Notes
Field writing
Short, sharp notes — anti-hype, globally minded, educational. Not financial advice.
Start here
Sovereign Starter
Prefer an ordered path over chronology? Five beginner modules — keys, money, Nostr, storage, and local AI — each with two notes already on the site.
Open Sovereign Starter →Latest
A Subaddress Is Not a Separate Identity: Janus Outputs, Sweeps, and the Seed Boundary
Subaddresses give you receive unlinkability against observers and honest payers. They do not give you identity isolation against an active counterparty, your own consolidation transactions, or a service's account database.
· Privacy & Monero · janus-attack, carrot, unlinkability
A Docs MCP Can Read the Whole Protocol. The Line to Watch Is Signing
A documentation MCP server is read-only by design, and that is what makes it easy to say yes to. The line worth watching is not what an agent can read, but whether anything in its reach can sign.
· Nostr · nostrbook, nip-01, nip-19
A Threat Model Is Not a Fear List: Writing Yours on One Page
A threat model is not a list of things to be afraid of. It is a short written record of what you are protecting, from whom, how bad failure would be, and which costs you accept to prevent it.
· Self-custody · security-planning, threat-model, backups
Release radar — weekly upstream release watch
October 2026
A Subaddress Is Not a Separate Identity: Janus Outputs, Sweeps, and the Seed Boundary
Subaddresses give you receive unlinkability against observers and honest payers. They do not give you identity isolation against an active counterparty, your own consolidation transactions, or a service's account database.
· Privacy & Monero · janus-attack, carrot, unlinkability
A Docs MCP Can Read the Whole Protocol. The Line to Watch Is Signing
A documentation MCP server is read-only by design, and that is what makes it easy to say yes to. The line worth watching is not what an agent can read, but whether anything in its reach can sign.
· Nostr · nostrbook, nip-01, nip-19
A Threat Model Is Not a Fear List: Writing Yours on One Page
A threat model is not a list of things to be afraid of. It is a short written record of what you are protecting, from whom, how bad failure would be, and which costs you accept to prevent it.
· Self-custody · security-planning, threat-model, backups
A Backup Is Not a Recovery: Seed Phrases, Passphrases, and the Restore You Haven't Tested
A seed backup is not a recovery plan. A backup is a copy of a secret. Recovery is a procedure that turns that copy back into usable keys, on a new device, under stress, possibly done by someone other than you.
· Self-custody · seed-phrase, bip-39, bip-32
Your nsec Is Not a Password: Nostr Key Handling and Signers
Your nsec is not a password. A password unlocks an account that someone else holds. On Nostr, the private key is the account: there is no server to reset it, no recovery email, and anyone who copies it can speak as you, indefinitely.
· Nostr · nsec, key-handling, nip-49
On-Chain Privacy Is Not Network Privacy: Node Choice, Tor, and I2P
On-chain privacy is not network privacy. Cryptography can hide who paid whom on the ledger, but whoever first sees your transaction, or your wallet's queries, can link them to your network address unless you route around that.
· Privacy & Monero · network-privacy, tor, i2p
Docker Is the Starting Line, Not the Boundary: Mapping Every Bridge Out of Agent Zero
Agent Zero's container is the default starting position, not the security boundary. The real boundary is the sum of the bridges you enable: which model endpoints each slot talks to, whether the host connector may read files or run code, which browser it drives, and which plugins, MCP servers, and extensions you install.
· Local AI · agent-zero, docker, host-bridge
A Preimage Is Not a Receipt: What a Settled Lightning Payment Actually Proves
A preimage proves the secret behind a payment hash was released, so that HTLC settled. It does not prove who paid, stay secret after release, or show that a particular person was credited.
· Bitcoin & Lightning · preimage, proof-of-payment, htlc
A Background Upgrade Is Not a Verified Pull: Ollama 0.40.2–0.40.3 and On-Disk Model Migration
Ollama 0.40.2 converts older models to llama.cpp-compatible builds the first time you run them; 0.40.3 pauses that background upgrade. A familiar tag is not a verified pull.
· Local AI · model-migration, gguf, open-weights
Pay-Once Is Not a Subscription: Autonomi Upload Payment vs Retrieval
Autonomi's documentation describes one payment event at write time, no recurring storage charge, and no fee to read.
· Storage · payment-model, pay-once, immutable-storage
A Seed Phrase Is Not a Wallet Definition: BIP 380 Output Descriptors
A mnemonic backs up entropy, not wallet definition. BIP 380 output descriptors name scripts, keys, paths, and checksums so watch-only and air-gapped sides agree on the same address set.
· Self-custody · bip-380, output-descriptors, wallet-backup
An Anonymity Set Is Not a Ring Size
An anonymity set is the group of candidates an observer cannot tell apart, not the count of candidates a transaction names.
· Privacy & Monero · zcash, ring-signatures, ironwood
A Decision Endpoint Is Not a Chat Completion: Ollama’s /v1/systemone and What You Can Verify
Ollama’s /v1/systemone returns choices, probabilities, and scores instead of chat text. A decision endpoint is a different contract from /v1/chat/completions — what you can check, and what you still trust.
· Local AI · decision-models, systemone, api-surface
September 2026
AUTH Is Not Identity: What NIP-42 Relay Challenges Actually Prove
NIP-42 AUTH proves possession of a private key for a connection-scoped challenge bound to a relay URL. It does not prove legal identity, payment status, or that the wider network endorsed you.
· Nostr · nip-42, auth, kind-22242
A Model Tag Is Not a Runner Contract: Ollama’s MLX Default and What Still Differs
Ollama v0.40.0-rc0 defaults supported Apple Silicon models to MLX under the same pull/run tags — a model tag is not a runner contract. Pin versions, name the engine, keep offline plans honest.
· Local AI · mlx, lm-studio, apple-silicon
Inference Residency Is Not Action Residency: Tools, MCP, and Local LLMs
Local open-weight sampling can keep prompts on machines you control — that is inference residency. Tools, Ollama web search/fetch, and MCP servers move side effects off-box. Grammar-constrained decoding (GBNF) stays local without remote tools.
· Local AI · tool-calling, gbnf, mcp
A Privacy Mode Is Not a Proof: Venice's Ladder from Policy to Enclave
Venice's Anonymized → Private → TEE → E2EE ladder is a claim about who sees plaintext — not a cryptographic proof by itself. What a client can verify (capability flags, nonce-bound attestation, binding checks) versus what remains trust (proxy retention, ZDR partners, full DCAP).
· Local AI · venice-ai, tee, e2ee
A View Key Is Not a Transparent Chain
A Monero private view key is an explicit allowance on an opaque ledger — not a second public chain. What sharing it reveals (incoming), what it does not (spend authority), and why bare view-key balances need key images.
· Privacy & Monero · view-keys, spend-keys, selective-disclosure
A BOLT-12 Offer Is Not a Reusable Invoice
An offer (lno) is a sticky precursor, not a reusable invoice — payers negotiate a fresh BOLT-12 invoice over onion messages, with blinded paths for tip-jar reachability.
· Bitcoin & Lightning · offers, onion-messages, blinded-paths
Successful Load Is Not Proof: Embedded GGUF Alignment and Wrong Tensors
llama.cpp b11019 fixed embedded GGUF data-section alignment so a successful load can no longer silently return wrong tensors — packaging integrity for sovereign local AI.
· Local AI · gguf, llama-cpp, oss-ai
Stale Lists and Dark Outboxes: When NIP-65 Metadata Lies by Age or Uptime
NIP-65 kind:10002 is discovery, not authority — stale relay lists and dark outboxes are the real failure modes; client merge is not consensus.
· Nostr · nip-65, relays, nostr
Autonomi Stores Immutable Chunks — Public Address vs Private DataMap
On Autonomi the storage primitive is the immutable content-addressed chunk. Public address and private DataMap are two retrieval stories over the same substrate — updates are new uploads.
· Storage · immutability, decentralized-storage, autonomi
PSBT and Air-Gapped Signing: Keep Keys Offline While the Network Watches
BIP 174 PSBT roles and air-gapped signing — online watch-only builds, offline signs, sealed packet in between. Protocol literacy, not a shopping guide.
· Self-custody · psbt, bip-174, bitcoin
NIP-65: Relay Lists, Outbox, and Inbox — Routing Beyond One Home Relay
NIP-65 kind:10002 advertises per-pubkey write (outbox) and read (inbox) relays — routing beyond one home relay, beside NIP-11.
· Nostr · nip-65, relays, nostr
Open Weights Were Not Enough: DeepSeek Harness, Cordis, and the Local Agent Stack
Open weights solve half of sovereignty. DeepSeek Harness and Cordis open the other half — a replaceable agent runtime you can point at local inference.
· Local AI · deai, oss-ai, local-ai
Local Chat Is Not Data Residency: RAG, Embeddings, and Vector Stores
A local chat UI is not data residency if embeddings or the vector index leave the building. RAG, local encoders, and on-disk stores versus hosted OpenAI-shaped APIs.
· Local AI · rag, deai, local-ai
From Rings of 16 to the Whole Chain: What FCMP++ Changes
Monero-class privacy already hides recipients, amounts, and spenders with a small ring. Full-chain membership proofs aim to replace that fixed decoy set with membership across the ledger — without a forced address migration.
· Privacy & Monero · fcmp, privacy, monero
Receive-Only Lightning Is Not a Merchant Stack
Most Lightning talk collapses into shop tills or routing businesses. Receive-only tips need a payable request, inbound capacity, and often a static entry point — not a merchant stack.
· Bitcoin & Lightning · bolt-11, receive-only, lightning
Autonomi Without the Data Map Is Just Scattered Ciphertext
Most decentralized storage talk collapses into cloud with a token. Autonomi’s educational core is client-side self-encryption, the Data Map, and pay-once content-addressed writes.
· Storage · self-encryption, decentralized-storage, autonomi
Self-Custody Without Verify-Before-Trust Is a Slogan
Not your keys, not your coins is still the right first sentence. It is also incomplete — keys only stay yours if software, addresses, and assumptions are things you can check.
· Self-custody · verify-before-trust, wallets, self-custody
NIP-11: How Nostr Relays Advertise What They Can Do
Nostr is often introduced as notes and other stuff transmitted by relays. That slogan is accurate and incomplete. Clients speak a small event protocol; relays decide what they store.
· Nostr · nip-11, relays, nostr
Local and Peer-to-Peer AI: Data Residency Without the Hype
Most people meet large language models through a browser tab or an API key. That path is convenient: someone else owns the GPUs, the ops, and the uptime. It is also a trust and dependency bargain.
· Local AI · open-weights, deai, local-ai
What Transparent Ledgers Leak — and Why On-Chain Privacy Matters
Most public blockchains advertise transparency as a virtue. That claim is half-true. Anyone can verify the rules of the system; anyone can also scrape a permanent graph of who paid whom.
· Privacy & Monero · privacy, monero, nostr