Method & Transparency
On the Work, the Method, and AI
The Physis project grew out of independent research that began in 2012. On this page I describe how the work accumulated, how the book Physis: A Theory of Power was written, and which tasks I used artificial intelligence for in its later stages. I set out the method's limits as openly as its strengths, for the reader to weigh.
The method described here also touches the thesis of Physis Theory: AI is a tool, and what it is depends on whose hands it is in.
The history of the research
Phase 1: Solitary (2012 – November 2024)
- 2012–2014
- I began examining Turkey's 2002–2012 period year by year; the survey spread from 2001 to 2014. Using almanacs and news archives I built chronological records of power. Processing a single year into finished form still takes about three months. Two volumes came out of this work: Turkey 2002–2008 and Turkey 2009–2012; the volumes were completed in March 2013. In time the recording method stopped being merely a working technique and became the way I read political life. I identified the power dynamics and the governing classes in this period; my observations on corporate entities matured in these same years. Having worked for years inside company structures, both governing and governed, fed those observations.
- 2014–2020
- I produced no new major work, but I kept up the record. The repetition of patterns in power relations while the actors changed strengthened the search for a historical framework.
- 2020–2023
- I began the Babil series in order to reread history through my own lens of power. The first volume, The Reign of the Aristocracy — then titled Ancient Ages — began in 2020 and was completed in 2022–23. Adapting my working technique to historical material, building the library and maturing a way of writing took about two years. I built the volume from collages of the sources I read and from my own notes; the sources are given in the volume's bibliography. While working on the second volume, The Rise of the Bourgeoisie, I brought the theory book forward.
The core of the theory was written before I began using AI. The Reign of the Aristocracy contains the tendency of power toward monopolization, the distinction between the law of nature and the law of men, the image of the power vacuum as a stone thrown into a still pool, the relation of freedom to the will and the will pool, the model of a community of six, the cycle of power and the biological basis of leadership. Most of these had not yet taken their present names or final definitions; their early forms remain in my dated working files. The framework's named concepts — the fraternity spike, the corporate-entity paradox, the V1/V2/V3/V4 evolution, ontological impermeability, the three-layered power conflict — were formulated on top of that core, during the writing of the theory book.
Phase 2: Encountering AI and writing the theory book (November 2024 – February 2026)
- November 2024 – February 2025
- I began trying out AI tools and saw the speed they gave me in research and checking. Writing the theory book had been planned for after the whole Babil series was finished; its turn had not come. Once I saw what the tool could do, I decided to begin while that capacity was in my hands, and changed the order of priority: AI did not produce this work, it changed its place in the queue. I began writing Physis: A Theory of Power on 21 February 2025. I opened the first conversation by asking for help in testing the model I had: governing classes, power conflict and corporate entities were the framework's starting point.
- February – September 2025
- The first writing cycle produced the rough draft of the book. I used AI for research, counter-arguments, gap detection and consistency checking. The text of that cycle and the references to thinkers were raw material for the next stage.
- October 2025 – February 2026
- I rebuilt the book on top of the rough draft. I settled the chapters, removed the section of proposed solutions, and clarified the mechanisms of resistance and fraternity. I weeded out most of the thinkers' references and examples added in the first cycle, and moved some of the rest into footnotes. By the end of February the book's structure and core mechanisms were in place.
Phase 3: Registration and first publications (March – May 2026)
- March – May 2026
- Three works were registered; Physis: 93 Falsifiable Hypotheses and the Preliminary Glossary were published on SSRN, and the Nomos Paradigm articles began to appear. In this period I used AI for translation from Turkish into English, for consistency of terms, for academic formatting and for editorial work.
Phase 4: Sharpening of the theory and completion of the book (June 2026 –)
- June – September 2026
- I worked out the book's mechanisms in detail, clarified its terminology, and established the links between examples and mechanisms. Typesetting and chapter-by-chapter editorial work were done. The mechanisms, the syntheses and the theory's frames are mine; in this cycle AI worked on verification, gap detection and the checking of footnotes.
- October 2026 –
- After a final reading, the chapters of the book are being published on Zenodo one by one, each with its own DOI, and the same texts appear on this site. Once the chapters are complete, the full version of the book will be published as a separate record.
- The coming period
- My priority is preparing the completed volume The Reign of the Aristocracy for publication and bringing out the English edition of Physis; the article series continue alongside. If time allows, an Almanak volume will also be published. After that I will return to the unfinished Rise of the Bourgeoisie. The Triade Engine v1.0 release and its dedicated technical publication are targeted for the first half of 2028.
Since February 2025 I have worked five or six days a week, eight to ten hours a day. AI sped up the cycles of research and checking; this tempo is part of the work I have kept up while turning the accumulation of earlier years into a book.
The four strands of the research programme
Almanak is my method of chronological recording; Babil provides the historical synthesis, Physis the theoretical framework, and Triade Engine the applied analytical extension. These strands were not designed from the outset as a single system; they developed as independent works that fed one another. First I saw the patterns in the records, then followed them in historical research, then built the theoretical framework. I state the framework's claims in a form that can be tested.
How I work
I come to a subject with knowledge and a view of my own. Where I do not know, I ask, and I compare what comes back with what I already know. As the research widens I order the material, gain command of the subject and sharpen my own angle. When I need a source I go to it; if I have no time and its use goes beyond a single paragraph, I buy the book and put it in my library. Where a source contains a passage I need, I place it in the work as collage, with its citation.
Research is also learning for me. Not every text that passes through my hands enters the book. Sometimes a single main heading, sometimes the history, etymology or biological context of a single concept, runs past two hundred pages in one AI session. The pages I have read and worked through in those sessions alone run into the tens of thousands; books and independent research are outside that count.
In moving from material to the book's text, the selection, the order and the route are mine. I do not count the research documents AI brings as the book's draft; the draft is the text I make by weighing that material and working it again. AI also brought category and content proposals from time to time; I either rejected them or rebuilt them according to the framework's own logic. At the level of the text I took suggestions from AI on shortening long sentences, correcting the language, smoothing transitions and redrafting. I also used its help in checking footnotes, in consistency of terms and in academic formatting. In this work I leaned on AI far more than in research and the checking of arguments; these are areas AI does better and in which I have no command. I weighed the suggestions and decided myself what would enter the text; that editorial help also left its mark on the wording of the final text. The book's construction, its concepts, its founding arguments and its historical syntheses are mine.
After writing a text I check it myself, and before taking it to its final form I run a separate consistency check with AI. As the work proceeds, differences of knowledge, perspective and language can arise between the chapter written first and the one written last; a second check helps me see them, and I weigh every warning it raises. When I changed the definition of freedom, for instance, I used AI to find where that change touched the text and tested the consistency of the definition and the examples.
In another case, I had described the power vacuum in my earlier volume through the waves of a stone thrown into a pool. In the first writing cycle AI added a thinker's reference, examples and a summary to that image. In the second cycle I rebuilt the principle through its mechanism; both the image and the reference were removed. A proposal having entered a draft does not mean it stays in the book.
What AI contributed
I used AI first as a search and research assistant: it answered questions, brought sources and texts, and helped me widen research headings. Then I used it for gap detection, internal consistency, argument testing and footnote checking. Working alone, it gave me an interlocutor that could point to the weak place in an argument and sometimes produced useful counter-arguments. Some of the points it found I repaired; on others I changed my mind. The quality of its proposals varied. This support shortened the writing considerably; the core of the research, formed over years, goes back to before any use of AI.
I did not use AI as a source of conceptual novelty. This is not a preference but a measured limit: asked to propose a new concept, the tool either renames an existing one or offers something superficial. Nor does it perform historical synthesis; left without a framework, it produces cliché. Synthesis needs a framework that makes the connections and sets the direction: if you do not know where you are going, following the tool's output takes a researcher far from what he aimed at. That the core of the framework was ready before AI was decisive for this reason.
AI and class
AI is a tool; in the hands of the upper class it concentrates hegemony, in the hands of the lower class it can be something else. For someone without academic infrastructure, editorial support, a translator or an assistant, AI is a lever that turns years of accumulated intellectual labour into structured work. The asymmetry is sharp: use by the upper class compounds with scale and capital, use by the lower class stays individual. I see my own case in the second category.
My relation to the classics
I know some of the basic concepts of modern social theory, and I collect the relevant works. Even so, I have not done comprehensive, direct and systematic reading of figures such as Bourdieu, Foucault, Gramsci, Marx and Weber; I left that until after the founding stage. I carried out the comparisons with my own work largely through summaries provided by AI. At points I judged critical I cross-checked across different sessions and tools. The thinkers' references AI added to the draft in the first cycle were reselected by me in the second.
This was a working choice, and it had three reasons: reading the classics before the framework had settled would have surrendered it to their field of gravity; a dialogue begun during the writing would have scattered connections that were still being woven; and measuring unfinished ideas of one's own against established texts would have eroded the confidence needed to carry the work on. In the early years the patterns had not yet become a framework against which I could compare; once the framework matured and the writing of the book began, I left deep reading of the classics until after the flow of writing. Instead of a broad literature survey I chose to research at the intersections I had selected. I make no claim to expertise in philosophy or biology; even so, one of the grounds the theory rests on is biology. My centre is power and history; in philosophy as in biology I follow what touches that field. Since the framework's founding stage is now complete, I regard direct comparison as part of the work and the criticism that follow.
The limit of indirect comparison is clear: AI summaries can contain errors, and I may have missed important similarities or distinctions. For that reason my own assessment of the work's originality remains open to testing, by my own reading and by others' independent examination. Appendix 1 of the book sets out these indirect encounters one by one and opens them to the reader's scrutiny.
Limits and checking
AI can produce false information, statistics, quotations and academic references. Some of the references it proposed in the early drafts were wrong. I reworked the footnotes together with the book in every pass, and had the citations checked again and again in separate sessions. I did not open each citation at its source and verify it myself. That method lowers the probability of error, but it does not take the place of checking every source one by one. Errors that come to light will be corrected in later versions.
AI can miss connections in long texts with many variables, and can give different answers to the same question in different sessions. So I compare research results against my own framework and recheck important changes within the whole of the text. Some lines I abandoned: the formal models I built with AI out of the thought experiment of a community of six, for instance, began at a certain point to confirm their own assumptions rather than the theory. Those models are not in the book; the thought experiment and its own simple arithmetic remain in the chapter on Freedom.
The nature of my relation to sources, and the absence of direct reading of the classics, are limits the reader should know when weighing the book. The project has also been run by one person since 2012 and developed outside the academic ecosystem: the work is published independently, not through a university, a publishing house or a research group. Both are at once an advantage and a limit. Working single-handed gives consistency and unity of vision, but it creates a single point of failure and does not scale. Independence made it possible to build the framework without the pressure of an institutional agenda, but the corrective pressure of peer review was absent during its formation, and critical reading begins directly in the reader's hands, without passing through an institutional filter. Since the later volumes of the Babil series are not complete, the basis for claims about dynamics that repeat through history is also open to development. Triade Engine is in development; until the targeted v1.0 and its technical publication appear, the empirical performance of the applied extension cannot be publicly assessed.
I built the framework so that it can be tested from the outset: where the hypotheses can break is written out in the book. I regard criticism not as a threat that would demolish the framework but as the ground on which it develops. The work from here runs in two directions: following the framework's implications, and actively trying to break it. I distinguish a single exception, a limit of the framework's scope and a genuine counter-example from one another; the first two narrow the definition, the third requires rebuilding the mechanism. I use academic tools not as the mechanism of production but to test the framework once it has been built. The book is not a closed system but the opening record of ongoing research.
I keep the working files of the first writing cycle, the Scrivener version history and the export of the Grok conversations. Since December 2025 I have made a habit of deleting conversations once they are finished, for reasons of security; the record for that period consists of the documents and the version history. These records and the published texts may allow the process I describe to be examined; the working decisions and the extent of the labour I report here are my own statement.
Why I share this
Within these limits, I am satisfied with how productive the method has been and consider it a useful way of working in its own right.
Explaining my method lets the reader see how the work was produced and within what limits it should be judged. This way of working may also be useful to another independent researcher; someone else can change it, improve it or criticise it. I share it not as a model for everyone, but as one concrete case of research carried on alone since 2012.
Artificial intelligence is an era technology, like electricity and the internet. It changed my own way of working for good, in productivity, in accuracy and in what can be brought to completion; and I am certain I am not alone in drawing on it, at least on the social-science side. In its later stages far more effective and more creative uses will appear, and the amount of work produced will grow many times over — independently of its quality. I am curious about the ideas and openings that capacity will bring out, and I record my own use as one of the examples of this era.
I reviewed and approved the AI proposals that entered the text. The book's founding ideas, arguments and historical syntheses are mine.
Registrations — Republic of Türkiye, Ministry of Culture and Tourism, Directorate General of Copyright:
Physis: A Theory of Power — Registration No: 2026/17534 (5 March 2026)
Triade Motor — Registration No: 2026/18239 (17 April 2026)
Almanak: Structured Power Chronology Recording Method — Registration No: 2026/18228 (17 April 2026)