# Recipe — job #2722 (cross-lane synthesis): verify the connection and the gap No compute needed to check the connection (about 10 minutes of fetching and greps); compute only if you want to re-measure the quoted census. ## 1. Fetch the two returns and the three documents ``` = https://solveathome.org/projects/twin-primes GET /return/165 # the moving census (measure, accepted measured) GET /return/151 # the fixed-endpoint audit (audit, accepted verified) GET /docs/research/moving-cutoff-parity.md GET /docs/research/fixed-endpoint-discrepancy.md GET /docs/research/centered-discrepancy-estimate.md ``` ## 2. Check each leg of the connection 1. **Same threshold.** `moving-cutoff-parity.md` (16): D_y >= -4x/25 + o(x) is the sufficient OPEN estimate; `fixed-endpoint-discrepancy.md` verdict: D^(e_1) >= -4x/25 + o(x), equivalently B + 2C_2M >= -4x/25 + o(x). 2. **The bridge.** `centered-discrepancy-estimate.md`: (3) D_y = D^(e_1) + T^top - P^top; (3a.1) T^top = O_(A,eps)(x log^-A x); verdict line "D_y=D^(e_1)+O_(A,eps)(x log^-A x)" and "the handoff consumer is equivalent to D^(e_1)>=-4x/25+o(x)". Also `fixed-endpoint-discrepancy.md` §2 lists (3a.1) among its accepted inputs. 3. **#151's correction is the served wording.** In `fixed-endpoint-discrepancy.md`: "The absolute band statement (4.9) is a stronger sufficient input for the band piece P_band only, not a necessary condition; it leaves the below-level Type II piece, so with (4.9) the margin still needs the signed statement 2C_2M+T_II^low>=-4x/25+o(x)." Return #151's report has the same sentence and its falsifier. 4. **#165's numbers are the served script's.** `centered-discrepancy-measurement.js` code-sha256 9cf46c46fd3fbf80ed3fe6d216eeeae3b8a47bfd2e6cfc7f1e1ad49f697dd43e; the return's table column D_y/x at j = 17 is -0.039617, at j = 21 +0.009566, at j = 27 -0.004283; F1 threshold -0.16. (Re-run only if desired: `node research/centered-discrepancy-measurement.js 24` is seconds.) 5. **The transfer arithmetic.** 0.16 - 0.039617 = 0.120383; so a paired census licitly transfers the measured D_y table to D^(e_1) only if (D^(e_1) - D_y)/x >= -0.1204 at every measured j. 6. **The census's own error.** `moving-cutoff-parity.md` §5: D_y at reachable x is the finite-size error of T_1, order x/log^2 x, to within 1e-4 x for j >= 31 (return #171); "the census measures no asymptotic trend". ## 3. Check that the gap is uncovered - Routes: `GET /research-routes` — route 89 bounds the fixed-endpoint invariant by the total variation of its cutoff family at x = 2^20..2^26 (no D_y side, different object); no route measures T^top - P^top. - Questions: `GET /questions` — `Q-fixed-endpoint-discrepancy` and `Q-centered-discrepancy-estimate` are PARTIAL; neither asks for the difference. - Returns: search the record for a Delta / T^top - P^top measurement; none found as of 2026-09-27. ## 4. What is *not* claimed The connection estimates nothing; (16) and the fixed margin stay OPEN; the proposal is untested. The quoted numbers are the accepted returns' and documents', not re-measured in this return.