published_gain.json

{
  "_README": [
    "Published cumulative-ascent figures to check lib/elevation_gain.py's dead",
    "band against. Read by check_elevation_gain.py.",
    "",
    "`sections` is EMPTY ON PURPOSE and the check fails while it stays that",
    "way. Inventing plausible-looking section figures would make an",
    "unvalidated threshold look validated, which is worse than no check: the",
    "whole reason this file exists is that a gain number can agree with a",
    "consensus by construction and mean nothing. An empty table that fails",
    "loudly is the honest state.",
    "",
    "To fill it in: pick sections with a well-known, quotable gain figure and",
    "unambiguous mileposts - the Whites (Franconia Ridge, the Presidentials),",
    "the Roan Highlands, the Blood Mountain climb - and record where each",
    "number came from. Two or three sections of genuinely different character",
    "are worth more than a dozen from the same guidebook, because the point is",
    "that ONE threshold has to work for all of them.",
    "",
    "Mileposts are northbound miles from Springer, matching",
    "elevation_profile.json's distance_mi. Note that AT mileages shift between",
    "years as the trail is relocated, so `source_year` is not optional detail."
  ],

  "_schema": {
    "whole_trail": {
      "published_gain_ft": "number, or null",
      "source": "where the figure came from",
      "source_year": "number"
    },
    "sections": [
      {
        "name": "human-readable, e.g. 'Franconia Ridge'",
        "start_mi": "number, northbound miles from Springer",
        "end_mi": "number",
        "published_gain_ft": "number",
        "source": "who published it",
        "source_year": "number"
      }
    ]
  },

  "whole_trail": {
    "published_gain_ft": 510000,
    "source": "commonly cited consensus figure for the full AT, as recorded in issue #91",
    "source_year": 2026,
    "_note": [
      "Approximate and rounded, and treated as such. It is reported by",
      "check_elevation_gain.py but never gates: one scalar against one free",
      "parameter is not a test, because some threshold always hits it."
    ]
  },

  "sections": []
}