The Black on the page: a first-attempt range marked low evidence; the Black panels load again
CI / test (pull_request) Successful in 40s

- The odds card for the Black: "Low evidence", "about 1 in N" first attempts, the 95% range written as
  "between 1 in X and 1 in Y", later attempts effectively never, why it is uncertain (every logged shadow is a
  rider who chose to post), and the steps (shadow -> reached the Black -> bonded). Dashed border.
- Valley map: the Black's routes are solved for a first attempt (they showed ~0% per attempt), labelled "on a first
  attempt", and "Why this route?" gives the same range and caveat.
- "The Black is not dice" panel: why the odds are uncertain, and what replaces the assumption (fresh-account bot
  runs to the tree).
- analysis/black/evidence.py wrote "odds_ratio": Infinity (no later-attempt shadow at all), which is not JSON:
  the page's two Black panels showed "could not be loaded". Now null, and the report says "unbounded".

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UCzTd31GphpMXDhQqWrCMQ
This commit is contained in:
2026-11-19 18:40:01 +00:00
co-authored by Claude Opus 5.5
parent d7befc8e18
commit b51a6c7039
6 changed files with 25 additions and 11 deletions
+10
View File
@@ -159,6 +159,8 @@
if (!b) { if (t2) t2.innerHTML = '<h3>Your first attempt is different</h3><p class="small muted">The shadow evidence could not be loaded.</p>'; return; }
const ff = b.form_first, fl = b.form_later, sr = b.same_route, as = b.after_shadow, gaze = as.answers["held its gaze"];
if (t2) t2.innerHTML = `<h3>Your first attempt is different</h3><p class="small">Among riders who helped their squadmate and climbed the tree, the shadow came in <b>${frac(ff)}</b> first attempts (${pct(ff[0] / ff[1], 1)}) and <b>${frac(fl)}</b> later ones. On the very same routes, first attempts saw it about ${Math.round(sr.first_shadows / Math.max(sr.expected, 1) * 10) / 10}× as often as chance would give. The shadow is not a dice roll you can repeat.</p>`;
const steps = $("#black-steps");
if (steps) steps.innerHTML = `How it breaks down: the shadow came in ${frac(ff)} reported first attempts (over-reported); after it, ${as.reached_black} riders reached the Black and <b>${as.bonded_black}</b> bonded (${pct(as.bonded_black / Math.max(as.reached_black, 1))}).`;
if (!box) return;
const br = b.best_route;
box.hidden = false;
@@ -170,6 +172,7 @@
${gaze ? `<li>At the Black, riders who held its gaze bonded in <b>${frac(gaze)}</b> logged answers.</li>` : ""}
<li>The bot has made ${b.bot.attempts.toLocaleString()} attempts on ${b.bot.accounts} accounts, ${b.bot.gated_tree} on the gated path, and seen the shadow ${b.bot.shadows} times: none of them was a first attempt${b.bot.earliest_attempt ? ` (its earliest was attempt ${b.bot.earliest_attempt})` : ""}, which fits.</li>
</ul>
<p class="small"><b>Why the Black's odds are uncertain.</b> The Black comes only after the shadow, and every logged shadow comes from riders who chose to post their run. Riders are much more likely to post a run with a shadow than one without, and no complete log (the bot, bulk logs) has a first attempt with a shadow yet, so the estimate assumes riders over-report the shadow as much as the meadow Blue. Fresh accounts the bot plays to the tree on their first attempt will replace that assumption; until then the odds are a range, not a figure.</p>
<p class="small muted">Method and every table: <code>analysis/reports/BLACK.md</code>; the numbers: <a href="data/black.json">data/black.json</a>.</p>`;
}
@@ -189,6 +192,13 @@
const art = (c) => window.DragonArt ? `<svg class="odd-art" viewBox="0 0 200 160" aria-hidden="true">${window.DragonArt.svg(POSE[c], { fill: `var(${DV[c]})`, eyeFill: "var(--bg)", dark: c === "Black" })}</svg>` : "";
$("#odds-colours").innerHTML = cols.map(({ c, first, p, pAll, lo, hi }) => {
const e = enc(c);
// The Black rests on one roll no complete log has measured (the shadow on a first attempt): a range, not a figure
if (first) return `<a class="odd odd-colour odd-low" href="#valley/${c.toLowerCase()}" style="--c: var(${DV[c]})">${art(c)}
<span class="pill warn">Low evidence</span>
<span class="big">about 1 in ${oneIn(p)}</span><span class="what">first attempts end bonded to a Black, on the best route</span>
${dots(Math.round(p * 1000), 1000, cssVar(DV[c]))}
<span class="how">Somewhere between <b>1 in ${oneIn(hi)}</b> and <b>1 in ${oneIn(lo)}</b> (95%). Later attempts: no shadow has ever been logged, so effectively never.${Math.abs(pAll - p) / p > 0.25 ? ` All form posts suggest about 1 in ${oneIn(pAll)}.` : ""} Every logged shadow comes from riders who chose to post their run, so this rests on an assumption until complete first-attempt logs exist. <span id="black-steps"></span></span>
<span class="go">See the route →</span></a>`;
return `<a class="odd odd-colour" href="#valley/${c.toLowerCase()}" style="--c: var(${DV[c]})">${art(c)}
<span class="big">${p >= 0.5 ? pct(p) : `1 in ${oneIn(p)}`}</span><span class="what">${p >= 0.5 ? "of runs" : "runs"} end bonded to ${/^[AEIOU]/.test(c) ? "an" : "a"} ${c}, on the best route${first ? ", on your first attempt" : ""}</span>
${dots(Math.round(p * 1000), 1000, cssVar(DV[c]))}