Most research either sits behind a paywall, behind a slide deck, or behind a private Discord. The reasoning that led to a result is almost never legible from the outside, so the alternatives never get argued in public. GIKSN exists to keep that record.

A paper here can be small. It can be a draft someone abandoned halfway. It can be a survey of a subfield that turns out to be already crowded. The format is borrowed from technical RFCs; the tone is closer to a working notebook.

Why the lab exists

AI and Deeptech get most of our cycles because that is where the lab thinks the next decade of leverage is. Hardware and Distributed Systems are the substrate everything else runs on, so we treat them as research targets in their own right, not plumbing. Cross-sector work is the point. Compute means little without the models on top of it, and the models mean little without the systems that let them run.

We bootstrap through the community. Funding comes after maturity and tangible output, not before. The site is the front door and the record. Telegram is the room where day-to-day work happens.

Who reads here

Three audiences, on purpose:

• Researchers looking for a place to write down half-formed ideas without waiting for a conference deadline.
• Builders who want to see the reasoning behind the tools and protocols the lab is shipping.
• The wider community reading quietly. Feedback is welcome. Applications to contribute are the path in.

What lives here

Four sectors plus a wire for lab updates. Every paper has an abstract, a body, a status, and a discussion thread. Authors are credited. Editors do not rewrite voice.

• Artificial Intelligence (AI). Foundational and applied work on models, agents, evaluation, alignment, and the road to AGI. Where the lab spends most of its cycles.
• Deeptech (DT). Bio, materials, energy, quantum, robotics. Research that sits at the physical frontier and takes years to compound.
• Hardware (HW). Silicon, accelerators, embedded systems, sensors. The compute substrate the rest of the frontier runs on.
• Distributed Systems (DS). Consensus, storage, coordination, protocols. The plumbing that lets frontier systems scale without silently breaking.
• Updates (UP). Announcements from the lab and the wider frontier. Cohort openings, collaborations, releases, program dates.

How statuses move

Status describes where the research is, not the editorial state of the document. Five stages:

• Exploration. Sketching. Open questions, no strong claims yet.
• Draft. Being written. The argument has a shape but is still being edited.
• Preprint. Public and open for critique. The lab is asking to be argued with.
• Published. Final version. The lab stands by it.
• Landmark. Foundational. Cited widely, referenced by later work.
• Product. Shipped and available. Used for entries that describe a tool or protocol the lab has built and released.

Status can move backward when reality demands it. That is the archive doing its job.

Contribution is gated

Anyone can read. Anyone can apply. Not everyone gets to work on the research. The wall exists because the work requires people who can actually understand it and continue it. Accepted applicants get contributor accounts on the platform and access tokens for the private Telegram channels.

The public Telegram channel is open. The private channels are where working groups coordinate. Both are linked from the community page once it lands.

Editing and credit

Authors keep editorial control of their own papers. Editors copy-edit for clarity and fix typos. Discussion is preserved verbatim, except where moderation removes a comment. In that case a placeholder is left in the thread with a one-line reason.

The record matters more than the verdict. We keep rejected directions because we want to remember what we considered.