you back to copying forms?
You are, if the pattern book is prescriptive. The trick is that the pattern book has to show the logic, not just the object. It has to show why the surround exists — because the wall is thick, because the opening needs a shadow line, because the rain has to be thrown clear of the sill. If it shows the logic, an architect can invent a new surround that does the same job. If it only shows the object, the architect copies the object.
The pattern book teaches principles through examples.
Principles through examples. And it should show multiple examples from different periods, so it's clear that the tradition is a range, not a single point. The Damascene courtyard house and the Irbid stone house and the Tel Aviv boulevard building are all in the same family. The pattern book shows the family, not the individual.
The code keeps the family coherent on the street.
The code keeps the family coherent on the street. The pattern book keeps the family alive in the architect's head.
Let me try to land this for Hannah. She asked how specific to get about height, cornices, street walls, proportions, openings, materials, ornament, roofs, arcades, balconies, courtyards, and the space between the building and the street. And the answer is: specific about all of them, but specific about the structural and the performative, and silent about the decorative.
Specific about the frame, silent about the object. And the object is where the architects live.
The two-document structure is how you get both. The code is the frame. The pattern book is the language. And the late additions are what the code has to leave room for.
The four-hundred-year test. Would people still want to live there. The answer is: if the code makes a good frame, and the pattern book keeps the language alive, and the city is allowed to grow, then yes. Because the thing that makes a city worth living in for four hundred years is not that it was finished. It's that it was never finished.
It's that it kept being added to.
It kept being added to. By people who were responding to what was there. And the code's job is to make sure that when they add, they add in a way that holds the street. And the pattern book's job is to make sure that when they add, they add in a language that belongs.
Modi'in's lesson is that you can get the frame right and still lose the language. And you can get the language right and still lose the growth.
You can lose it either way. The two documents are how you don't.
Herman, last question. If Hannah builds this city, and it works, what does it look like in a hundred years?
It looks like a place where the buildings are all different and the street is all one thing. Where the stone is real stone and the courtyards are real courtyards and the shade is real shade. Where somebody in year eighty added a building that nobody in year one could have imagined, and it fits, because the code told them where to stand and the pattern book told them how to speak.
And the dirt?
There's some dirt. Because a city that's been lived in for a hundred years has dirt. That's not a failure. That's the evidence.
Thanks, Herman.
Thanks, Corn.
