Definition #
[Design Friction] is the forces acting on an operator at the moment of commitment that keep him from building what his read supports.
You saw it. You did not build it. Something was acting on the decision while you were making it, and what opened instead is a smaller version of what you knew was there.
Not the shortfall in the build. That is the evidence these forces left behind, and it is where you go looking for them.
Child of [Friction]. The resistance is indirect, as it is everywhere in that family: it never reduces what the read makes possible, only how much of it gets committed to.
Mechanism #
The read is free and the build is not. Reading costs attention and standing. Building costs capital, a construction window, a menu you have to teach, a cast you have to retrain, and a public position you will be living with in front of everyone who knows you. The read produces a picture at no cost and the build makes you pay for it, which is why the two so rarely match.
Reversibility sets the price. The drag on any design decision is proportional to how hard it is to take back. Changing a garnish costs nothing. Changing a service model costs a season. Changing what the operation is costs everything you have put in it so far. Operators feel this correctly, and then let the feeling make the decision rather than pricing it and deciding.
So the build shrinks toward the reversible. Every element that could be walked back survives the design. Every element that could not gets deferred to next year. What opens is technically the design and structurally a different thing: the safe fraction of it, executed well, competing against operations that made the same subtraction.
Then ownership sets in and runs the other direction. Once it is built and has your name on it, reading the market honestly risks a conclusion you cannot afford, which is that the thing you built no longer fits what is available. That is not stubbornness. It is a rational refusal to run a read whose likely output is an expensive redesign. It is also why operations defend builds instead of redesigning them, and why a build keeps the shape it had on the day it was committed.
The unfelt half is what you think a restaurant is. Before any cost gets weighed, the design space has already been narrowed by what an operation of your type is understood to do. Category vocabulary arrives with the elements pre-selected, and options that were never considered do not present themselves as options declined. The design comes out looking reasonable because it matches what reasonable looks like in the category.
Two drags, two different repairs. The felt one is a cost you can name, price, and decide to pay. The unfelt one is a boundary on the design space that you cannot pay your way past because you do not know it is there. Working harder at the design does nothing to the second.
The result reads as a market problem. An operation built to the safe fraction of a good read performs like an operation built to a mediocre read. The numbers cannot tell the difference, so the operator concludes the read was wrong and reads more conservatively next time. The drag compounds into the next design cycle.
Both Roads carry it and only one can buy out of it. On Road 1 the design space is throughput, cost, and conversion, and most of it is genuinely purchasable — somebody else has built the layout, the system, the model. On Road 2 the design is the value you read and nobody else read, so no purchase exists. Buying a design for a Road 2 operation is buying the mean and installing it where the differentiation was supposed to be.
Load-Bearing Distinction #
Not the cut list it produces. The difference between what you read and what you built is the record these forces left. This term is the forces that were acting while each of those decisions was made. Stopping at the list produces an operator who knows what he did not build and not one thing about why, which is the whole of what is workable.
Not [Read Friction]. Read friction acts on the picture and holds it below what you could have seen. Design friction acts at the moment of commitment and holds the build below the picture you did see. They stack, and most operations carry both, but they are separated by a specific moment: the point where a decision has to be committed and paid for.
Not [Default Gravity]. Default gravity is an installed default overriding a ruling in execution. The design exists and gets overridden downstream. Design friction operates before that, on whether the ruling is ever made at full size.
Not a budget. Money is one of the costs and usually not the deciding one. Standing, reversibility, and what you would have to say to your cast and your Guests do more work than capital. Operators with the money still build the safe fraction.
Not caution. Caution is a decision made with the price named. Design friction is mostly a decision made with the price unnamed, which is what lets it feel like judgment rather than avoidance.
Not [Ordinary By Design]. Ordinary by design is a ruling: the element that does not need to be remarkable gets built plain on purpose. Design friction is a ruling that never got made, on an element that did need to be remarkable.
Not [Sameness Machine]. The sameness machine supplies a finished design so the work appears already done. Design friction is what happens to a design you did the work on yourself. They meet, badly: an operator carrying heavy design friction is exactly who the supplied design is easiest to sell.
Without this term, the shortfall between a good read and an ordinary operation has no name, and it gets blamed on the read. That sends the operator to work on the one part of the chain that was functioning.
Diagnostic Tests #
Test One — The Cut List. Write what you saw when you designed the operation, and what you opened with. The difference is the cut list, and it is evidence rather than the finding. Every item on it had something acting on it at the moment you decided, and the rest of these tests are how you name what that was. Most operators have never written the list down and can produce it in five minutes.
Test Two — The Reversibility Sort. Sort that cut list by how hard each item would have been to undo. If the cuts cluster at the hard-to-undo end, the design was not made on merit. It was made on reversibility.
Test Three — The Deferral Audit. Name what you have been going to do next year for more than two years. That is not a plan. It is a cut you have not admitted to.
Test Four — The Price Question. Take the largest deferred item and name what it actually costs: money, time, the retraining, the public position. Most operators find the item they have avoided for three years costs less than they have spent that whole time working around its absence.
Test Five — The Category Boundary. Name three things an operation like yours does not do. Then say why for each. If the answer is that it is not done, you have found a boundary on the design space rather than a decision you made.
Test Six — The Last Redesign. When did you last change something structural rather than fix, refresh, or add. If the answer is the build, the design has been frozen since the day it was committed and the market has not.
Test Seven — The Honest Read Test. Ask yourself what you would have to change if you read your market accurately today. If there is a part of the answer you would rather not know, you have located the protective drag directly.
Test Eight — The Road Test. Name whether the design element in question is throughput and cost, or read and value. The first can be bought. The second cannot, and buying it installs the mean exactly where the difference was supposed to be.
Family Position #
Sits inside Perspective, Operating Principles. Child of [Friction], positioned one stage downstream of [Read Friction] on the same chain: the read produces the picture, the design commits to it, and each stage has its own drag. Earns a separate entry because commitment cost and ownership protection are specific to building and do not appear in the read at all.
Perspective read. Design friction feeds back into the read and degrades it, which is the part operators miss. A read whose output keeps getting cut stops being run at full strength, because there is no point seeing what you already know you will not build. The read narrows to the buildable, quietly, and the operator experiences it as getting more realistic. That is the most expensive form this term takes, because it removes the one thing upstream of every other decision.
Product read. This is where the evidence shows. The Product is the direct output of the design, so the cut list is visible in the room to anyone who knows what was on it. A GX built to the safe fraction of a strong read delivers competently and distinguishes nothing, and it will be described by Guests in exactly the terms every other operation in the category is described in. The protective drag then locks it, because the Product carries the operator’s name and redesigning it means saying out loud that the build was wrong.
People read. The heaviest cuts are usually here, because People design is the least reversible thing in the operation. A new station, a new position, a real development structure, a different pay architecture — all of them are slow to install, expensive to unwind, and visible to everyone while they settle. So the design gets cut to what the current cast can already hold, which converts a design decision into a staffing accident. The cast then reads the ceiling accurately and the capable ones leave.
Performance read. The cut list arrives on the stage as chronic workaround. Every element that was designed out leaves the work it was supposed to carry somewhere else, usually on people improvising through a Friday. The operation runs at capacity every night with no obvious fault, and the actual cause is a structural element that was never built because it was hard to reverse.
Profit read. Design friction reads as prudence on the ledger and it is the most expensive line nobody has. Capital not committed shows as capital preserved; a build not made shows as a cost avoided. What does not show is the margin the unbuilt element would have carried, because no ledger records what was not designed. That asymmetry is why the conservative design keeps winning the internal argument while the operation stays in category range on every line.
Cross-References To Locked IP #
Parent:
- [Friction] — the family this term is one child of
Related:
- [Read Friction] — the stage upstream, holding the picture below ability
- [Value Congruence] — how well the built operation matches what was read, which this term degrades on the design side
- [Uncertainty Capacity] — what an operator needs to commit to a build he cannot fully see the end of
- [Question Dumb Shit] — the instrument that works directly on the category boundary
- [Five-Actor Test] — what validates a design element before it is committed
- [Ordinary By Design] — the deliberate ruling design friction is most often mistaken for
- [Meaningfully Differentiated Value] — the output the cut list removes
- [Constant Expiry] — why a design frozen at commitment stops fitting
- [Designed Architecture] — what a build carries when the design was committed at full size
Opposing patterns:
- [Default Architecture] — what fills the design space the operator did not commit to
- [Sameness Machine] — the supplied design that removes the commitment cost and the differentiation with it
- [Default Gravity] — the downstream override, after the design was made
- [Million Dollar Mediocrity] — a strong read built at the safe fraction, executed well
- [Static Decline] — the frozen build reading as stability while the market moves
Why This Matters #
An operator can do the hardest part of the work correctly and get nothing for it. He reads his market honestly, sees what is available, and designs to it. Then he builds two thirds of it, because the other third was expensive and hard to undo. What opens performs like an ordinary operation, and the numbers cannot tell him why.
So he does the natural thing and doubts the read. The next cycle he reads more conservatively, which produces a smaller design, which produces a smaller result, which confirms the doubt. Inside three cycles the reading has narrowed to what he already knows he will build, and the most valuable capability in the operation has been quietly switched off.
That is the reason this needs its own name rather than sitting inside the parent. The failure looks like a bad read and it is not. The read worked. The commitment did not. Those take entirely different work.
It also names what the safe build actually costs. Cutting the irreversible element feels like protecting the operation, and every ledger agrees, because capital not spent shows up and margin never earned does not. The subtraction never gets priced, so it keeps winning, and the operation stays inside category range for years with nobody able to point at the decision that put it there.
And it names who the supplied design gets sold to. An operator who has already cut his own design is carrying an unfilled space he knows is unfilled. The instrument that arrives claiming to fill it is not a temptation to a careless operator. It is a relief to a careful one.
Operating Consequence #
Write the cut list and keep it. What you read, what you built, what the difference was. The design is not finished until the subtraction is on paper, and it has to stay on paper, because an unrecorded cut becomes a thing you never thought of.
Price the irreversible element before deciding it. Money, time, retraining, the public position. Named, the decision is a decision. Unnamed, reversibility decides it and calls itself judgment.
Refuse reversibility as the sort order. Merit sorts the design. How hard something is to undo is an input to the price, never the ranking.
Stop deferring anything twice. A second deferral converts a plan into a cut. At that point either commit to it or strike it, but do not carry it a third year as intention.
Run the read at full strength regardless of the build. Read what is there, not what you expect to build. The moment the read starts pre-filtering for buildability, you have lost the upstream and everything downstream shrinks with it.
Separate an honest ordinary from a cut. Elements that do not need to be remarkable get built plain on purpose and named as such. Everything else that came out plain came out plain because you did not pay, and it belongs on the cut list, not in the ruling column.
Re-open the design on a schedule. Not fixes and not additions. Structural review of what the operation is, run against a current read, on a cadence you set rather than when something breaks.
Buy on Road 1 only. Throughput, cost, and conversion can be purchased. A design element carrying your differentiation cannot, and putting a purchased one there installs the mean in the position that was supposed to carry the difference.
What Changes Tomorrow #
Write two columns on one page. On the left, what you saw when you designed this operation — everything, including what you knew you were not going to do. On the right, what you actually built.
Circle everything on the left that is not on the right. That is your cut list, and it is the first time most operators have looked at it as one object rather than as a series of reasonable decisions made months apart.
Take the largest item on it and price it now, at current cost rather than the cost it carried when you cut it. What would it take to build. What would it take to undo if it failed. Who would have to be retrained and what would you have to say publicly about the change.
Then set that price against what its absence has cost you since you cut it. Count the workarounds the stage has been running, the position in the market you did not take, and the years the operation has been sitting in the same range.
Most operators find the number they have been avoiding is smaller than the accumulated cost of avoiding it, and that the reason it went uncut for so long is that the second number never appeared anywhere.
If it is worth building, commit to it with a date. If it is not, strike it from the list and stop carrying it as next year. What you are not allowed to do is leave it where it has been, which is unbuilt, unpriced, and still counted as part of the plan.
