research/Interest

2 records · yaml · accepts undeclared fields

Fields

FieldTypeRequiredMeaning
slug string matching /^[a-z0-9][a-z0-9-]*$/ required Unique identifier for this interest; also the filename and the vocabulary token other records use to reference it.
stub exactly true optional Marks a thin record: hidden from default listings and search, but still served at its own URL.
renderings list of object (additional keys allowed) optional Generated human-facing pages built from this record; regenerate them rather than editing them.
    ↳ path non-empty string optional On-disk location of the generated page, omitted when it is rendered on demand.
    ↳ published string required Public URL where the generated page is served.
    ↳ kind non-empty string required Which kind of generated page this is.
provenance any shape (not yet constrained by the schema) optional How this record was captured, including caveats that must survive into its renderings.
edges list of object (additional keys allowed) optional Typed graph links from this record to other records.
    ↳ name non-empty string required Display name of the entity at the far end of this link.
    ↳ openalexId non-empty string optional OpenAlex identifier that disambiguates the linked entity.
    ↳ kind non-empty string required What kind of relationship this link represents.
    ↳ provenance any shape (not yet constrained by the schema) optional How this link was established.
name non-empty string required Display title shown for this interest on the public research homepage.
blurb non-empty string required Short explanation of why this interest matters, written in the homepage voice.
status one of: active | dormant required Whether this interest is still being actively pursued or has gone dormant and dropped off the homepage.
priorArt non-empty string required Note on prior art: related work already known about or rediscovered while pursuing this interest.

Rules

Defects

No defects in the 2 record(s) checked.

Example record

/Users/eshao/.config/mnt/data/research-interests/interests/anti-slop.yaml

# InterestRecord v0 - schema authority: _template.yaml.
slug: anti-slop
name: "Anti-slop: detecting and evading AI-text detectors"
blurb: >-
  Whether faithful, human-authored-feeling prose can survive contact with a production
  AI-text detector like Pangram, and what a generator-discriminator-oracle loop can and
  cannot learn about the machine signal it is trying to erase. Spans four linked
  questions: what remains of the signal once every surface tell is stripped (review 1),
  whether an adversarial hill-climb loop can keep improving without exhausting its
  signal or Goodharting its evaluator (review 2), whether a production detector can be
  reliably beaten and at what cost to the text (review 3), and whether self-play lets a
  generator durably out-run a fixed or co-evolving classifier (review 4).
status: active
priorArt: "Wu et al., A Survey on LLM-Generated Text Detection (2023, rev.) - https://arxiv.org/abs/2310.14724; Zhang et al., A Survey on Self-play Methods in Reinforcement Learning (2024, rev. 2025), open-challenges section names non-transitivity/cycling as unsolved - https://arxiv.org/abs/2408.01072; Arvind Narayanan's public Pangram-evasion attempt, cited as the review baseline in review-001 and review-003."

Where it lives

Declared in/Users/eshao/.config/mnt/mdr/skills/research-records/assets/schemas/interest.ts
BindinginterestDataStore
Directory/Users/eshao/.config/mnt/data/research-interests/interests
Files*.yaml