Live Threshold

Language English Español Português العربية Русский 中文

This page answers:What should this line's threshold be, so that it is not a guess?

What is this small life-chain called (may be left empty)
Anchor instance —— which thing has already reached some level (left empty it is written as “an unnamed instance”)
★ The level the anchor instance has already reached (required) —— this is the unit U of “one standard level”
Height —— the height this thing should reach; it decides how many tiers the threshold has (detail level ∝ height)
Override the positive-threshold multiplier by hand (empty = use the default tier for the height)
Override the negative-threshold multiplier by hand (empty = use the default tier for the height)

What this page is computing (the conventions first, so nothing is ambiguous)
This mechanism is the user's own synthesis of their practice, not an academic theory — no DOI, no named proposer, no peer review. Words like significance, confidence interval, p-value or effect size will not appear on the page, because this mechanism has not one statistic in it. The only academic items it can be set against are the names NCA / QCA (both ask “how far short counts as short”), for comparison only, never mixed in.

Its six struts: ① one main life-chain carries several small life-chains; ② each small life-chain is determined by a height; ③ the detail level of a threshold ∝ the height; ④ a threshold may not be decided off the top of your head — take one thing that has reached a standard and fix that; ⑤ thresholds split into positive / negative, running respectively on the upper / lower rail of the timeline; ⑥ reaching a suitable quadrant starts the next one. The upshot is point ⑦: the threshold is alive.

“The threshold is alive” — alive where, exactly: it is not that the threshold floats by itself, it is that the anchor moves up — whenever a measurement ≥ the positive threshold T⁺ appears, that run's achieved level is fixed as the new anchor U′, and the next generation of thresholds is deduced from it. So block ④ below tells you “when to re-anchor, and what the new anchor is”.

How the four quadrants are drawn: the horizontal axis is level (the measurement against the lower rail T⁻), the vertical axis is time (checkpoint progress against 50%) —— 🟢 Early (on-rail and under halfway) · 🔵 On-rail (on-rail and past halfway) · 🟡 Warning (below the lower rail and under halfway; two in a row counts as 🔴) · 🔴 Failed (below the lower rail and past halfway). Also, as soon as any checkpoint is ≥ T⁺ you can start the next chain immediately, without waiting for the timeline to finish.

What is the user's own wording and what is this algorithm's own decision (separated honestly — do not confuse the two)
· From the user's wording (6 points): a main chain carries several small life-chains · each one is determined by its height · the detail level depends on the height · the threshold is fixed from an instance that has already hit the standard (never decided off the top of your head) · positive and negative thresholds run on the top and bottom of the timeline · a suitable quadrant starts the next one.
· Decisions this algorithm makes (4 points, none of them a number from the user's wording): ① the multipliers on the height ladder (1.0/1.3/1.6/2.0/2.5 and 0.8/0.7/0.6/0.5/0.4) — the wording only says “the detail level depends on the height” and gives no number at all; ② the number of checkpoints N = the height — this reads “detail level” as “how many checkpoints”, which is this algorithm's interpretation; ③ the expected line E_k = T⁻ + (T⁺−T⁻) × k/N (a linear climb) — there is no expected line in the wording; this is an auxiliary line added to judge quadrants; ④ the 50% boundary of the four quadrants — the wording only says “the suitable quadrant” and never says how quadrants are drawn.
Items ①② above can be overridden by input fields on the page; ③④ are written into the algorithm.

Unconfirmed (what we are not sure about, flagged here)
· Unconfirmed ①: the NCA effect size d was not checked against the original literature this round (no Crossref lookup, no reading of the source text), so this page only mentions the concept name and does not give its formula or its DOI — writing one out would be making it up.
· Unconfirmed ②: QCA's consistency threshold likewise — nominal comparison only.
· Unconfirmed ③: whether the “height” in the user's text has fixed tiers (whether it really is 1–5) is unconfirmed; this page works with 1–5, and anything outside that range is treated as the default height 3 with a notice.
· Unconfirmed ④: this mechanism has no real-world calibration case at all — the T⁺/T⁻ that comes out of the page is an arithmetic result and does not mean it has been validated on any real business.
· Unconfirmed ⑤: algorithm text / design document: no written original document for this mechanism could be found this round (only the user's spoken account), so this page's conventions come entirely from the user's wording plus this algorithm's own decisions, with no third-party document to cite.

Algorithm file: /_shared/algo/live-threshold.js (pure front-end, zero external dependencies, no CDN, no LLM call). The full algorithm document (including the nine-axis scale upgrade plan): on the algorithm-port machine, /opt/inventory/live_threshold.md; the threshold audit across the whole corpus: /opt/inventory/threshold_audit.md.

What you can do next

No downstream slices yet.

Generated by /opt/slices_build/build.py from the spec /_shared/slices/live-threshold.json — to change this page, edit the spec and re-run the generator; never hand-edit the HTML.