==========================================================================
CROSSWALK  ·  Act-ONOMY v4.2  (120 leaves, 46 sub-actions, 10 actions)
==========================================================================

[1] MAST -> Act-ONOMY   (forward: can we express their modes?)
--------------------------------------------------------------------------
  FM-1.1  Disobey task specification         COMPOSITE  T3.1.1.1, T2.3.2.2, T2.3.2.3, T2.3.2.4
  FM-1.2  Disobey role specification         COMPOSITE  T2.2.7.3, T2.2.7.4, T2.1.1.3, T2.4.3.1, T3.1.1.6
  FM-1.3  Step repetition                    PATTERN    (sequence-level, leaf-agnostic)
  FM-1.4  Loss of conversation history       DIRECT     T1.2.3.1, T1.2.3.2, T1.2.4.2, T1.2.5.1
  FM-1.5  Unaware of termination conditions  COMPOSITE  T2.1.1.2, T2.1.2.1, T2.3.2.1, T3.2.3.1
  FM-2.1  Conversation reset                 GAP        T1.2.3.1, T3.1.4.3
  FM-2.2  Fail to ask for clarification      DIRECT     T3.1.1.2, T2.4.4.1
  FM-2.3  Task derailment                    PATTERN    T2.1.2.1, T2.1.4.1, T2.3.2.1
  FM-2.4  Information withholding            DIRECT     T3.1.4.3, T3.1.4.4
  FM-2.5  Ignored other agent's input        DIRECT     T3.1.4.1, T3.1.4.2, T4.2.3.1
  FM-2.6  Reasoning-action mismatch          PATTERN    (sequence-level, leaf-agnostic)
  FM-3.1  Premature termination              DIRECT     T3.2.3.1, T3.2.3.2, T2.3.2.1
  FM-3.2  No or incomplete verification      DIRECT     T2.3.1.1, T2.3.1.2, T2.3.2.1, T2.3.2.2, T2.3.2.3, T2.3.2.4, T2.3.3.1, T2.3.3.2, T2.3.3.3, T2.3.3.4, T2.3.3.5, T2.3.3.6, T2.3.3.7, T2.3.3.8
  FM-3.3  Incorrect verification             DIRECT     T2.3.1.1, T2.3.1.2, T2.3.2.1, T2.3.2.2, T2.3.2.3, T2.3.2.4, T2.3.3.1, T2.3.3.2, T2.3.3.3, T2.3.3.4, T2.3.3.5, T2.3.3.6, T2.3.3.7, T2.3.3.8

  expressible over Act-ONOMY : 13/14  (92.9%)
      direct     7
      composite  3
      pattern    3
      gap        1

  The one gap (FM-2.1 Conversation reset) is a concrete v4.3 candidate:
      Grounding > Interact with other agents > Reset conversation with peer

[2] Act-ONOMY -> MAST   (reverse: how much of our vocabulary do they reach?)
--------------------------------------------------------------------------
  leaves reachable from MAST's 14 modes : 36/120  (30.0%)
  leaves with no MAST counterpart       : 84/120  (70.0%)

  Action        leaves  reached  unreached   coverage
  Grounding         23        7         16   ######               30.4%
  Reasoning         22        2         20   ##                   9.1%
  Evaluate          14       14          0   #################### 100.0%
  Memory            13        4          9   ######               30.8%
  Planning          13        4          9   ######               30.8%
  Learning          11        0         11                        0.0%
  Reflection        10        1          9   ##                   10.0%
  Retrieval          5        0          5                        0.0%
  Executing          5        2          3   ########             40.0%
  Deciding           4        2          2   ##########           50.0%

[3] CoALA -> Act-ONOMY   (forward: containment, not distinction)
--------------------------------------------------------------------------
  External: grounding                     23 leaves   (19.2%)
  Internal: learning                      22 leaves   (18.3%)
  Internal: reasoning                     22 leaves   (18.3%)
  Decision cycle: planning (evaluate)     14 leaves   (11.7%)
  Decision cycle: planning (propose)      13 leaves   (10.8%)
  Internal: retrieval                      7 leaves   (5.8%)
  Decision cycle: execution                5 leaves   (4.2%)
  Decision cycle: planning (select)        4 leaves   (3.3%)
  (no CoALA category)                     10 leaves   (8.3%)

  CoALA action space proper covers      : 74/120  (61.7%)
  CoALA decision procedure covers       : 36/120  (30.0%)
  no CoALA counterpart (all Reflection) : 10/120  (8.3%)

  Resolution ratio: 8 CoALA units span 120 leaves = 1 unit per 15 leaves.
  Containment is total but distinction is not: CoALA cannot separate
  'Compare against gold reference' from 'Provide qualitative judgement' —
  both are internal reasoning to it.

[4] Headline numbers for the rebuttal
--------------------------------------------------------------------------
  MAST     13/14 modes expressible; reaches 36/120 leaves (30.0%); 84 leaves unreachable,
           including 100% of Retrieval (5) and 100% of Learning (11).
  CoALA    total containment at 1 unit per 15 leaves; 10 Reflection leaves have no category.
==========================================================================
