Abstract
Shomosite is my site for essays, product theses, and evolving research notes. It is not arranged around announcements or a career narrative. It is arranged around inquiry: ideas enter when they are stable enough to help a serious reader, and they keep changing when criticism shows that they should.
I am Shomodip De, a writer-builder interested in AI, psychology, productivity, knowledge work, and the problem of how human beings create better explanations. This site exists because inquiry is not finished when a page first appears. It needs a place where claims, examples, products, and cross-references can be corrected over time.
What This Site Is For
I do not think of this site as a stream of posts. A stream is useful when the main fact about a piece of writing is when it appeared. Most of the writing I care about is not like that. A question may sit quietly for months, return through a book, reappear inside a product problem, and then become clearer only after another essay exposes the weakness in the first framing.
Shomosite exists to give that process a durable surface. The pages here are allowed to keep their identity while their contents improve. A piece can begin as a stable enough explanation, then gain a qualification, lose a false claim, point to a better neighboring page, or be partially superseded by a sharper argument. That is not a defect in the form. It is the form doing its job.
The site has three main regions. Prose is where I work through arguments. Product is where selected ideas become systems I might build or already care about building. Docs explains the site itself: its design, architecture, and reading conventions. These divisions are practical, not metaphysical. An essay may produce a product idea. A product may force a better essay. A design note may clarify why the site has to behave a certain way.
The Reader I Am Writing For
The primary reader is a serious stranger: someone who arrived through one page, sensed that there may be more behind it, and wants to understand what kind of site they have entered.
I am not writing mainly for a feed, a professional bio, or a room of people who already agree with me. I am writing for a reader who can tolerate a living argument: one that makes claims directly, but does not pretend that its current form is final. I want such a reader to see the work, not only the conclusion.
That is also why this page is about the site before it is about me. A short biography can tell you that I am a writer, researcher, and product manager. It can tell you that I have worked on B2B platform products in enterprise ontology and data productization. It can tell you that I care about AI, psychology, productivity, and human existential wellbeing. Those facts are useful, but they are not the real explanation. The real explanation is the working pattern: questions become pages, pages become a corpus, the corpus changes the questions, and some questions become products.
The Fun Criterion
I take interest seriously. Not because every interesting thing is important, and not because pleasure is an argument, but because sustained interest is often the first sign that a question has energy in it. The fun criterion is my name for that signal.
Fun, in this sense, is not amusement. It is the arrival of inspiration, creative thought, and the energy to act. When an idea keeps returning with that quality, I pay attention. It may be wrong. It may be premature. It may lead nowhere. But if it produces enough pressure to make me read, think, write, or build, then it has already passed a threshold that many dutiful topics never cross.
This matters because writing without live interest usually becomes imitation. It reaches for established topics, established forms, and established vocabulary. It may become correct in a thin way, but it does not generate much. The pieces I want to write begin when a question becomes hard to leave alone.
Error Correction
The pages here are not permanent verdicts. They are attempts that remain open to correction.
By error I mean more than a false claim. A page can be wrong because a sentence states something untrue. It can also be wrong because the explanation is weak, the framing is misleading, the emphasis is distorted, or the page does not yet know how it relates to another part of the corpus. A missing relation is a real error when the relation would make the idea more intelligible.
That is why error correction happens at several levels. I may revise an individual page. I may add a new page that makes an older page look incomplete. I may change the relation between two pieces. I may change the broader world model that gives the pieces their meaning. The unit of correction is not only the sentence. It is also the page, the neighborhood around the page, and the pattern that holds the site together.
This is not a performance of uncertainty. I dislike vague writing that hides from criticism by refusing to say anything definite. A page should make claims clearly enough to be wrong. But once it can be wrong, it should also be able to improve.
Essays And Products
Most ideas should remain essays. Some should become products.
That distinction matters. Products are much more selective than essays because building asks for a different kind of commitment. A page can explore an idea because it is alive enough to examine. A product asks whether the idea deserves tools, interfaces, workflows, maintenance, and users. Very few idea spaces justify that extra burden.
When an idea does cross that line, product work is not separate from the writing. It is one way of putting the idea to practical use. A product turns an explanation into constraints: what the system must remember, what it must hide, what it must make easy, what it must prevent, and what kind of person it assumes the user is.
The correction loop remains the same. My products, product theory, and product-building skill improve through criticism just as essays do. A bad product idea may expose a weak essay. A practical constraint may reveal that a theory was too loose. A working tool may create new questions the original essay could not have predicted. That is why products belong here, but only selectively.
AI In The Workshop
AI is part of my working environment, but not the author of the work.
I use it as a grunt-work assistant: source research, link maintenance, editing, summarizing, checking, converting, reorganizing, and keeping the site machinery from becoming too expensive to maintain. Those tasks matter because a serious corpus creates tedious work around itself. If the upkeep becomes too heavy, the writing system decays.
AI is also one of my research subjects. It changes what can be delegated, what can be inspected, what kinds of tools become cheap, and what kinds of intellectual laziness become tempting. So it appears here in two roles at once: as a tool in the workshop and as an object on the bench.
The boundary is important. AI can help move material, test explanations, find sources, and make maintenance less punishing. It cannot decide what I find interesting. It cannot take responsibility for the world model. It cannot replace the criticism that makes a page worth keeping. That is the point of using AI in the workshop rather than treating it as a substitute for judgment.
How To Read The Site
You do not have to read this site in order.
Start wherever the question is live for you. If an essay gives you enough, stop there. If a term points elsewhere, follow it. If a margin note opens beside a paragraph, treat it as a second voice in the page: qualification, context, or objection. If a product page appears near an essay, read it as a practical extension of a smaller number of ideas, not as a catalog item.
The site is meant to reward return. A page may be clearer later than it is now. A weak relation may become a strong one. A product note may move from thesis to system. An old essay may become less central because a better explanation has appeared beside it.
That is not housekeeping around the work. That is the work.