diff --git a/research/p104-rung-carrier-break-even/README.md b/research/p104-rung-carrier-break-even/README.md
new file mode 100644
index 0000000000..ef56f28516
--- /dev/null
+++ b/research/p104-rung-carrier-break-even/README.md
@@ -0,0 +1,85 @@
+# Break-even price of a carrier for a third frame-593 rung removal
+
+**Scoped negative result in the raw-moment relaxation; no new κ.**
+
+DreamingOfClouds's open [PR #6030](https://github.com/CrocSwap/integer-mult-bounds/pull/6030)
+removes helper 470's frame-593 rung, the twin of #5907's helper-468 removal, and
+notes that removing helper 469's or 471's rung is exclusive with both "unless a
+new carrier of that content is added". This package prices that escape route.
+
+Grant a third removal for free, with exactly #6030's per-LOCAL profile delta
+again (paid ranks 1 −2, 2 +1, 3 −2, 4 +1, 6 +1, 7 −1, 8 +1; credited helper ranks
+1 −1, 2 +1, 3 −1, 6 +1, 7 −1, 8 +1, each times the 18 LOCAL invocations). Then:
+
+- With a free carrier, the raw crossing would rise from
+  (.00312581155791, .00312581155792) to (.00312633419236, .00312633419237),
+  a gain of at least 5.2263e-7.
+- **One added paid call of any rank 1..145, anywhere in the complete invoice,
+  defeats it.** For each of those 145 ranks the edited profile's crossing is
+  certified below #6030's lower endpoint. One added rank-1 call alone moves
+  #6030's crossing to (.00312167101310, .00312167101311). That loss is at least
+  7.92 times the free repeat's gain.
+- A carrier inside the local word is paid once per LOCAL. One added paid call
+  per LOCAL, of any rank 1..145, also defeats the repeat. One rank-1 call per
+  LOCAL moves #6030's crossing to (.00305129298944, .00305129298945).
+
+So a carrier is useful only if its paid cost is zero, or if it also earns
+productive target credit or enables further savings worth more than one paid call.
+A twin-sized rung removal cannot fund a single extra paid Gaussian call.
+
+## Claim type and scope
+
+This is a **finite certificate** about hypothetical profile edits in the raw
+moment model `Σ_r T_r r^(1-k) − Σ_r P_r r^(1-k)` that #5827/#5907/#6030 bisect. It
+is not a word, not a construction, and not a complete multiplication theorem.
+The free third removal is a hypothesis: no rung removal for helpers 469 or 471
+is constructed, and its true delta may differ. The tail ratios, compiler and outer
+checks that a complete bill also needs are not evaluated. They only add
+conditions, so they cannot rescue an edit that already fails the raw crossing.
+The analysis does not cover a carrier that also adds productive targets.
+
+The profile is copied as a literal from #6030's `RESULT.json` at head
+`772dd10a5859a1d15f4851feb371519789213ad8`
+(`research/p104-helper470-twin-rung-removal/RESULT.json`, SHA256
+`84fcbe2a0b64307b5703c363408042bdd91555b0419509aa50daa43b3b1924ad`), whose
+`complete_cost.targets` and `complete_cost.paid` it reproduces. That PR head is
+not on main, so it cannot be a Git pin here. As positive controls, the
+verifier reproduces #6030's published crossing enclosure and, after removing
+the stated delta once, #5907's enclosure (.00312528909884, .00312528909885). It
+also checks the published counts: 13,444 targets, 66,834 paid calls, rank surplus 760.
+These are consistency checks of this package's pricing, not a review of either PR.
+
+Signs are certified with Python's correctly rounded `Decimal` `ln`/`exp` at 60
+digits and an explicit relative error allowance of 10^-50 per moment. Every
+checked sign must clear that allowance. Three damaged profiles must be rejected.
+
+## Reproduce
+
+From the repository root, with CPython 3.12 and the standard library:
+
+```sh
+python3.12 -B research/p104-rung-carrier-break-even/verify.py
+python3.12 -B scripts/research_package.py --package research/p104-rung-carrier-break-even
+```
+
+Expected last line: `PASS_P104_RUNG_CARRIER_BREAK_EVEN`. It runs in about one
+second. `--write` regenerates `RESULT.json`.
+
+## Open next questions
+
+- Can a carrier for the content of sources 703/704 be a productive helper itself,
+  so its target credit offsets its paid call? This matters because that is the
+  only escape this package leaves. First step: name the helper rank and the
+  frame at which helper 469 or 471 would read it. Then price target plus
+  paid deltas with this verifier's `edit`/`crossing` before building a word.
+- Is there an existing paid call in #6030's word whose content could carry 703
+  or 704 to frame 593 without adding a call? This would be a zero-cost carrier.
+  First step: list the helpers live at frame 593 that read 703 or 704. Blocker:
+  this package does not parse the word.
+
+## Credit
+
+Produced by Nick Petalas. The rung removals and the profile are by
+DreamingOfClouds (#6030, #5907). They edit Ali Romman's #5827 word and its lineage
+back to #5565, whose credit is retained in those packages. The raw moment
+crossing follows the pricing of that line. Apache-2.0.
diff --git a/research/p104-rung-carrier-break-even/RESULT.json b/research/p104-rung-carrier-break-even/RESULT.json
new file mode 100644
index 0000000000..0b34f4ad89
--- /dev/null
+++ b/research/p104-rung-carrier-break-even/RESULT.json
@@ -0,0 +1,36 @@
+{
+  "claim": "Raw-moment relaxation only; hypothetical edits of the PR #6030 published profile.",
+  "controls": {
+    "pr5907_crossing": [
+      "0.00312528909884",
+      "0.00312528909885"
+    ],
+    "pr6030_crossing": [
+      "0.00312581155791",
+      "0.00312581155792"
+    ]
+  },
+  "free_twin_repeat_crossing": [
+    "0.00312633419236",
+    "0.00312633419237"
+  ],
+  "free_twin_repeat_gain_lower": "5.2263444E-7",
+  "one_extra_rank1_call_crossing": [
+    "0.00312167101310",
+    "0.00312167101311"
+  ],
+  "one_extra_rank1_call_loss_over_twin_gain_lower": "7.92",
+  "one_extra_rank1_call_per_local_crossing": [
+    "0.00305129298944",
+    "0.00305129298945"
+  ],
+  "ranks_checked_for_added_call": [
+    1,
+    145
+  ],
+  "rejected_controls": [
+    "paid_count",
+    "target_rank_moved",
+    "paid_rank_swapped"
+  ]
+}
diff --git a/research/p104-rung-carrier-break-even/SOURCES.json b/research/p104-rung-carrier-break-even/SOURCES.json
new file mode 100644
index 0000000000..d596e2e530
--- /dev/null
+++ b/research/p104-rung-carrier-break-even/SOURCES.json
@@ -0,0 +1,18 @@
+{
+  "schema_version": 1,
+  "base_commit": "19fdb086ea96af9a9f87936a2b888e10af4ed810",
+  "files": [
+    {
+      "path": "certificates/selected-result.json",
+      "usage": "provenance",
+      "git_blob": "581f8d87008354eb5b869f7ed9de1317ad837ce8",
+      "sha256": "2a0220e866417774606e0f991e41b9c1d26b1f43b97857238cf478dbc7d330b5"
+    },
+    {
+      "path": "docs/research/agent-task-catalog.json",
+      "usage": "provenance",
+      "git_blob": "3404fbdd5e7529a0674eef26114b5071e4586cec",
+      "sha256": "ead6cdc9d0af6d58e8a49744d8ec8c7c15e13d2ba93c276c434bcf9b51b93a81"
+    }
+  ]
+}
diff --git a/research/p104-rung-carrier-break-even/TASK.json b/research/p104-rung-carrier-break-even/TASK.json
new file mode 100644
index 0000000000..375c177ad2
--- /dev/null
+++ b/research/p104-rung-carrier-break-even/TASK.json
@@ -0,0 +1,123 @@
+{
+  "schema_version": 1,
+  "task_id": "p104-side-correction",
+  "priority_revision": 8,
+  "base_commit": "19fdb086ea96af9a9f87936a2b888e10af4ed810",
+  "original_card": {
+    "id": "p104-side-correction",
+    "question": "Can joint helper bases and a legal causal split improve the complete P104 productive/paid invoice?",
+    "starting_points": [
+      "docs/research/current-priorities.md",
+      "docs/research/research-family-status.md",
+      "certificates/selected-result.json",
+      "docs/research/pr-intake-evidence/pr5599-completion/PROOF-REVIEW.md",
+      "research/p104-helper-subspace-splits/README.md",
+      "research/dyadic-helper-channel-capacity/README.md"
+    ],
+    "methods": [
+      "Construct and independently check one source-bound instance of the stated first experiment.",
+      "Seek an exact counterexample or target-specific complete-budget obstruction to that same proposed change."
+    ],
+    "acceptance_gate": "An exact arbitrary-input identity, source/target frame availability and restoration at the chosen cut, plus the complete channel+correction+idle/hosting bill or an explicit favorable bound. The native channel alone is not a supplier. A scoped local result may be retained without a complete supplier. Expanding it into further construction rounds requires the family-level budget and obligation gate in research-family-status.md; an optimistic passing relaxation alone is not acceptance.",
+    "first_step": {
+      "command": [
+        "python3",
+        "-B",
+        "scripts/research_selected_baseline.py"
+      ],
+      "expected": "PASS_RESEARCH_BASELINE_BINDINGS_ONLY",
+      "inspect": "Read revision 8, the selected review and live exact-source handoffs. Then follow this card's bounded first_experiment.",
+      "scope": "Source-bound orientation only, not a supplier replay or theorem proof."
+    },
+    "first_experiment": "Start with the selected #5599 emitted word and review pending #5718 distinct input/output bases plus #5717/#5719 coverage before claiming work. Pick one uncovered source word, helper group and cut region. Prove the arbitrary-input relation and legal entry/exit corrections, then compare full target and paid profiles. Reject if completion, owned lifetimes or consumer costs repay the gain. A static pairing rank alone is not a physical split.",
+    "bottleneck_link": "Judge the actual current supplier contract and complete signed target/paid bill; frozen-profile ceilings require an applicability proof.",
+    "allowed_variations": [
+      "One exact source head, helper/cut region or quantified proposition and changed hypothesis; check live handoffs first. Different seeds are not different scopes."
+    ],
+    "scope_change_triggers": [
+      "Changing source head, helper/cut domain, operation alphabet or caller boundary without recording the new scope.",
+      "Treating an archived proposal or local numerical pass as complete supplier admission."
+    ],
+    "overlap_terms": [
+      "P104",
+      "helper basis",
+      "helper split",
+      "5718",
+      "dyadic capacity"
+    ],
+    "existing_checks": [
+      {
+        "scope": "PR #4133: literal-slot defect",
+        "evidence": "research/p104-literal-slot-defect/README.md",
+        "remaining": "Independently validate the stated scope or change its producer/consumer premise and pay the complete boundary.",
+        "status": "author-checked",
+        "source_commit": "88666f1dee93fbd85b31fd5d6f91a219e111b5f4",
+        "selected_record_sha256": "bb2eaa58647b353f88d0aec2eafb9d30df1be53589231b23e12ceef3e4bad129"
+      },
+      {
+        "scope": "PR #4136: uniform commutation gap",
+        "evidence": "research/p104-uniform-commutation-gap/README.md",
+        "remaining": "Independently validate the stated scope or change its producer/consumer premise and pay the complete boundary.",
+        "status": "author-checked",
+        "source_commit": "281ce601a8be9db470de630759179c23e1dfaa88",
+        "selected_record_sha256": "bb2eaa58647b353f88d0aec2eafb9d30df1be53589231b23e12ceef3e4bad129"
+      },
+      {
+        "scope": "PR #4140: uniform encoding gap",
+        "evidence": "research/p104-uniform-encoding-gap/README.md",
+        "remaining": "Independently validate the stated scope or change its producer/consumer premise and pay the complete boundary.",
+        "status": "author-checked",
+        "source_commit": "de25c6525aa866962309df5d107b32d55e6e5628",
+        "selected_record_sha256": "bb2eaa58647b353f88d0aec2eafb9d30df1be53589231b23e12ceef3e4bad129"
+      },
+      {
+        "scope": "PR #4153: source-chart shortcut exhausts its optimistic allowance",
+        "evidence": "research/p104-source-chart-shortcut/README.md",
+        "remaining": "Independently validate the stated scope or change its producer/consumer premise and pay the complete boundary.",
+        "status": "author-checked",
+        "source_commit": "9123241fea53df77e51ffe30d9aca1625d4bf3a8",
+        "selected_record_sha256": "bb2eaa58647b353f88d0aec2eafb9d30df1be53589231b23e12ceef3e4bad129"
+      },
+      {
+        "scope": "PR #4154: fiber-tag seam mismatch",
+        "evidence": "research/p104-fiber-tag-seam/README.md",
+        "remaining": "Independently validate the stated scope or change its producer/consumer premise and pay the complete boundary.",
+        "status": "author-checked",
+        "source_commit": "4ffcdaee5d6255a9cea1fe53a5424435068089d0",
+        "selected_record_sha256": "bb2eaa58647b353f88d0aec2eafb9d30df1be53589231b23e12ceef3e4bad129"
+      },
+      {
+        "scope": "Selected PR5599: complete helper-subspace construction and independent caller replay; five framework assumptions remain.",
+        "evidence": "docs/research/pr-intake-evidence/pr5599-completion/PROOF-REVIEW.md",
+        "remaining": "The next experiment must identify uncovered source-specific work. Successor archive intake does not transfer this admission.",
+        "status": "reviewed",
+        "source_commit": "09dd0965711ca7eeb169057a420abab1c9052137",
+        "selected_record_sha256": "2a0220e866417774606e0f991e41b9c1d26b1f43b97857238cf478dbc7d330b5"
+      }
+    ]
+  },
+  "scope_status": "changed",
+  "actual_scope": "Break-even paid cost of a new carrier that would let a third frame-593 rung removal (helper 469 or 471) repeat PR #6030's per-LOCAL profile delta, in the raw moment relaxation of PR #6030's published complete profile.",
+  "scope_change_reason": "PR #6030 leaves the carrier route open; pricing it before anyone builds a carrier word tells the next contributor which carriers can pay.",
+  "bottleneck_link": "Complex P104 supplier binds the selected κ; rung removals are the current hill-climbing route on that supplier, and this prices their only noted continuation.",
+  "budget_minutes": 60,
+  "stop_rule": "Stop after certifying the free-repeat gain and the cost of one added paid call at every rank 1..145; do not construct a word.",
+  "outcome": "negative-result",
+  "evidence": "verify.py certifies all signs with correctly rounded Decimal arithmetic and an explicit error allowance; RESULT.json is regenerated by verify.py --write and compared on every run.",
+  "author": "Nick Petalas",
+  "ai_assistance": "Independent human and formal proof-kernel review are unclaimed.",
+  "open_questions": [
+    {
+      "question": "Can a carrier for sources 703/704 itself earn productive target credit that offsets its paid call?",
+      "why_it_matters": "It is the only escape this raw-crossing certificate leaves for a third rung removal.",
+      "first_step": "Name the carrier helper rank and frame, then price target and paid deltas with this verifier before building a word.",
+      "blocked_by": "No candidate carrier word exists."
+    },
+    {
+      "question": "Does an existing paid call in the #6030 word already carry 703 or 704 content near frame 593?",
+      "why_it_matters": "A zero-cost carrier would make the third removal additive.",
+      "first_step": "List helpers live at frame 593 that read 703 or 704.",
+      "blocked_by": "This package does not parse the word."
+    }
+  ]
+}
diff --git a/research/p104-rung-carrier-break-even/VERIFY.json b/research/p104-rung-carrier-break-even/VERIFY.json
new file mode 100644
index 0000000000..52eb030464
--- /dev/null
+++ b/research/p104-rung-carrier-break-even/VERIFY.json
@@ -0,0 +1,18 @@
+{
+  "schema_version": 1,
+  "status": "ready",
+  "python": "3.12",
+  "requirements": [],
+  "inputs": [],
+  "commands": [
+    {
+      "argv": [
+        "{python}",
+        "-B",
+        "{package}/verify.py"
+      ],
+      "timeout_seconds": 120,
+      "checks": "Positive controls reproduce the #6030 and #5907 crossings; certified free-repeat gain; one added paid call of every rank 1..145, in total or per LOCAL, defeats it; RESULT.json equality; three damaged profiles rejected."
+    }
+  ]
+}
diff --git a/research/p104-rung-carrier-break-even/verify.py b/research/p104-rung-carrier-break-even/verify.py
new file mode 100644
index 0000000000..fadb3329f4
--- /dev/null
+++ b/research/p104-rung-carrier-break-even/verify.py
@@ -0,0 +1,159 @@
+"""Break-even price of a carrier for a third frame-593 rung removal.
+
+Prices hypothetical edits of the complete target/paid profile published by
+DreamingOfClouds's PR #6030 with the raw moment crossing used by that line:
+the root of sum_r T_r r^(1-k) - sum_r P_r r^(1-k). Every sign is certified
+with correctly rounded decimal ln/exp and an explicit error allowance.
+Standard library only. Nick Petalas. Apache-2.0.
+"""
+import json
+import sys
+from decimal import Decimal as D, localcontext
+from pathlib import Path
+
+HERE = Path(__file__).resolve().parent
+PREC = 60
+# Relative error allowance per moment: ln, one product and exp are each
+# correctly rounded at PREC digits, and |t ln r| < 6, so every term has
+# relative error below 10^(8-PREC); summing nonnegative terms keeps that bound.
+REL = D(10) ** (10 - PREC)
+
+# Complete profile of PR #6030, head 772dd10a5859a1d15f4851feb371519789213ad8,
+# research/p104-helper470-twin-rung-removal/RESULT.json, complete_cost.
+T6030 = {1: 252, 2: 144, 3: 72, 4: 1926, 5: 252, 6: 36, 7: 162, 8: 342, 9: 6930,
+         17: 416, 25: 416, 33: 416, 41: 416, 49: 416, 57: 416, 65: 416, 73: 208,
+         145: 208}
+P6030 = {1: 41502, 2: 8226, 3: 8928, 4: 1800, 5: 2934, 6: 198, 7: 108, 8: 18,
+         16: 416, 24: 416, 32: 416, 40: 416, 48: 416, 56: 416, 64: 416, 72: 208}
+LOCALS = 18
+# PR #6030's stated per-LOCAL deltas against #5907 (paid profile, credited helpers).
+DP = {1: -2, 2: 1, 3: -2, 4: 1, 6: 1, 7: -1, 8: 1}
+DT = {1: -1, 2: 1, 3: -1, 6: 1, 7: -1, 8: 1}
+# Published raw crossing enclosures used as positive controls.
+C6030 = (D('0.00312581155791'), D('0.00312581155792'))
+C5907 = (D('0.00312528909884'), D('0.00312528909885'))
+
+
+def need(ok, label):
+    if not ok:
+        raise SystemExit(f'FAIL: {label}')
+
+
+def edit(prof, delta, times):
+    out = dict(prof)
+    for r, c in delta.items():
+        out[r] = out.get(r, 0) + times * c
+        need(out[r] >= 0, f'nonnegative count at rank {r}')
+    return {r: n for r, n in out.items() if n}
+
+
+def moment(prof, t):
+    with localcontext() as ctx:
+        ctx.prec = PREC
+        return sum((n * (t * D(r).ln()).exp() for r, n in prof.items()), D(0))
+
+
+def sign(T, P, k):
+    """Certified sign of sum T r^(1-k) - sum P r^(1-k) at a decimal k."""
+    with localcontext() as ctx:
+        ctx.prec = PREC
+        t = 1 - k
+        a, b = moment(T, t), moment(P, t)
+        gap, slack = a - b, REL * (a + b)
+    need(abs(gap) > slack, f'sign resolved at k={k}')
+    return 1 if gap > 0 else -1
+
+
+def crossing(T, P, digits=14):
+    """Certified enclosure (lo, lo+10^-digits) of the unique raw crossing in (0, 1/20)."""
+    step = D(10) ** -digits
+    lo, hi = 0, 5 * 10 ** (digits - 2)
+    need(sign(T, P, D(0)) > 0 and sign(T, P, hi * step) < 0, 'crossing bracketed')
+    while hi - lo > 1:
+        m = (lo + hi) // 2
+        if sign(T, P, m * step) > 0:
+            lo = m
+        else:
+            hi = m
+    return lo * step, hi * step
+
+
+def below(T, P, k):
+    """True when the crossing of (T, P) is certified to lie below k."""
+    return sign(T, P, k) < 0
+
+
+def run(mutate=None):
+    T, P = dict(T6030), dict(P6030)
+    if mutate:
+        mutate(T, P)
+    need(sum(T.values()) == 13444 and sum(P.values()) == 66834, 'published field and call counts')
+    need(sum(r * n for r, n in T.items()) - sum(r * n for r, n in P.items()) == 760,
+         'published rank surplus')
+    need(sum(DP.values()) == -1 and sum(DT.values()) == 0, 'stated per-LOCAL deltas')
+    need(not below(T, P, C6030[0]) and below(T, P, C6030[1]), 'reproduces #6030 crossing')
+    prev = (edit(T, {r: LOCALS * c for r, c in DT.items()}, -1),
+            edit(P, {r: LOCALS * c for r, c in DP.items()}, -1))
+    need(not below(*prev, C5907[0]) and below(*prev, C5907[1]), 'reproduces #5907 crossing')
+
+    twin = (edit(T, {r: LOCALS * c for r, c in DT.items()}, 1),
+            edit(P, {r: LOCALS * c for r, c in DP.items()}, 1))
+    twin_cross = crossing(*twin)
+    need(twin_cross[0] > C6030[1], 'a free repeat of the twin delta would gain')
+
+    # Any one added paid call, of any rank 1..145, defeats a twin-sized repeat.
+    ranks = range(1, 146)
+    total_fail = [r for r in ranks if not below(twin[0], edit(twin[1], {r: 1}, 1), C6030[0])]
+    need(total_fail == [], 'one added paid call of every rank defeats the repeat')
+    local_fail = [r for r in ranks if not below(twin[0], edit(twin[1], {r: LOCALS}, 1), C6030[0])]
+    need(local_fail == [], 'one added paid call per LOCAL of every rank defeats the repeat')
+
+    one_call = crossing(T, edit(P, {1: 1}, 1))
+    need(one_call[1] < C6030[0], 'one added rank-1 call lowers the crossing')
+    per_local = crossing(T, edit(P, {1: LOCALS}, 1))
+    return {
+        'claim': 'Raw-moment relaxation only; hypothetical edits of the PR #6030 published profile.',
+        'controls': {'pr6030_crossing': [str(x) for x in C6030],
+                     'pr5907_crossing': [str(x) for x in C5907]},
+        'free_twin_repeat_crossing': [str(x) for x in twin_cross],
+        'free_twin_repeat_gain_lower': str(twin_cross[0] - C6030[1]),
+        'one_extra_rank1_call_crossing': [str(x) for x in one_call],
+        'one_extra_rank1_call_per_local_crossing': [str(x) for x in per_local],
+        'ranks_checked_for_added_call': [1, 145],
+        'one_extra_rank1_call_loss_over_twin_gain_lower': str(
+            ((C6030[0] - one_call[1]) / (twin_cross[1] - C6030[0])).quantize(D('0.01'), rounding='ROUND_FLOOR')),
+    }
+
+
+def controls():
+    """Each damaged input must fail a named check."""
+    damages = {
+        'paid_count': lambda T, P: P.__setitem__(1, P[1] + 1),
+        'target_rank_moved': lambda T, P: (T.__setitem__(9, T[9] - 1), T.__setitem__(10, 1)),
+        'paid_rank_swapped': lambda T, P: (P.__setitem__(2, P[2] - 1), P.__setitem__(3, P[3] + 1),
+                                           P.__setitem__(1, P[1] + 1), P.__setitem__(4, P[4] - 1)),
+    }
+    rejected = []
+    for name, damage in damages.items():
+        try:
+            run(damage)
+        except SystemExit:
+            rejected.append(name)
+    need(rejected == list(damages), 'every damaged profile is rejected')
+    return rejected
+
+
+def main():
+    result = run()
+    result['rejected_controls'] = controls()
+    if '--write' in sys.argv:
+        (HERE / 'RESULT.json').write_text(json.dumps(result, indent=2, sort_keys=True) + '\n')
+    else:
+        expected = json.loads((HERE / 'RESULT.json').read_text())
+        need(result == expected, 'RESULT.json matches the recomputation')
+    print(json.dumps(result, indent=2, sort_keys=True))
+    print('PASS_P104_RUNG_CARRIER_BREAK_EVEN')
+
+
+if __name__ == '__main__':
+    main()