§15. Layer 1: Reachability

Jason St George. "§15. Layer 1: Reachability" in Next Generation Stores of Value: Privacy, Proofs, Compute. Version v3.2. /v/3.2/read/part-iii/15-layer-1/

Layer 1: Reachability

All of this math still travels as packets.

Every proof, every private swap, every verified FLOP ultimately crosses a handful of cables and radios that a small number of operators can see and shape. A store of value that survives yield-curve control but dies when a few IXPs collude is not a store of value; it is an overlay.

Comms resistance is the stack’s oxygen: the property that air still flows when someone leans on the hose.

Prior art this chapter stands on.

The layer’s components are deployments, not proposals, and the design borrows deliberately: BIP-324 encrypted P2P transport [Mehta et al. 2023] (shipped in Bitcoin Core 26.0, enabled by default in 27.0 [Bitcoin Core 2024]) is the wire-format template for encrypted peer transport; Tor [Dingledine et al. 2004], I2P, and the pluggable-transport lineage—obfs4, meek, Snowflake, and the refraction-networking line ending in Conjure [Frolov et al. 2019]—are the first-class transports and the camouflage fallbacks; BOLT12 Offers [NoBS Bitcoin 2024b] are the receiver-private rendezvous pattern; and the VerifyReach measurement program of §15: Layer 1: Reachability follows the OONI methodology [OONI 2012–] — volunteer vantage points, canary requests, ASN-stratified public aggregation — that has measured real-world censorship at scale since 2012. The thesis’s contribution is not any of these mechanisms; it is binding them into a monetary stack whose falsifiability depends on them, and treating their telemetry as monetary instruments. Full entries in Sources.

Threat Model: What We Must Survive

The comms threat model has four main faces:

Backbone controls.
IP and prefix blocking, BGP blackholing, DNS and SNI filtering, DPI-based resets, QoS throttles on known P2P handshakes.

Exposure and linkage.
Static addresses and reusable invoices let adversaries map who is getting paid; mempool and gossip surveillance reveal patterns.

Eclipse and routing capture.
A node that only talks to a small set of peers in the same ASN can be effectively isolated.

Platform risk at the edge.
App-store takedowns, CDN firewalls, corporate endpoint policies that classify P2P clients as malware.

Design Rules: Protocol Posture Under Pressure

1. Transports must be encrypted by default, and must not mistake encryption for camouflage. Clear-text, distinctive handshakes invite classification, and BIP-324-class v2 encrypted transport removes them. But a stream of uniform-random bytes is itself a fingerprint: Wu et al. (USENIX Security 2023) [Wu et al. 2023] documented that the Great Firewall has, since late 2021, blocked fully encrypted traffic heuristically—by byte-entropy and printable-ASCII tests on the first packet—without needing to identify the protocol at all. “Boring on the wire” therefore splits into two distinct properties, and the stack needs both: encrypted transports (BIP-324, obfs4) that defeat content inspection and protocol fingerprinting, and look-like-something transports that defeat entropy heuristics by resembling traffic the censor cannot afford to block. Earlier versions of this document treated the first as sufficient. It is not.

2. The stack must be transport-agile. No single path (TCP/TLS, Tor, I2P, QUIC) should be a single point of failure, and the fallback ladder must include at least one look-like-something transport per §15: Layer 1: Reachability.

3. Receiver privacy must be the default. Payment rendezvous should be “addressless”—BOLT12 Offers, path-blinding, and shielded pools.

4. Settlement must be refund-safe under squeeze. Atomic, adaptor-signature flows with clear timeout and refund semantics.

5. Edge admission must remain open. No “special” relays or whitelisted entry points.

6. Comms health is part of the telemetry regime. If reachability silently degrades, neutrality degrades with it.

Mechanisms: What Needs to Ship

Encrypted, multi-path P2P transports.

Full nodes speak encrypted P2P by default (BIP-324-class, built on a Noise-style authenticated key exchange—Noise is a handshake framework and supplies the encryption, not the camouflage), with Tor and I2P as first-class citizens. Against censors that fingerprint protocols, obfs4 removes the recognizable handshake and is the encrypted fallback of first resort. Against censors that block on entropy, a second class of transport is needed, and it is first-class here rather than an afterthought: TLS mimicry via uTLS, so the client handshake is byte-identical to a mainstream browser’s; Snowflake, which carries traffic inside WebRTC sessions indistinguishable from video calls, brokered through ephemeral volunteer proxies; meek and domain fronting, which ride inside TLS to a large CDN whose collateral cost of blocking is high; and refraction networking (Conjure) [Frolov et al. 2019], in which a cooperating transit ISP redirects flows addressed to unused IP space in its range, so the censor sees a connection to a decoy and blocking it means blocking the ISP. Each of these has been defeated somewhere; none has been defeated everywhere at once, and the reachability program of §15: Layer 1: Reachability exists to say which one is working where. Relays in high-interdiction ASNs advertise the transports they accept, and the client’s fallback ladder is ordered by measured success in its region rather than by preference.

Transaction-origin privacy.

The threat model of §15: Layer 1: Reachability names mempool and gossip surveillance, and encryption on the wire does nothing about it: a well-connected observer who receives a transaction first from node $A$ learns that $A$ originated or was near the origin, encrypted link or not. Transaction relay therefore uses Dandelion++-style two-phase propagation [Fanti et al. 2018]—an anonymity phase in which the transaction is forwarded along a random privacy graph with no fan-out, followed by a diffusion phase from a node that is not the origin—so that the first-spy estimator’s advantage is bounded rather than total. Layer 1 owns the relay graph; whether the transaction’s contents are also private is a Layer 5 question.

Receiver-private routing for payments.

BOLT12 Offers [NoBS Bitcoin 2024b]: receiver-private, reusable invoices. Two distinct mechanisms are in play and must not be conflated. Onion routing hides the route from the intermediaries: each hop learns only its predecessor and successor. Blinded paths (BOLT12) hide the receiver from the sender: the receiver constructs the last hops of the route and encrypts them so the sender can pay without learning the receiver’s node identity. The first protects both parties from the network; the second protects the receiver from the payer. Federated ecash [Fedimint Project 2024–25] and shielded pools provide endpoint firebreaks—points at which the payment graph is cut and cannot be followed through—with one residual to state plainly: a federated ecash mint is custodial. The federation holds the funds, and the firebreak is bought by trusting a threshold of its members not to steal or freeze them. It is a privacy tool with a custody cost, not a non-custodial one, and users choosing it are choosing that trade.

Settlement survivability.

Adaptor-signature atomic swaps as the default for cross-asset payouts. “Abort & refund” as a first-class, well-signposted action.

Topology and anti-eclipse hardening.

Peer-set diversity and rotation; connections spread across geographies and ASNs; gossip protocols favor multiplicity of paths.

How Layer 1 Anchors the Triad

Communications resilience is not an orthogonal concern; it is the medium in which the triad either lives or suffocates.

For Privacy,

encrypted transports and receiver-private endpoints are what keep the network itself from nullifying ledger-level secrecy. Shielded pools and privacy coins are only as private as their membrane to the outside world. If every shielded payout can be traced to a stable IP graph and a static address, “lawful privacy” collapses into a thin veneer over a rich flow-of-funds analysis. Layer 1’s job is to ensure that settlement paths are as hard to pin down as the flows they carry: private not only in state, but in motion.

For Proofs,

communications resilience keeps verification a public act rather than a priestly privilege. A world where only a small set of well-positioned nodes can fetch and check proofs is a world where VerifyPrice has quietly become an internal KPI rather than a public commodity. If any honest machine with modest connectivity can still reach a verifier over at least one path, proof markets and receipt ledgers remain subject to universal scrutiny. When reachability fragments, the “public” in public verification becomes aspirational.

For Compute,

Layer 1 ensures that useful-work mining and verified inference do not devolve into “whoever still has a clear line to the router wins.” Proof-of-Useful-Work schemes depend on open admission: a wide, geographically and topologically diverse set of provers and miners competing to satisfy claims. If adversaries can choke ingress to a few ASNs or clouds, PoUW degenerates into a club good. Comms health metrics and open-admission design at the edge keep compute supply neutral and keep the token’s claim (“backed by globally demanded work”) from turning into “backed by whichever datacenter the regulator likes.”

All of this math still travels as packets. Layer 0 keeps the machines honest and powered; Layer 1 keeps them in conversation when it is no longer convenient for them to be. The triad’s monetary ambitions depend on both. Without verifiable machines, we do not know what happened. Without resilient communications, we do not know it in time, or at all.

Residual risks: Encrypted transport and obfuscated routing are ingredients, not complete answers to censorship. Residual threats include traffic analysis (pattern and timing correlation), active probing (protocol fingerprinting despite camouflage), endpoint seizure (physical confiscation of devices), long-term sustained shutdowns (where no transport can help), and legal coercion at the social layer. Layer 1 reduces these surfaces; it does not eliminate them. The honest claim is degradation resistance, not censorship immunity.

VerifyReach: Communications Telemetry

Layer 1 introduces VerifyReach as the communications analogue of VerifyPrice:

• Reachability metrics: fraction of vantage points from which key services are reachable.

• Degradation patterns: which networks experience blocking or throttling.

• Transport diversity: percentage of traffic over each transport class.

Target SLOs:

• Core infrastructure reachable from $\geq 95\%$ of sampled ASNs in uncensored regions, where “uncensored” means outside the pre-declared list of known-censored regions maintained by the observatory (the “Censorship regimes” row of the sampling frame below). That list is a published governance parameter; a region moves onto it by measured, sustained filtering, not by assertion.

• $\geq 70\%$ reachability for known-censored regions, measured over the ASNs sampled inside those regions.

• p95 time-to-first-connection $< 10$s uncensored; $< 30$s censored.

• In uncensored regions, no country-level view sees more than $10$% persistent reachability degradation without triggering incident handling. In known-censored regions, where $30$% degradation is already within the SLO, this trigger would fire permanently and mean nothing; there, incident handling is triggered instead by a fall of $\geq 10$ percentage points from that country’s trailing 90-day reachability, so the rule reads on change rather than on a level the region never met.

The denominators are stated because the falsification test of §17: Layer 3: Identity & Claims reads on them: that test counts ASNs in uncensored regions, not globally, so that the $70\%$ censored-region SLO and a $>20\%$-unreachable failure condition are not in contradiction. A region can be at $65\%$ reachability and be an incident; it cannot, by itself, falsify the thesis.

VerifyReach feeds into Layer 6 governance: if reachability collapses in a region, incident response kicks in (alternative transports are promoted, routing is adjusted, and the degradation is visible on public dashboards).

VerifyReach Measurement Specification

Like VerifyPrice, VerifyReach requires a rigorous measurement methodology to prevent gaming and ensure credibility.

Sampling Frame:

VerifyReach sampling frame requirements.

Vantage Points:

• Community nodes: Volunteer-run measurement agents (similar to OONI probes [OONI 2012–]), paid small amounts from the assurance budget (§22: Layer 6: Governance & Telemetry) — not in Work Credits, which mint only against registry workloads.

• Independent labs: At least 3 organizations (academic, NGO, commercial) run measurement infrastructure. No single operator controls $>30\%$ of vantage points.

• Diversity requirement: Vantage points must span $\geq 20$ countries and $\geq 100$ ASNs for measurements to be considered valid.

The two threshold levels do different jobs and are not redundant. The sampling frame targets $\geq$500 ASNs and $\geq$50 countries; the validity floor is the minimum below which measurements are not published at all ($\geq$100 ASNs, $\geq$20 countries). Continent arithmetic follows the five-continent convention used throughout this document ($\geq$10 per inhabited continent).

Metrics:

VerifyReach metrics and targets. Targets shown as uncensored / known-censored.

The ttfc and succ targets bind both, because they measure different things: ttfc percentiles are conditional on successful connection and say nothing about how often connection fails, while succ$_1$/succ$_2$ measure the unconditional success rate and say nothing about how slow the successes were. Neither subsumes the other.

Adversarial Robustness:

VerifyReach adversarial robustness measures.

Publication:

• Raw measurement data (anonymized to protect vantage point operators) published daily.

• Aggregated dashboards updated hourly.

• Quarterly reports summarizing regional trends, incidents, and transport effectiveness.

• All measurement code is open-source and reproducible.

Monetary Consequences:

If VerifyReach for a region falls below thresholds:

• WC minted by operators in that region may face regional risk premiums (higher collateral requirements).

• Corridors primarily serving that region are flagged; users see warnings before transacting.

• Incident response is triggered: alternative transports promoted, routing adjusted.

This makes VerifyReach not just a dashboard metric, but an input to economic risk pricing.

New here? Start with the one-minute version.

Tip: hover a heading to reveal its permalink symbol for copying.