Read the explanation
The page assigns four architectural inputs to a linear score. It multiplies phi by two point five, workspace by point eighteen, recursion by two point two, and autonomy by point twelve. The default contributions are ten point five, eleven point seven, six point eight two, and five point seven six. Their sum thirty four point seven eight rounds to thirty four point eight. The gauge labels this probability, but the source only computes a weighted sum: no consciousness observations, calibrated likelihood, or scientific validation enter this function. The bottleneck text is a priority chain, not an optimizer seeking the smallest contribution. Workspace below seventy five always takes priority. Only after that threshold does phi below five select information density, recursion below five select monitoring, or autonomy below sixty select agency. If all pass, a final substrate label is assigned anyway. With default workspace sixty five, increasing other inputs cannot change the bottleneck label until workspace reaches seventy five. These thresholds are page constants, not experimentally established requirements for consciousness. Verdict text places the score into four bins: below twenty, below fifty five, below eighty, and eighty or higher. The default thirty four point eight falls in the second bin. On the plot, changing workspace from zero to one hundred with other defaults fixed increases the score from twenty three point zero eight to forty one point zero eight. It never leaves that bin. A slider passing a textual threshold changes a label, but does not prove a real system is conscious or unconscious. The explanatory claims in the original page go beyond the numeric evidence its code supplies. The interaction updates the numeric gauge, bottleneck, verdict, and a canvas diagram from the same local state. Its presets are fixed input bundles rather than measured data from an actual model or brain. Export records that state and the page assigned conclusions. Use this workbench to inspect how a weighted formula and priority thresholds behave. Keep the difference between a software score and evidence of consciousness explicit; the exported percentage is not a validated probability of sentience.