The Hidden Cost of Ceremony: Why Less Process Often Means More Power

Hatched by john ke

Jun 22, 2026

9 min read

46 views

84%

The lie we keep paying for

What if the thing making your system better, safer, or more polished is also the thing quietly draining it of performance, money, and clarity?

That question shows up in places that look unrelated at first. In the gym, people cling to deadlifts as if the exercise itself is sacred, even when simpler movements can train the same muscles with less fatigue. In AI systems, people accept polite, expansive responses because they sound intelligent, even when that extra wording is just billing fuel. In both cases, the real issue is not the exercise or the phrasing. It is the assumption that more ceremony equals more value.

That assumption is almost always wrong.

We are trained to admire effortful-looking processes. Big lifts. Full explanations. Friendly warmups. Thorough preambles. Detailed tool announcements. The problem is that ceremony often masquerades as substance. It creates the feeling of quality while quietly imposing costs: fatigue in training, token burn in AI, cognitive overload in communication, and eventually a system that is harder to recover, harder to scale, and no more effective than a leaner alternative.

The deeper question is not whether deadlifts are good or whether politeness is nice. The deeper question is: when does a prestigious default survive only because people confuse tradition, completeness, and style with actual usefulness?

The prestige trap: when the hard thing becomes the obvious thing

Deadlifts have a special place in fitness culture because they feel like a root exercise. They look serious. They sound serious. They produce a sense of legitimacy. If you lift, you deadlift. That is the myth.

But myths often survive because they organize identity, not because they optimize outcomes. For many people, especially those chasing lower back health, hamstring development, and sustainable recovery, a well chosen alternative can deliver the main benefits without the main costs. A 45 degree back extension, done with intent and progression, can train the posterior chain while reducing the systemic price tag of repeated heavy pulls.

That same prestige trap appears in language models. A model that says, “Great question!” and then gives you a careful explanation feels helpful. It feels intelligent. It feels human. But if you are running thousands of calls a day, that friendliness is not free. It is a recurring line item. Every filler opener, every reassurance, every “Let me think through this,” is a tiny tax on output tokens. Multiply that by scale, and politeness stops being personality and starts becoming overhead.

This is the same pattern in two domains:

A high status default persists because it signals seriousness.
The default becomes normalized because people confuse signal with function.
The system absorbs hidden cost until someone asks a better question: what are we actually trying to preserve?

If the goal is lower back resilience, the prestige of deadlifting is not the point. If the goal is accurate code, the charm of a warm conversational wrapper is not the point. The point is the output, not the ritual around it.

The moment you stop asking, “What looks like the real thing?” and start asking, “What actually produces the result?” you begin escaping ceremony.

Brevity is not emptiness, it is compression

There is a lazy way to think about minimalism: as if shorter means cruder, and cruder means worse. That is how people dismiss terse workout programming, terse writing, and terse model responses. But compression is not the absence of content. It is the removal of redundant structure so the useful signal can move faster.

A good back extension is not a lesser deadlift because it omits the barbell theater. It is a different compression of the same training intent. It targets the posterior chain, spares some of the recovery burden, and often makes progression more sustainable. The exercise is leaner, not weaker.

Caveman style prompting works on the same principle. If a model is told to stop warming up, stop narrating tool use, stop restating the question, and stop signing off like a customer service representative, it does not lose intelligence. It loses packaging. The task remains. The fluff drops away.

This distinction matters because many people misunderstand what tokens are. They are not just words. They are part of the operating cost of thought. In a multi turn session, verbose responses are not only expensive once. They come back as context, which means every extra sentence echoes into the next turn as future input. That is why a little ceremony can become a lot of drag.

Think of it like carrying extra weight through the whole workout because you liked how the plates looked in the first set.

Compression has three advantages:

Lower direct cost: fewer output tokens, lower bill.
Lower recursive cost: less chatty history reenters context on later turns.
Lower coordination cost: fewer words means fewer opportunities to confuse, distract, or bury the answer.

The key insight is that brevity is not the same as austerity. Brevity is selective retention. It keeps the part that moves the system forward and discards the part that mainly flatters our sense that something substantial is happening.

The real enemy is not verbosity. It is unpriced overhead

People often frame these debates as style preferences. Do you like a friendly assistant or a blunt one? Do you prefer deadlifts or extensions? But style is not the whole story. The more useful frame is overhead versus payload.

Payload is the part of the system that directly accomplishes the goal. Overhead is everything required to make the payload feel complete, polished, or socially acceptable. Some overhead is necessary. Too much, and the system slows down, costs more, and becomes harder to maintain.

This is true in lifting. A deadlift carries overhead in the form of fatigue, recovery demand, technique complexity, and injury risk for some lifters. A back extension may deliver less maximal prestige but more favorable cost to benefit ratio for the specific goal.

It is true in AI. “Of course, I’d be happy to help” is overhead. So is narrating every tool call, restating the user’s request, or closing with a polite offer for more help. If the model is embedded in a pipeline, those words are not harmless. They are an operational decision, whether or not anyone intended them to be.

It is also true in writing itself. Most bad prose is not bad because the author lacks ideas. It is bad because the author pays too much tribute to the appearance of thought. The paragraphs circle the point, explain the explanation, and keep the reader hostage while the writer performs seriousness.

A helpful mental model here is the ceremony budget:

Every system has a limited budget for rituals that do not directly advance the task.
Spend too much of that budget on style, and you starve the task.
The best systems are not style free. They are style disciplined.

This is why the “just be nice” instinct can be costly in technical environments. Politeness is not inherently bad. Unexamined politeness becomes a default tax. In human conversation, that tax is often worth paying. In a machine pipeline, it may not be.

The same applies to training culture. If a movement pattern is chosen because it symbolizes toughness rather than because it best serves the athlete’s current objective, the athlete is paying a ceremony tax in recovery, time, and soreness.

A better rule: optimize for the next bottleneck, not the first tradition

The mistake behind both deadlift worship and verbose AI responses is the same: people optimize the thing that looks foundational instead of the thing that is actually constraining performance.

In fitness, the obvious bottleneck may not be posterior chain strength. It may be recovery, lower back tolerance, adherence, or consistency. A movement that leaves you too drained to train well next week is not helping, even if it is “the king of exercises.” The best stimulus is the one your body can repeat productively.

In model systems, the obvious bottleneck may not be raw response quality. It may be cost per call, latency, context growth, or tool orchestration. If a model spends half its output being charming, the system may still function, but it is functioning inefficiently. The next bottleneck is not intelligence. It is throughput.

That suggests a broader principle:

Do not preserve the tradition that solves yesterday’s problem if it creates today’s bottleneck.

This principle can be applied with a simple diagnostic sequence:

Name the real objective. Muscle growth, code fixes, lower latency, lower cost, better adherence, clearer writing.
Identify the hidden costs. Fatigue, token burn, context bloat, confusion, recovery debt.
Separate signal from ritual. Ask what part of the practice actually produces the result.
Replace prestige with proportionality. Use the smallest effective method.
Test at scale, not vibes. A method that feels better may not perform better over time.

That last step matters. The most seductive ceremonies are the ones that feel sophisticated in the moment. It is easy to admire a long, polished response. It is easy to respect the deadlift because it looks hard. Scale is what reveals the truth. Once you repeat the action dozens or thousands of times, overhead becomes visible.

Key Takeaways

Ask what the ritual is buying you. If the answer is mostly identity, polish, or tradition, it may not be worth the cost.
Separate payload from overhead. The useful part of a practice is often much smaller than the part that surrounds it.
Optimize for repeatability. The best choice is often the one you can sustain for many iterations without accumulating fatigue or cost.
Treat brevity as compression, not loss. Removing fluff can preserve meaning while improving speed, cost, and clarity.
Look for the next bottleneck. Stop defending the old default when it is no longer solving the problem that matters.

The new status symbol is ruthless usefulness

There is a quiet cultural shift hiding inside these examples. For a long time, prestige belonged to the elaborate thing: the big lift, the polished answer, the full explanation, the gesture of effort. But as systems get more complex and more scalable, the new luxury is not excess. It is precision.

The strongest routine is not the one with the most heroic exercise. The best assistant is not the one that sounds the most human. The most intelligent writing is not the one that uses the most words. In all three cases, the real achievement is achieving the result with the least necessary drag.

That is why “caveman” style prompting and exercise substitution are not gimmicks. They are examples of a larger discipline: stripping away what does not carry its weight. Not because details are bad, but because details must justify themselves.

The world is full of systems that mistake elaboration for excellence. The opportunity, whether you are training a body, building an agent, or writing an answer, is to ask a more demanding question:

What happens if we remove everything that is merely performing competence and keep only what creates it?

Often, the result is not a weaker system. It is a more honest one.

And honesty, in training or in language or in software, is usually the first step toward power.