LOLM is developed through a working group that meets weekly. This document details the structure for the process by which we surface, construct, and editorially manage new entries in the project.
An entry in LLM should be neither purely a technical history nor a media-theoretical critique of a term, but rather a way to open avenues to further thinking on both of these fronts. The project is designed to contextualize the way in which terms in the NLP technical literature emerge, come to prominence, and leak historical and philosophical meaning through their circulation.
In contrast to a ‘collected edition’ of contributions, where each entry constitutes a standalone essay that develops a self-contained argument, LOLM entries are conceived more like modules in an open source codebase. Where possible, they should draw on and refer to explanations in other entries. They should also be brief and as tightly encapsulated as possible. In practice, this means that entries should normally be around 5-10 paragraphs. (See the entry for LLM as an example.)
Each entry should target at least one of the aims listed in the bullet points below. Though entries can achieve multiple of the aims below, they only need to address one, as we want to keep a relatively low barrier to entry for new members of the working group / new terms. As a general rule, we prefer to have multiple entries for closely related terms rather than entries that cover multiple terms.
New terms are brought to the LOLM working group through a steward. Any member of the working group can self-elect as a new term’s steward, and all working group sessions will begin with an open call for new term-stewards. Not all working group members need to self-elect as a term-steward, and there is no limit to the number of concurrent terms for which one person can self-elect as steward.
A term’s steward brings a new term to the working group with at least one and ideally multiple of the following:
Once a new term has been presented by its steward, the working group will collectively discuss it. The aim of this discussion is to fill out the body of the aims listed above in the ‘General method’. Post-discussion, the working group will then elect an editor for the term who will work with the steward to fill an entry in LOLM (see next section).
An editor is responsible for filling out an entry in LOLM itself by working with the term-steward and other invested/interested parties.
This contribution will ultimately take the form of a pull request to the LOLM GitHub repository. All sections of LOLM are written in Typst, and LOLM is envisioned as being published concurrently as a website (https://lolm.ohrg.org/), a PDF, and an EPUB through the Rheo toolchain. (Merging a pull request to the main branch of the LOLM GitHub repository will trigger a recompilation of all three of these formats.)
Whereas the term-steward is responsible for surfacing the dimensions of a term that should be documented in its LOLM entry, the term-editor is responsible for compiling these many threads into a single and coherent entry in the project. The construction of this entry may benefit from separate meetings between steward and editor, and perhaps even sub-working groups if there are many interested parties.
There is no strict timeframe on a term’s transition from discussion (via steward) to page (via editor). Each working group session will assess the terms that are in flight, asking status updates of steward/editor pairs and whether the contribution would benefit from any further discussion or help.
The LOLM working group uses the following tools and platforms: