How These Essays Are Made
A look at the process on Reality is Malleable
The essays on Reality Is Malleable are built through an AI-augmented writing process.
That does not mean pressing a button and asking a chatbot to produce an essay. It means using AI as part of a larger workflow: exploratory conversation, research synthesis, outline building, multi-model critique, section-by-section drafting, and final human judgment.
I use AI because I think this is where knowledge work is going. Not as a replacement for curiosity, taste, reading, lived experience, or judgment, but as an amplifier for them. A good AI workflow can help map an idea faster, test it from more angles, find weak spots, organize sprawling thoughts, and turn scattered fragments into something coherent enough to share.
The essays here are the result of that process.
They begin with questions, hunches, observations, books, papers, news items, conversations, and half-formed tensions that seem worth exploring. AI helps me examine those materials, but the direction, priorities, arguments, and final choices remain mine.
Here is how the process usually works.
1. The Starting Point
Most essays begin with a spark.
Sometimes it comes from something I read in the news. Sometimes it comes from an academic paper, a book, a historical pattern, a technology announcement, or a strange little contradiction in everyday life. Sometimes it emerges during an exploratory conversation with an AI model, where one tangent opens into something larger.
At this stage, the idea is usually not an argument yet. It might only be a question:
Why does this trend feel underexplored?
What happens if this second-order effect matters more than the obvious headline?
Is there a deeper pattern connecting several things that are usually discussed separately?
The goal at the beginning is not to force an answer. It is to notice that something is worth thinking about.
2. Exploratory Conversation
Once an idea catches my attention, I start talking it through with AI.
This is the messy stage. I test angles, follow tangents, ask questions, challenge assumptions, and try to figure out whether the idea has enough substance to become an essay.
The conversation might bring in economics, technology, geopolitics, philosophy, logistics, culture, history, science fiction, or whatever else seems relevant. If I have been reading a book, paper, or news story that connects to the topic, that gets added to the mix and examined as part of the broader exploration.
This stage is not about producing clean prose. It is about mapping the territory.
Some paths dead-end. Some ideas turn out to be obvious once examined closely. Others reveal hidden structure. The point is to think widely before deciding what the essay is actually about.
3. Research and Source Integration
The essays are not built only from conversation.
When relevant, I bring in material from outside the chat: news articles, academic papers, books, reports, historical examples, technical documentation, personal observations, and previous essays I have written.
AI helps examine these materials in context. It can summarize, compare, extract implications, and connect them to the developing argument. But the important part is not just gathering references. It is figuring out what role each piece of information plays.
Does it support the argument?
Does it complicate it?
Does it reveal a counterpoint?
Does it belong in the essay at all?
A useful source is not automatically included. It has to earn its place in the structure.
4. Finding the Shape of the Essay
After enough exploration, the idea usually starts to clarify.
At that point, I ask AI to help organize the conversation into a structured outline. This turns a messy field of notes, arguments, examples, and tangents into something more deliberate.
The outline is not treated as sacred. It is a working map.
Sections move around. Some get cut. Some expand. New connective tissue appears. Weak arguments get flagged. Stronger framings emerge. The goal is to find the shape that best serves the idea, rather than forcing the idea into the first structure that appears.
This is often where an essay becomes real.
5. Multi-Model Review
I usually do not rely on a single AI model.
For many essays, I bring in several different models and ask them to critique the outline, stress-test the argument, identify blind spots, suggest missing considerations, or point out where the framing is unclear.
Different models have different strengths. Some are better at structure. Some are better at skepticism. Some are better at technical detail. Some are better at noticing rhetorical drift or weak transitions.
I treat this less like asking one assistant for an answer and more like running the idea through a small review panel.
The goal is not to obey every suggestion. It is to expose the essay to enough different forms of critique that the final version becomes sturdier.
6. Tone and Framing
Once the structure is solid, I think about tone.
Not every essay should sound the same. Some subjects need a direct analytical voice. Some need a more speculative or reflective approach. Some can carry a bit more humor. Some need restraint because the topic itself is already complicated enough.
AI helps build tone guides, test openings, refine transitions, and identify where an essay should stay plain versus where it can afford a more vivid turn of phrase.
The aim is not decoration. The aim is fit.
The tone should serve the argument.
7. Section-by-Section Drafting
I rarely draft an entire essay in one pass.
Instead, I work section by section. This keeps the process focused and reduces the chance of the essay becoming generic, rushed, or structurally mushy.
Each section is treated as its own small project. I may draft it, revise it, question whether it belongs, adjust its role in the larger argument, or rewrite it completely. If a new idea appears during drafting, it can still be integrated, but only if it improves the whole piece.
This modular process is slower than asking for a full essay at once, but it produces much better work.
It also keeps me involved at every stage.
8. Revision and Cohesion
Once a full draft exists, the essay goes through revision.
This includes checking transitions, trimming repetition, improving section flow, strengthening weak claims, smoothing awkward phrasing, and making sure the argument actually builds from beginning to end.
AI is useful here because it can hold a large amount of structure in view. It can notice repeated points, identify unclear leaps, and suggest where a reader might get lost.
But revision is not just cleanup. It is where the essay becomes more honest about what it is trying to say.
Sometimes a section gets cut because it is interesting but unnecessary. Sometimes a paragraph gets expanded because the idea is doing more work than expected. Sometimes the conclusion changes because the essay has evolved while being written.
That is normal.
9. Final Human Pass
The final pass is mine.
This is where I decide what stays, what goes, and what no longer sounds like something I would actually say. I adjust phrasing, remove over-polished lines, check the argument against my own judgment, and make sure the essay reflects my thinking rather than merely producing a smooth imitation of it.
AI can assist, challenge, organize, and draft.
It does not get the final vote.
The finished essay has to match my standards, my curiosity, my worldview, and my sense of what is worth putting into public.
Why Use This Process?
Because I think AI-augmented workflows are going to become a normal part of serious thinking and writing.
Used badly, AI can produce slop: generic, hollow, derivative text with no real perspective behind it.
Used well, it can become a powerful thinking partner. It can help one person explore more angles, compare more interpretations, stress-test more assumptions, and build more ambitious projects than they could manage alone.
That is the experiment behind this site.
Reality Is Malleable is not an attempt to hide the use of AI. It is an attempt to use AI openly, deliberately, and thoughtfully as part of a larger intellectual workflow.
The human part is still the important part: curiosity, judgment, taste, skepticism, lived experience, and the willingness to keep asking better questions.
AI expands the workspace.
It does not replace the person working inside it.