ROMAN Applied Sciences. Nodarama
Roman Applied Sciences diagram connecting systems, expression, and trust.
Innovation 01

RANGES

Roman Alpha Numeric Glyph Encoding System.

RANGES is a compact, shape-preserving naming system for code built around a simple idea: AI does not always need more language. Often, it needs clearer structure.

The hidden cost of descriptive code

Modern frontend code often accumulates names such as main-navigation-container-wrapper-inner. Each name may sound reasonable in isolation, yet repeated across HTML, CSS, scripts, prompts, diffs, and AI responses, descriptive naming can become a second layer of code that the system must continually interpret.

That extra language has a cost. It increases character count, consumes context, creates more opportunities for inconsistent naming, and makes targeted changes harder when several identifiers sound almost—but not quite—the same.

RANGES replaces that naming bloat with one brief semantic name and a compact relational glyph. The goal is not to remove readability. The goal is to preserve the shape of the system while removing unnecessary linguistic noise.

A coordinate system for code

RANGES stands for Roman Alpha Numeric Glyph Encoding System. Its glyphs can describe nesting depth, relationship, object type, and instance position.

<body class="Body-Ax">
  <main class="Site-Bx">
    <header class="Head-Cx">
      <div class="Title-Dx">
        <img class="Logo-Eo">
      </div>
    </header>
    <section class="Hero-Cx">
      <h1 class="Title Do"></h1>
      <img class="Art-Do">
    </section>
  </main>
</body>

The capital letter indicates depth. Elements sharing the same letter are structural peers. A lower-case suffix can indicate whether something is a container or an object: x for a container and o for an object. Numbers can distinguish repeated instances such as Button Eo1, Button Eo2, and Button Eo3.

The notation may use a space or a dash. Teams may encode position, branch, family, or sequence differently. The important rule is not rigid conformity; it is local consistency.

Shape preservation improves scannability

Code is not only language. It is also shape: hierarchy, containment, sequence, siblings, parents, children, and terminal objects. RANGES makes those relationships visible in the identifiers themselves.

For a person, this makes indentation levels and component boundaries easier to recognize at a glance. For an AI system, it provides compact structural coordinates that can help it follow where an element sits without repeatedly inferring the hierarchy from verbose names.

A short identifier can still carry rich information when it preserves position, relationship, and type.

Less naming noise may reduce AI drift

AI models are linguistic systems. When a project contains many long, semantically similar identifiers, the model must distinguish between names that differ only by a few words or suffixes. That creates room for substitutions, invented variants, and broad rewrites when a narrow edit would have been enough.

RANGES reduces the number of competing words while preserving a stable location signal. That can make instructions more direct: change Title Do, insert after Hero Cx, or replace Button Eo2. The system becomes easier to target precisely, whether the change is made by a human, an AI, or a deterministic recorder.

This does not eliminate hallucinations by itself. It reduces one source of ambiguity: verbose, overlapping naming.

Efficiency compounds across the workflow

Shorter repeated identifiers reduce the amount of text carried through source files, prompts, responses, diffs, and logs. The exact savings depend on the tokenizer, language, project shape, and repetition pattern, but the effect compounds because the same names appear many times.

  • More code can fit inside a fixed context window.
  • Smaller prompts can reduce processing work and response time.
  • Usage-based plans may cost less when fewer tokens are sent.
  • Lower token volume can reduce compute demand and associated energy use.
  • Smaller files are faster to transmit, index, compare, and search.

These are potential outcomes, not fixed guarantees. They should be measured against real projects. The useful principle is simple: every unnecessary repeated token is work performed again.

Precision surgery instead of broad rewrites

Compact, stable identifiers make manual search easier and support exact-locator workflows. A developer can find a structural node quickly, and an execution system can target the same text without guessing.

This changes the relationship with AI coding. Instead of asking a model to regenerate a large file to repair one detail, the user can request a narrow replacement against a recognizable coordinate. That makes changes easier to review, easier to diff, and easier to reverse.

A guideline, not another language

RANGES is intentionally lightweight. It can be explained in minutes, introduced gradually, and adapted to the structure of an existing project.

  • Use one brief semantic name.
  • Add a relational glyph.
  • Let capital letters communicate depth or family.
  • Use suffixes to distinguish containers, objects, steps, or forks.
  • Add numbers only when instance identity matters.
  • Choose one interpretation and apply it consistently within the file.

The purpose is not to force every team into one vocabulary. It is to give humans and AI a shared structural map.

Privacy through structural abstraction

RANGES also points toward a broader privacy model. Sensitive or proprietary identifiers can be replaced locally with compact structural coordinates while relationships, hierarchy, dependencies, and sequence remain intact.

An external AI could reason over the abstract shape while a local system retains the original mapping. The response could then be rehydrated into the private environment and executed under local control.

RANGES alone is not a complete privacy boundary. Comments, literals, routes, data, and business logic may still reveal meaning. Privacy requires deliberate local preprocessing, retained mappings, and controlled reconstruction. What RANGES contributes is a readable structural layer for that workflow.

Not minification. Not obfuscation.

Minification optimizes for machines by removing human readability. Obfuscation optimizes for concealment. RANGES is different: it attempts to preserve human scannability and machine usefulness at the same time.

Its strongest promise is not merely shorter code. It is a more precise interface between people, software, and AI—one where shape carries meaning, locations stay visible, and small changes can remain small.

The next interface for AI-assisted development

AI coding has largely inherited human naming conventions without asking whether those conventions are optimal for machine collaboration. RANGES asks a different question: what is the minimum readable representation that preserves everything needed to reason and act precisely?

The system is still evolving, and that is part of its value. It is open enough to test, measure, challenge, and improve. Its thesis is practical: stop making AI rediscover structure from prose, and start giving it structural coordinates directly.

Stop talking to AI only through descriptive names. Start targeting code through shape, relation, and position.
Roman Applied Sciences · Nodarama
Innovation 02

Prompt Honesty

The Prompt Stack and validating authorship with Reverse Citations.

Did AI write this article? Yes. But also no. It could not have written this without me.

Introducing the Prompt Stack

AI did not arrive carrying my experiences. It did not have a question or an idea circling around its head. It did not have a personal motivation, emotional vulnerability, insight, or initiative to begin writing this.

Before this article existed, there were questions, complaints, assumptions, corrections, disagreements, research requests, and decisions. There were also mistakes.

AI misunderstood me. I corrected it. It introduced useful language. I decided whether that language accurately represented what I had already described. It proposed structures. I accepted some, rejected others, and repeatedly set direction.

Traditional citations show where our information came from. A Reverse Citation shows where the work began.

1. The Universal Checkpoint

One of the behaviours humans often recognize as evidence of intelligence in other species is the ability to learn and teach the use of tools.

Generative AI may be the most broadly accessible cognitive tool humanity has ever created.

My own view is that AI represents a shared human checkpoint: a vast baseline of accumulated language, information, and patterns from which more people can begin.

When polished output becomes easier to produce, the finished surface becomes a weaker indicator of the thought behind it. More of its distinguishing value moves into the journey.

  • The original complaint.
  • The first question.
  • The unexpected connection.
  • The evidence request.
  • The disagreement.
  • The correction.
  • The decision to remove something.

2. The Fallacy of Linguistic Gatekeeping

Established industries, academic disciplines, and professional communities develop specialized language for good reasons. Terminology can compress complicated ideas and help people communicate precisely.

But terminology can also become a shortcut for judging capability. Memorizing terminology can present itself as capability. It is not a substitute for doing the work.

A Prompt Stack can show when a human described an experience or process before AI supplied the accepted terminology used to describe it.

Complaint or opportunity → question → idea.

3. The Authority of Experience

My perspective on AI does not come only from reading trend reports or observing other people use it. I have engaged AI across creative work, emotional reflection, business design, branding, strategy, software development, and system architecture.

That experience exposed what I think of as the 90% Complexity Trap. The beginning of a project can move astonishingly fast. The final stretch requires something different from generation. It requires governance.

The most consequential human contribution is often the moment the person says: “No. That is not what I asked for.”

4. A Personal Mission of Self-Directed Validation

I invested deeply in graduate-level education and completed all but one course of an MBA. I eventually chose to leave without the credential.

That decision changed how I think about validation. I am attempting to demonstrate capability through the systems I build, the arguments I publish, and the results I eventually produce.

The mountain of work disappears into the result. The Prompt Stack is not a substitute for results. It is a way to stop the process from becoming invisible.

5. The Three-Tiered Shield

The solution is not to hide AI use. Nor is it to flood every reader with an unreadable transcript.

  1. The Artifact: the polished article, report, presentation, software, or creative work.
  2. The Authorship Record: the chronological Prompt Stack, including questions, corrections, disagreements, and redactions.
  3. The Contribution Map: a readable summary of the most meaningful human and AI contributions.

6. What the Prompt Stack Can Show

  • Who introduced the original problem or thesis.
  • Which lived experiences shaped the argument.
  • What constraints governed the work.
  • Where AI supplied terminology or structure.
  • Where the human requested evidence.
  • When the human rejected an interpretation.
  • Which alternatives were considered.
  • What was intentionally excluded.
  • How the finished direction evolved.
  • Who accepted responsibility for publication.

7. What It Cannot Prove

A Prompt Stack is evidence. It is not omniscience.

It does not automatically prove that every statement is true, that no material was omitted, that the work is original, that the work is good, or that legal authorship and copyright have been established.

Its purpose is narrower: preserve the human process well enough that people can evaluate it with more information.

8. From Question to Working Tool

An observation became a question. The question became an idea. The idea became a method. The method became a working product.

This is the capability of AI: not unattended generation, but the ability to help a person investigate, articulate, and build something that began with their own curiosity and direction.

9. Download AuthMor Today

AuthMor is available as a tool for capturing and preserving a Prompt Stack. It can be used privately, published with completed work, or shared selectively.

The purpose is simple: people should have access to a practical way to preserve and demonstrate their human contribution to AI-assisted work.

Read the Complete Record

This artifact has been adapted from its original to better fit the editorial format of the website. The original document includes the article, complete Prompt Stack, summary, and actual output compiled from AuthMor—rhymes with author.

Concept, thesis, lived experience, direction, corrections, and final approval: Roman Kiefer
View the Original Prompt Honesty PDF Try AuthMor Prompt Capture for Free
Innovation 03

The Closed-Open

The book that helps you close it.

Lessons from gay relationships to help build yours.

Most of us are given a book about relationships long before we are old enough to question who wrote it.

Dating leads to commitment. Commitment leads to marriage. Love means forever. Fidelity means one thing. A good relationship moves through the appropriate stages toward the appropriate destination.

Then two actual people arrive, and the book gets considerably less useful.

Gay relationships have had an interesting relationship with that script. For a long time, many of us simply weren't in it. That came with obvious costs. But it also meant having to ask questions other couples could sometimes avoid.

What counts as intimacy? What does loyalty require? What does commitment actually mean? And perhaps most importantly:

Are we actually answering the same questions?

I have spent a lot of time thinking about that one. Not because I have perfected relationships. My learning style is on a need-to-know basis. I've apparently needed to know quite a lot.

We fall in love. Then what?

Fall in love is an interesting phrase. Falling happens to us. Relationships are full of decisions. We decide what we offer, what we accept, what we disclose, what we forgive, and what we are prepared to carry for each other.

The feeling may arrive long before the agreement. That creates a problem we do not talk about enough.

Two people can love each other sincerely and still be participating in two different relationships. I eventually wrote this down:

The relationship you have is the one they offer you. The relationship they have, is the one you offer them. They may never be the same one.

That is not necessarily betrayal. It may simply be two truths. Which leaves an uncomfortable question about the lie.

The relationship contract

Nothing says romance like contract law. So let's consider what a contract asks people to do: understand what is being offered, understand what is expected in return, know the terms, and recognize when those terms need to change.

That sounds cold until negotiation starts to look like curiosity, disclosure like care, consideration like generosity, and performance like commitment.

The point is not that relationships should become legal documents.

The point is that we frequently make agreements we have never actually discussed. We inherit words like partner, faithful, open, monogamous and committed, then assume the word completed the negotiation for us.

It didn't.

I have come to believe there may be only one universal rule: shared agreement.

Not agreement with me. Not agreement with a relationship expert. Agreement with the person who is actually in it with you.

And you cannot meaningfully agree to terms you do not understand.

Some things are not negotiable

People aren't contracts. Some things can be negotiated. Some things probably shouldn't be.

We have become very comfortable telling people to heal.

Heal before dating. Heal your wounds. Heal until nothing another person does can destabilize you.

I'm not convinced human beings work that cleanly.

I think there are some wounds that heal, and some that just need tenderness.

There is also a difference between being capable of enduring something and agreeing to endure it indefinitely.

And somewhere inside that distinction is a difficult question:

Am I helping carry something because circumstances temporarily require more of me, or is my effort preventing me from discovering what the other person would contribute if I stopped doing their part?

That is an uncomfortable question. I think it can also be a loving one.

Four ways of being together

At the centre of The Closed-Open is a relationship model built around four states:

Closed. Open. Open-Closed. Closed-Open.

They sound almost annoyingly similar. That is useful.

Because the differences are not primarily about sex. They are about where the work lives.

Trust. Transparency. Loyalty. Repair. Future planning. Contribution.

What happens when something breaks?

Do we repair, or simply return?

And what relationship actually exists beneath the name we have given it?

Two Truths & A Lie

There was a game we used to play at work: Two Truths & A Lie.

I did not expect it to become relationship theory, but my experience has taught me that everything is relationship theory—and none of the theory is as useful as understanding the relationship you have and the relationship they experience.

Those truths do not automatically become identical because you love each other.

The question becomes whether two different realities can coexist once both people can actually see them.

And then there is the lie, and learning how there can be a lie without a liar.

Because who among us has never made a prediction that was mis-labelled, misunderstood, or misrepresented and offered as a promise?

We inherited a script that asks sincere people to promise something neither of them can know.

In that sense, the book we were given makes a liar out of everyone who would dare promise forever.

This is why I am interested in learning and sharing what can actually be demonstrated.

Choice. Effort. Loyalty. Repair. Planning. Contribution.

The wrong order

Most of the important things in my life have happened in the wrong order.

School → Credential → Career.
Nope.
Career → Credibility → School → None of the above.

Hurt → Healing → Readiness → Love.
Nope.
Hurt → Love → Healing → Readiness.

Knowledge → Decision → Action.
More like:
Action → Knowledge → Decision.

The prescribed sequence, taught me nothing. My own sequence taught me everything.

That may be the larger argument underneath the book. We are given sequences for education, success, healing and love.

But life doesn't always happen in that sequence. Usually it happens the hard way.

Perhaps mastery is not completing the prescribed sequence.

Perhaps sometimes it is understanding the choice you are making well enough to make it completely. Even knowing you may choose differently later.

Especially knowing that.

So what is The Closed-Open?

It is not a book arguing that everybody should have an open relationship. It is not a book arguing that everybody should be monogamous.

And despite the title, it is not particularly interested in telling you whether to stay or leave.

I can't teach you what is right for you, nobody can, but I can teach you what I have learned about how to write a book of your own.

Gay relationships are the case study. Human relationships are the subject.

The book will explore trust and transparency, contracts and tenderness, love and endurance, contribution and choice.

It is being written to ask what happens when two sincere people discover that they have been living two sincere versions of the same partnership.

There will be four relationship states. There will be two truths. There will be a lie somewhere.

I'm just not convinced there needs to be a liar.

And the goal is to close the book.

The receipts

There is another experiment happening alongside this one. I am developing The Closed-Open with AI.

Through ROMAN Applied Sciences, I have also been working on Prompt Honesty: methods for preserving the human contribution behind AI-assisted work.

So I will publish this as more than a manuscript.

\ I will ship it as a test case and demonstration of AuthMor, which captures and shares the Prompt Stack: the author's contribution and inner dialogue as preserved in the prompts themselves.

The corrections. The contradictions. What AI suggested. What I accepted. What I rejected.

And AI assessments comparing what can be seen from the complete collaboration with what remains visible when only the human prompts are available.

Not because a prompt history proves everything.

It doesn't.

But if we're going to publish work developed with AI, trust me seems like a strange place to stop asking questions.

I'd rather show the work.

That fits comfortably inside what ROMAN Applied Sciences is becoming:

SYSTEMS. EXPRESSION. TRUST.

Nodarama. Sentuas. Roman Ulysses. Everyone Ever. AuthMor. The Closed-Open.

Different projects, but increasingly the same questions.

How do we build better systems?
How do we express what we actually mean?
And how do we know what—or who—to trust?

For now, I'm writing the book. I may invite people to help me publish it. And if I do, I intend to show my work.

AI will write the book. I just need to finish filling in the pages first.
ROMAN Applied Sciences presents · The Closed-Open · The book that helps you close it.
Enablement 01

Nodarama
Deterministic AI

Bespoke automation made to measure.

Nodarama Bespoke Automation is made to measure and patent pending. It is AI-powered and executed transparently on an enterprise-friendly, logic-governed kernel that runs locally.

The system allows enterprises to build their own competitive differentiation instead of renting many separate SaaS components, gluing them together, and calling the result an employee experience.

Nodarama enterprise automation diagram.
Learn More About Nodarama
Enablement 02

AuthMor

Authorship through prompt validation.

AuthMor is a free, donation-supported extension for Chrome and Firefox that extracts your prompts from an AI conversation, helping you build a complete Reverse Citation, or Prompt Stack.

Observe Install and Capture

AuthMor

Download AuthMor

Download AuthMor for Chrome AuthMor for Firefox Coming Soon Support AuthMor with an Optional Donation
Music 01, 02, 03

Roman Ulysses &
EVERYONE EVER

Songs, commissions, performance, and licensing.

Song Directory

Coming soon. For now, you can stream Roman Ulysses and Everyone Ever on Spotify, Apple Music, Amazon Music, and other major streaming services.

Purchase, Perform, or Commission Work

Song lyrics, commissions, and performance opportunities are coming soon.

Licensing

Licensing information is coming soon.