Middle-Out Architecture: Free Movement Between Fragment, Work, and Knowledge System

Inspired by Cristina Marras's explanation in the field of digital humanities, the logic of knowledge production in Leralink can be understood as a sort of "middle-out" approach. In this approach, knowledge production neither necessarily begins with a macro, pre-determined plan, nor does it inevitably have to move toward a larger structure from the disorganized accumulation of notes and small fragments. The starting point can be at any level, and the author, researcher, or producing group can freely navigate between different levels throughout the process.

Most content creation and publishing systems impose one of two paths on the user. In the top-down approach, the final shape of the work is defined first: for example, a book, course, article, journal issue, or research collection is created; then chapters, sections, and content units are placed within it. Before starting, the writer must more or less know what they want to create, what the final structure is, and where each section will fit into it.

In the bottom-up approach, work starts with small units: notes, definitions, observations, quotes, sources, images, audio files, or independent arguments. These units are gradually accumulated, linked, and categorized, and may ultimately turn into an article, book, course, or larger structure. In this case, the final shape of the work is not clear from the beginning and emerges during the course of knowledge production.

Leralink accommodates both paths, but does not stop at either. Its key feature is the ability to constantly move between these two directions. The user can start with a small piece, create a few related pieces, turn them into an essay, and then place that essay within a larger book or course. In the opposite direction, they can first design a collection, book, or curriculum, outline the structure of chapters and sections, and then gradually produce the necessary pieces within this structure or retrieve them from the knowledge library.

Therefore, middle-out in Leralink does not mean finding a fixed point in the middle of a hierarchy. The "middle" here is an operational and fluid position: each level can be both the destination for combining smaller units and the starting point for building larger units. A piece can be a fundamental unit of writing at one moment, and at another, become part of an article, lesson, or book. That same article can later be placed as a unit within a journal issue, a thematic collection, or a larger training program.

This logic is linked to the concept of continuous authoring, which I tried to design in Leralink in a way that is currently desirable for me. In the traditional model, a work usually passes through several relatively separate stages: planning, writing, editing, finalizing, and publishing. Publication is considered the end point of the work, and any subsequent change requires a new print run, revision, or version. In Leralink, authoring can be an ongoing process. Pieces are gradually created, modified, completed, translated, and connected to new sources; larger structures can also evolve alongside the growth of the pieces. Another example of an mismatched experience occurs in software like Word and Google Docs. Of course, Google Docs must be credited for somewhat addressing the problem by adding tabs. Google Docs or Microsoft Word give the writer an infinite canvas for writing.

A researcher, for instance, might begin by recording a few scattered observations. Later, they might transform these observations into independent concepts, sources, and arguments. Some of them may be published in a public note, others may form an academic paper, and a broader collection of those same fragments may lead to a book. The publication of a note or paper does not mean the end of the fragments' lives. They can be revised, presented in another language, used in an educational course, or connect with new fragments. The issue is that all these stages occur in a scattered, unrelated manner and in fragmented domains, rather than in a single, coherent ecosystem.

In such a model, the author is not forced to choose between "free writing" and "structured design." They can move bottom-up when ideas are forming, and redesign the structure top-down once the overall form of the work becomes clearer. While designing a book, they might realize that some chapters do not yet have the necessary fragments; or while working on a fragment, they might discover that it could be the starting point of an independent project. The system must support both movements, without the content being copied, losing its relationship with its sources, or becoming locked into a specific structure.

The same possibility exists in publishing. A book might initially be defined as a complete project and written chapter by chapter. Conversely, it might be formed by combining a collection of essays and fragments produced over several years. Even after publication, its fragments can be used in other works, without losing their identity, origin, history, or relationship with the first book.

The middle-out approach becomes even more important in collaborative projects. Not all contributors necessarily work at the same level. One person might design the architecture of the entire course or book; another might produce content fragments; a researcher might find sources; a translator might develop a language version; and an editor might review the relationships between fragments. Leralink’s architecture does not place these activities into a rigid hierarchy, but rather connects them at different levels of a shared system.

This perspective also aligns with the collaborative logic of the middle-out approach. In some applications of this concept, middle-out is an effort to reduce the division between the top-down knowledge of experts and the knowledge arising from the experience of practitioners and communities. Neither is the specialized structure eliminated, nor do scattered experiences alone replace it; the two must influence each other in a shared space. Research related to collaborative projects has also described middle-out as a space for dialogue between specialized and experiential knowledge, rather than the complete dominance of one over the other.

In Leralink, this principle can take on a technical and institutional form. The macro structure of the project, scientific standards, and the validation system may be defined by an institution or a group of experts; but fragments of knowledge can be produced and completed by researchers, translators, teachers, artists, students, or members of a community. The structure from above influences the content, and the actual content can also refine the structure. If new fragments show that the initial classification is inadequate, the project's architecture must be alterable. If a project grows from a set of small contributions, it must be possible to define an appropriate structure, standard, and publication format for it.

In this regard, Leralink does not draw a hard line between the "work" and the "process of producing the work." A book, course, article, or archive is not a final, closed container; it is a temporary and concrete manifestation of a growing network of knowledge fragments. Structure gives meaning and direction to content, but fragments also make new structures possible.

The consequence of this architecture is that starting work is no longer dependent on having a complete map. The user can begin with a single fragment, a question, a source, or an idea. At the same time, someone who has the blueprint for a digital university, encyclopedia, book series, or educational program from the outset can also define the macro structure and gradually cultivate it. Both paths are valid and, more importantly, the chosen paths are not irreversible.

From this perspective, Leralink is not a system that merely serves to store final content, but rather an environment for the structural growth of knowledge. The smallest units can become large works; large works can be opened up again into reusable, translatable, and combinable units; and the author can move between details and the whole, production and design, exploration and organization throughout this process. This is the practical meaning of middle-out in Leralink: an architecture that determines neither the beginning nor the end, but rather enables continuous movement between different scales of thinking and building.

من این افتخار را داشتم که در یک برنامه پژوهشی و در همکاری با کریستینا ماراس شرکت کنم. او در کار مشترکش با همکاران دیگر در زمینه رویکرد میان‌گستر کار بسیار ارزشمندی کرده است. رویکرد پراگماتیک او و اهمیتی که به عدالت اپیستمیک در مقوله علم باز دارد الهام بخش من بوده است و امید مرا در پیگیری چارچوب کاری لرالینک بیشتر کرده است.