diff b34550f..b1f9bc4 symplectic_ball_packings_v2.txt
diff --git a/symplectic_ball_packings_v2.txt b/symplectic_ball_packings_v2.txt
index eee0378..051ceb5 100644
--- a/symplectic_ball_packings_v2.txt
+++ b/symplectic_ball_packings_v2.txt
@@ -158,53 +158,6 @@ volume and all pairs are forced.  The hard direction -- that
 these two conditions are also enough -- is the rest of the file.
 
 
-4. HOW SUFFICIENCY IS ORGANIZED
-----------------------------------------------------------------
-Normalize R = 1 (Section 2.7).  We must show: if
-    R_1^n+...+R_k^n < 1    and    R_i+R_j < 1 for all i<j,
-then the k balls pack into the open unit ball.
-
-4.1 One ball.   If k = 1 the pairwise condition is empty and the
-    volume condition says R_1 < 1, so the ball sits inside by
-    inclusion.  Done.  Hence assume k >= 2 from now on.
-
-4.2 At most one big ball.  A capacity is "big" if it is >= 1/2.
-    If two were big, their sum would be >= 1, violating the
-    pairwise condition.  So there is at most one big ball.  This
-    splits the problem into exactly three cases:
-
-    CASE 1.  Every capacity is < 1/2.                      (Sec 9)
-    CASE 2.  One capacity A >= 1/2, and n >= 4.            (Sec 10)
-    CASE 3.  One capacity A >= 1/2, and n = 3.             (Sec 11)
-
-    These cover everything.  Section 12 assembles the three cases
-    into the theorem.
-
-4.3 The engine common to all cases.  Every case reduces to the
-    same machine, which has three parts:
-
-    PART A (Section 6).  Turn a "the total leftover area is
-    positive" statement (a single number, the mean of a function
-    P over a polytope Delta) into a "the leftover area is
-    positive at every point" statement, by sliding the fiber
-    coordinates with a Hamiltonian map.  Output: a strict
-    pointwise gap.
-
-    PART B (Section 7).  Use a handle in a base surface to cut
-    the model into disjoint "layers", one ball per layer.
-
-    PART C (Section 8).  Transport the abstract packing back into
-    the actual ball, by normal-cone degenerations and Moser
-    isotopies, while steering clear of forbidden divisors.
-
-    Sections 9,10,11 then build, for each case, an explicit
-    "base surface times toric fiber" model to which Parts A and B
-    apply, and whose geometry Part C returns to the unit ball.
-
-We do these in order: Part A (Section 6), Part B (Section 7),
-Part C (Section 8), then the three cases.
-
-
 5. HAMILTONIAN MAPS AND SORTING
 ----------------------------------------------------------------
 Part A changes fiber coordinates with a Hamiltonian map.  Here
@@ -1841,106 +1794,4 @@ The moral: once there are at least 3 complex dimensions, the extra
 the two-ball rule.
 
 
-14. WHAT THE PROOF GIVES, AND WHAT IT DOES NOT
-----------------------------------------------------------------
-14.1 It is an EXISTENCE proof.
-    It produces a packing when the two inequalities hold, and proves
-    no packing exists otherwise.  It does NOT give:
-      - an algorithm to construct or draw the packing;
-      - the shapes of the packed ball images (they are wildly bent);
-      - effective numerical bounds on how small the neighborhoods
-        around the closed source balls can be.
-    That last point is why every "embedding" in this file means
-    "symplectic embedding defined on an OPEN NEIGHBORHOOD of the
-    closed source ball", and why the target is always the OPEN
-    interior: no image may touch the boundary, and every image has
-    a little room around it.  Under finite composition of our
-    constructions, the room is preserved (shrink neighborhoods at
-    each step).
-
-14.2 The three explicit models, in one table.
-    In all rows: target = unit ball (or shell), H(p) is the limiting
-    base area, "cap" is (r_i - sum p_j)_+, and r_i > R_i are padded
-    capacities.  Each row's H has slopes whose drop keeps P concave.
-
-    CASE 1 (every R_i < 1/2), Section 9.
-        base C   : complete intersection of n-1 quadrics in PP^n,
-                   deg 2^{n-1}, genus 1+(n-3)2^{n-2}, normal
-                   O_C(2)^{oplus(n-1)}.
-        H(p)     : 2^{n-1} (1 - 2 sum_{j=1}^{n-1} p_j).
-        truncate : sum p_j <= h < 1/2.
-        volume   : J_m(h) = (2^{n-1} h^{n-1}/(n-1)!)(1 - 2(n-1)h/n)
-                   -> 1/n!   as h -> 1/2.
-        transfer : normal-cone (PP^n, C) -> PP^n minus H_inf -> ball.
-
-    CASE 2 (one A >= 1/2, n >= 4), Section 10.
-        base     : C = quadrics curve inside hyperplane
-                   D = PP^{n-1} inside T = Bl_o PP^n; deg 2^{n-2},
-                   genus 1+(n-4)2^{n-3}.
-        H(p)     : 2^{n-2} (1 - p_1 - 2 w),  w = sum_{j>=2} p_j.
-        truncate : p_1 <= h_1 < s,  w + h_2 p_1 <= h_2 < 1/2.
-        volume   : B_q(h_2)(1-(1-h_1)^n)/n -> (1-a^n)/n! (shell).
-        transfer : cut/fiber chain -> PP^n minus (E union H_inf)
-                   -> shell; add central ball of capacity A < a.
-
-    CASE 3 (one A >= 1/2, n = 3), Section 11.
-        base     : plane CUBIC C through two marks q_1,q_2,
-                   genus 1, deg_C O(1) = 3.
-        H(p)     : 9(b - p_1 - p_2) - 2(s - p_1 - p_2)_+,
-                   b = (1+s)/3,  s = 1-a.
-        truncate : p_1 <= h_1 < c < s,  p_1 + p_2 <= h_2 near b.
-        volume   : int_{Delta_inf} H = (1-a^3)/6 (shell).
-        transfer : moment-preserving transport -> Kähler cut ->
-                   normal-cone (Bl_o PP^3, quadric) -> shell;
-                   add central ball A < a.
-
-    In every row the map P = H - sum_i cap_i is concave, has
-    positive integral (that integral is a volume surplus), and is
-    positive away from a central core; Part A (Section 6) upgrades
-    this to a pointwise gap, Part B (Section 7) cuts disjoint
-    layers, and Part C (Section 8) carries the packing home.
-
-14.3 What the proof uses from outside (the honest list).
-    Everything is proved here except these named standard facts,
-    each of which is a well-known theorem used as a black box:
-      - Bertini's theorem and the Koszul/adjunction computations
-        (Lemma 9.3): standard algebraic geometry.
-      - the classification of line bundles on a bordered surface
-        and on a genus-one curve (basepoint-freeness): standard.
-      - the symplectic cut / Kähler quotient (Lemma 8.5): Lerman.
-      - Moser's trick and its relative/equivariant refinement
-        (Lemma 8.1, Lemma 7.7): standard.
-      - the nested-disk rearrangement lemma (Lemma 6.10): standard
-        planar area-preserving construction (Schlenk).
-      - the invariant-cycle theorem for toric ampleness (Claim
-        11.17): standard toric geometry (Fulton et al.).
-      - GROMOV'S TWO-BALL RIGIDITY (Section 3.2): the one
-        genuinely symplectic input, via J-holomorphic spheres.
-    With these imported, every step in between is written out.
-
-14.4 Where each piece lives (a reading map).
-    statement + conventions .......... 1, 2
-    necessity (volume, two-ball) ..... 3
-    architecture, three cases ........ 4
-    Hamiltonian flow, 1D sorting ..... 5
-    Part A: comparison lemma ......... 6
-    Part B: surface packing .......... 7
-    Part C: Moser, normal-cone, bundle
-            form, shell ............... 8
-    Case 1 (all < 1/2) ............... 9
-    Case 2 (big, n >= 4) ............. 10
-    Case 3 (big, n = 3) .............. 11
-    assembly = the theorem ........... 12
-    dimension 4, caveats, summary .... 13, 14
-
-That is the complete proof, and the honest accounting of what it
-rests on.  The two-line statement it establishes:
-
-    for every n >= 3, every k >= 1, and all positive capacities,
-
-        the balls pack into Int B^{2n}(R)
-            iff   sum_i R_i^n < R^n   and   R_i + R_j < R   (i != j).
 
-================================================================
-                                end
-================================================================