Dear Internet Area,
Cross-posting to int-area, architecture-discuss, IPPM, and irtf-discuss
because the work below spans multiple WGs and an IAB-level architectural
question. Apologies for the breadth -- a single thread on int-area is
the only place all four audiences read together.
Over the last two weeks the MVPS family of individual Internet-Drafts
closed at NINE documents (all now on Datatracker): seven concrete
instantiations and two architectural capstones. Several of you have
privately said the collection is hard to navigate. This message is
the one-page map.
No WG action is requested. All nine are individual submissions.
----------------------------------------------------------------------
The one-line statement
----------------------------------------------------------------------
MVPS = (V, B, (C, H), D^2, Pub) on a bounded simplex.
For any architecture A satisfying axioms MVPS-A1..A5 (D-8),
the reactive latency floor at planetary scale is
R* = max{ tau_causal, tau_sampling, tau_information,
tau_consensus, tau_coupling } (D-9)
and MVPS is the first architecture in which max{...} reduces
to tau_causal (the speed of light).
At antipodal distance: classical-Internet floor ~ 60-300 s
(BGP-keepalive- and BGP-convergence-bound, RFC 4271 Section 10).
MVPS floor ~ 145-196 ms (causality-bound). Speedup ~ 1220x.
----------------------------------------------------------------------
The nine drafts (read in this order if new)
----------------------------------------------------------------------
CAPSTONES (read first if you only want the architecture)
--------------------------------------------------------
D-8. draft-melegassi-iab-mvps-architecture-00 [IAB / IRTF]
"MVPS Architecture: Specification Conformance for the
Multi-Vantage Path-Coherence Drafts"
Defines MVPS as a 5-tuple with five axioms (MVPS-A1..A5).
Invariance Theorem: any architecture satisfying A1..A5
inherits Theorems 1, 2, 3, 3', 4, 5, 9 of the v4.0 catalogue
plus L_DL plus Stein's Lemma under A4 by 10 mechanical
substitution steps -- no new mathematics introduced.
D-1..D-7 below are proved conformant; BGP-4 [RFC 4271],
BFD [RFC 5880], DNS [RFC 1034/2181], and TCP RTX [RFC 6298]
are catalogued as STRUCTURALLY non-conformant, and the
specific axiom each violates is precisely the tau_sampling
floor that D-9 computes for it.
D-9. draft-melegassi-iab-mvps-planetary-floor-00 [IAB / IRTF]
"Planetary Coherence Floor: Composition Theorem for
Reactive Latency in Multi-Vantage Network Infrastructure"
States and proves PCF (above). Each of the five floors
reduces by finite chain to v4.0 results plus RFCs:
tau_causal from special relativity + n_fiber = 1.467
(ITU-T G.652); tau_sampling from RFC 4271 Section 10
(BGP keepalive 30 s; convergence 60-300 s), RFC 5880
Section 6.8.1 (BFD M * MinTx), RFC 2181 (DNS TTL_min),
RFC 6298 Section 2.4 (TCP RTO_min = 1 s);
tau_information from Stein's Lemma under A4;
tau_consensus from Theorem 9 geometric-median bias;
tau_coupling from Pub_A subscriber-arrival envelope.
Sharpness Corollary PCF.1 (each floor attained by an
explicit construction) and Falsification Corollary PCF.2
(the exact empirical signature whose observation would
refute the theorem) are both stated. Receipt
scripts/validate_planetary_floor.py exit 0 (PASS).
SEVEN INSTANTIATIONS (the proofs that D-8's axioms are non-vacuous)
-------------------------------------------------------------------
D-1. draft-melegassi-ippm-mvps-bundle-00 [IPPM]
Bundle envelope and coherence algebra C := (C_1, C_2, C_3) in
[0,1]^3. Mahalanobis decision D^2 with FAR control. The
canonical instantiation; everything else specialises it.
D-2. draft-melegassi-mvps-incremental-be-00 [OPSAWG]
Bandwidth-Efficient MVPS. Sherman-Morrison-Woodbury + CRDT
cells. T_BE Pareto: ~25x bandwidth down at ~2x CPU up.
D-3. draft-melegassi-coherence-bfd-00 [BFD]
Coherence-BFD. RFC 5880 extension, 5-state machine, 10 TLVs,
sub-second coherence detection. Demonstrates that A2 (bounded
triple) can ride on existing BFD wire format.
D-4. draft-melegassi-mvps-ddos-resilience-00 [OPSAWG]
DDoS resilience profile. Cell-aware geometric median strengthens
A5; volume-independent detection across 11 scenarios.
D-5. draft-melegassi-mvps-ai-coherence-00 [AIPM]
AI-coherence extension. C_2 instantiated with W_2 / CKA on
replica embeddings. Theorem C-5.6: A5 holds with Byzantine-
robust C_2^gm on the embedding ball.
D-6. draft-melegassi-ippm-mvps-coherence-leadtime-00 [IPPM]
Closed-form lead-time on rank-low propagating signals.
L_ZD.1' linear, L_ZD.2' exponential (worm-class), L_ZD.3
sparse-direction SIGN-REVERSAL boundary. MC SIGN-CLAIM 9/9.
D-7. draft-melegassi-ippm-mvps-orbital-coherence-00 [IPPM]
Satellite-segment profile. Mixed-medium causal C_1 (vacuum-c
on ISL legs, fiber-c on terrestrial) + TLE-derived C_3^pred.
Seven theorems T-1..T-7. Stein-Lemma + KL chain rule additive
across N independent ground vantages (L_ORB.1/2/3 PASS).
----------------------------------------------------------------------
Why exactly nine
----------------------------------------------------------------------
Mutually independent. Each of the nine proves something the
others do not: D-1 the algebra; D-2 the bandwidth-efficient
encoding; D-3 the BFD ride; D-4 the cell-aware Byzantine
strengthening; D-5 the AI surface; D-6 the closed-form lead-
time; D-7 the orbital/multimodal mapping; D-8 the abstract
specification; D-9 the operational composition floor.
Jointly exhaustive. Beyond these nine, no further capstone
is derivable from existing material without introducing
either new measurement or new mathematics. Once D-8 names
the axioms and D-9 names the world number, the structural
picture is closed.
----------------------------------------------------------------------
Where to read everything (one page)
----------------------------------------------------------------------
Combined landing page (chain diagram + downloads + live evidence):
https://catellix.com/v11-evidence.html
Datatracker (each draft has a permanent URL):
draft-melegassi-ippm-mvps-bundle-00
https://datatracker.ietf.org/submit/status/163258/
draft-melegassi-mvps-incremental-be-00
https://datatracker.ietf.org/submit/status/163422/
draft-melegassi-mvps-ai-coherence-00
https://datatracker.ietf.org/submit/status/163424/
draft-melegassi-coherence-bfd-00
https://datatracker.ietf.org/submit/status/163436/
draft-melegassi-mvps-ddos-resilience-00
https://datatracker.ietf.org/submit/status/163435/
draft-melegassi-ippm-mvps-coherence-leadtime
https://datatracker.ietf.org/submit/status/163523/
draft-melegassi-ippm-mvps-orbital-coherence
https://datatracker.ietf.org/submit/status/163530/
draft-melegassi-iab-mvps-architecture
https://datatracker.ietf.org/submit/status/163538/
draft-melegassi-iab-mvps-planetary-floor
https://datatracker.ietf.org/submit/status/163537/
Formal proof companions (for reviewers who want the chase):
MVPS_MATHEMATICAL_EXISTENCE_PROOF_V4.txt (the v4.0 catalogue)
MVPS_ARCH_PROOF.txt (D-8 Invariance Thm)
MVPS_PCF_PROOF.txt (D-9 PCF + corollaries)
MVPS_DETECTION_LATENCY_LEMMA.txt (L_DL, used by D-9)
MVPS_LEAD_TIME_LEMMA.txt (L_LT.A, used by D-6)
MVPS_ORBITAL_PROOF.txt (used by D-7)
Numerical receipts (SHA-256 stamped, exit code = PASS/FAIL):
evidence/planetary_floor_receipt.json (D-8 + D-9)
evidence/orbital_error_exponent_receipt.json (D-7)
evidence/detection_latency_lemma_receipt.json (L_DL)
Reference implementation (pure Python, 0 dependencies) is also
linked from the landing page above.
----------------------------------------------------------------------
Honest caveats (kept explicit so reviewers do not have to dig)
----------------------------------------------------------------------
- D-9 PCF speedup of ~1220x is the ratio of the BGP-convergence
worst case (300 s, RFC 4271 Section 10) to a deployed MVPS
instantiation at N = 1000 fiber vantages antipodal (~246 ms).
Against the pure causal floor (~196 ms fiber) the ratio is
~1531x. Section 7 of D-9 makes both numbers explicit so they
cannot be confused.
- D-7 orbital remains gated by Operational Hypothesis H-5 (path-
identity exposure at the vantage). No major LEO operator
publishes such mappings today; absent H-5 the framework
degenerates to a single-axis C_1 detector. OP-1 (Starlink
TLE reference simulation) and OP-2 (exposure protocol) are
explicitly catalogued open problems.
- D-6 MC magnitude is within +/- 40% of closed form only in the
fast-growth regime (Slammer-class, N >= 30); slower growth
(T_d > 30 s) confirms the SIGN claim but not the magnitude.
This is stated honestly in Section 5.5 of D-6.
- D-8 inherits Stein-additivity ONLY under axiom A4 (conditional
independence of vantages). Architectures that violate A4
(BGP-style correlated propagation along AS paths) are bound
above by tau_sampling -- exactly as catalogued in D-9 Section 6.
----------------------------------------------------------------------
Reviewer asks
----------------------------------------------------------------------
1. (Architectural) Is the conformance pattern of D-8 (axiom set
+ Invariance Theorem) acceptable in the IAB / RFC 1958 /
RFC 3439 / RFC 2475 architectural-document tradition, or
should it be re-cast as a category-theoretic functor?
D-8 Section 6.2 explains why the weaker conformance answer
was chosen; pushback welcome.
2. (PCF composition) Does the five-floor decomposition in D-9
Section 5 miss any latency component that should be a sixth
floor? Candidates that were considered and rejected (with
reasons) are listed in D-9 Section 8.
3. (Non-conformance catalogue) D-8 Section 7.3 catalogues
BGP-4, BFD, DNS, TCP-RTX as non-conformant. Reviewers from
the respective WGs are explicitly invited to dispute the
specific axiom assignment for their protocol.
4. (Operational) D-9 Section 11 invites large operators to
measure the tau_causal floor in the wild on a conformant
MVPS deployment; the family's final open question is
empirical, not mathematical.
Best regards,
Leonardo Melegassi
Catellix Research, Andradina/SP, Brazil
mailto:[email protected]
https://catellix.com/v11-evidence.html_______________________________________________
Int-area mailing list -- [email protected]
To unsubscribe send an email to [email protected]
