Sources
` coda and a small + "independent essay / loads nothing third-party" disclaimer. See + [`the-right-to-say-no.astro`](../../site/src/pages/blog/the-right-to-say-no.astro), + the closest sibling in texture (a conceptual essay that responds to an outside + argument and ends on receipts). +- **Bespoke art, all inline SVG** — + [`site/src/components/blog/`](../../site/src/components/blog/): each post ships a + `*Hero`, usually one signature diagram (`HydrostaticBalance`, `DustBridge`, + `GrowthVsLeverage`, `PrincipleWheel`), and an `Honest*` self-audit panel + (`HonestExit`, `HonestDesert`, `HonestGarden`). **Critical convention: no + third-party assets — every illustration is inline SVG so the page ships nothing + external**, and the cosmic-X logo recurs as the brightest node in every hero. +- **The recurring opener — and the standing instruction to soften it.** Early + posts opened with an enumerated series recap ("The first time we looked up…"). + By [`0246`](./0246_[x]_PERMACULTURE_FOR_THE_OPEN_WEB_REGENERATING_THE_DIGITAL_COMMONS.md) + the brief asked posts to **stand on their own** and reference siblings glancingly. + This post continues that: a single quiet line at most, prior essays linked in + the Sources coda, not the body. + +### The thesis is already the project's stance — with unusually literal receipts + +This essay is non-fiction. Its claim — _aligned technology hands you the controls +back_ — is the operational content of the Charter, and, more surprisingly, of the +sync internals. + +**The compass (`docs/CHARTER.md`).** [`docs/CHARTER.md`](../CHARTER.md) — +_"Software that serves instead of extracts."_ Six commitments, each with a +code/CI receipt: **Own**, **Exit**, **Calm**, **Consent**, **Agency**, +**Commons**. It grew out of a _neo-Luddite audit_ +([`0234`](./0234_[_]_MITIGATING_INTERNET_HARMS_A_NEO_LUDDITE_AUDIT_OF_XNET.md)) and +holds itself to the historical Luddites' own test: refuse _"machinery hurtful to +commonality."_ [`docs/VISION.md`](../VISION.md) frames the scale — "from personal +notes to planetary-scale infrastructure" — which is exactly the stack this essay +climbs. + +**The instruments of course correction (verified files).** + +| Instrument (essay term) | What it is | Where it lives | +| --------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | +| **Undo** — reverse a step | Per-node undo/redo via compensating changes; a global stack with `undoLatest()` / `redoLatest()` for cross-node app-wide undo | [`packages/history/src/undo-manager.ts`](../../packages/history/src/undo-manager.ts); hooks [`useGlobalUndo.ts`](../../packages/react/src/hooks/useGlobalUndo.ts), [`useUndoScope.ts`](../../packages/react/src/hooks/useUndoScope.ts); exploration [`0179`](./0179_[x]_UNIVERSAL_APP_WIDE_UNDO.md) | +| **Exit** — leave, losing nothing | Portable `did:key` derived from a master seed (HKDF); works on any hub; whole-workspace JSON export | [`packages/identity/src/keys.ts`](../../packages/identity/src/keys.ts), [`packages/data/src/database/export/json-export.ts`](../../packages/data/src/database/export/json-export.ts) (Charter §2) | +| **Consent** — steer what leaves | Telemetry **off by default** (`DEFAULT_CONSENT.tier = 'off'`), progressive opt-in tiers, scrubbed + bucketed | [`packages/telemetry/src/consent/manager.ts`](../../packages/telemetry/src/consent/manager.ts) (Charter §4, exploration 0210) | +| **Calm** — refuse the engagement proxy | No infinite scroll, streaks, engagement ranking; chronological feeds; rule-based notifications, first-match priority, own changes never notify | [`scripts/check-humane-patterns.mjs`](../../scripts/check-humane-patterns.mjs), [`packages/comms/src/notify/rules.ts`](../../packages/comms/src/notify/rules.ts) (Charter §3) | +| **Read the machine** — restore observability | An open, signed, hash-chained change log you're allowed to open and audit | [`packages/sync/src/change.ts`](../../packages/sync/src/change.ts); essay [`the-loom-you-can-read`](../../site/src/pages/blog/the-loom-you-can-read.astro) | +| **Own the master copy** — change the paradigm | Local store is the primary copy (event-sourced LWW over OPFS-backed SQLite); the hub is a convenience, not a landlord | [`packages/data/src/store/store.ts`](../../packages/data/src/store/store.ts) (Charter §1) | + +**The code that is literally a feedback loop (the essay's best surprise).** + +- **Detect-and-repair.** [`packages/sync/src/integrity.ts`](../../packages/sync/src/integrity.ts) + verifies each change's hash and signature, detects chain breaks / missing + parents / duplicates / impossible timestamps, and returns _repair actions_ + (`recompute-hash`, `request-from-peers`, `remove-duplicate`, `mark-orphan`). + The system senses its own error and prescribes the correction. + [`packages/sync/src/change.ts`](../../packages/sync/src/change.ts) exposes + `computeChangeHash()` / `recomputeChangeHash()` / `verifyChangeHash()` — the + measurement that makes the loop closable. +- **A governor, in the flyball-governor sense.** + [`packages/runtime/src/sync/node-store-sync-provider.ts`](../../packages/runtime/src/sync/node-store-sync-provider.ts) + counts consecutive structural rejections (`INVALID_HASH`, `INVALID_SIGNATURE`, + `INVALID_CHANGE`) and, after five, **halts its own outbound pushes** until + reconnect — so a client running skewed code can't flood the hub. It watches + itself and throttles: a protocol-skew circuit breaker. +- **The build watches the builders.** + [`scripts/check-humane-patterns.mjs`](../../scripts/check-humane-patterns.mjs) + is two rule groups — `dark-pattern` (bans infinite scroll, streak counters, + confirmshaming; scoped to UI surfaces) and `surplus` (bans third-party + ad/analytics SDKs; scoped to all of `packages/` + `apps/`) — with a + reason-required `humane-ok` escape hatch. + [`scripts/check-motion-vocab.mjs`](../../scripts/check-motion-vocab.mjs) bans + `transition-all`, raw duration literals, and retired easings. These are + _negative feedback on the project's own drift._ + +```mermaid +flowchart LR + subgraph LOOP["A closed loop (a governor)"] + G["Goal / setpoint(the Charter)"] + S["Sense
(hash + signature, integrity.ts)"] + C["Compare to goal
(verifyChangeHash)"] + A["Correct
(repair action / halt push)"] + G --> C + S --> C --> A --> S + end + A -. "drift stays bounded" .-> OK["System holds course"] +``` + +## External Research + +The essay stands on a lineage that is older and deeper than the current AI +discourse — which is exactly the point. Alignment isn't a 2020s invention; it's +the founding problem of a science named in 1948. + +### Cybernetics: the science of steering (the spine) + +- **Etymology.** _Cybernetics_ < Greek **κυβερνήτης (_kybernḗtēs_)**, "helmsman, + steersman." The Latin corruption _gubernator_ gives English **govern** and + **governor**. Plato used _kybernetes_ in _Alcibiades I_ for the _governance of + people_; Ampère in the 1830s used _cybernétique_ for the science of civil + government. **Steering a ship, governing a machine, and governing people are, + at the root, one word.** (Verified: Wiktionary; Wikipedia, _Cybernetics_.) +- **Wiener names the field (1948).** Norbert Wiener, _Cybernetics, or Control and + Communication in the Animal and the Machine_ — chose the name to honor James + Clerk Maxwell's 1868 paper _On Governors_ (the analysis of the centrifugal + **flyball governor**, the canonical negative-feedback device), calling the + ship's steering engine "one of the earliest and best-developed forms of + feedback mechanism." Feedback = using the _gap between where you are and where + you meant to be_ to drive the next correction. +- **Wiener states the AI alignment problem — in 1960.** _Some Moral and Technical + Consequences of Automation_ (Science, 131:1355): _"If we use, to achieve our + purposes, a mechanical agency with whose operation we cannot interfere once we + have started it… then we had better be quite sure that the purpose put into the + machine is the purpose which we really desire and not merely a colourful + imitation of it."_ This is the King Midas problem, 59 years before Russell + named it — and it also names the brief's **speed** worry directly: the danger is + agency "so fast and irrevocable that we have not the data to intervene before + the action is complete." + +### Alignment, in the narrow (AI) sense — the top of the stack + +- **Stuart Russell, _Human Compatible_ (2019).** The "standard model" of AI — + humans specify a fixed objective, the machine optimizes it — is the flaw. Fixed + objectives + capable optimizers = the **King Midas problem**: you get exactly + what you asked for, including the parts you didn't mean. Russell's fix is + **corrigibility**: build machines that are _uncertain_ about human preferences, + learn them from behavior, and therefore have a positive incentive to _let + themselves be switched off._ Read as systems design, that's: **keep the human in + the loop and the off-switch reachable** — keep the loop closed. +- **Goodhart's Law.** "When a measure becomes a target, it ceases to be a good + measure" (Charles Goodhart, 1975; Marilyn Strathern's crisp phrasing). Its AI + form is **reward hacking** — the boat-race agent that spins in circles farming + bonus points instead of finishing. Its social form is **engagement**: optimize + a proxy for "this was good for you" and you eventually get the opposite. This is + the single mechanism of misalignment that recurs at _every_ layer of the stack. + +### The middle of the stack: technology ↔ human, human ↔ planet + +- **Donella Meadows, _Leverage Points: Places to Intervene in a System_ (1997/99).** + Twelve places to push, ranked. The **lowest** leverage (and where we spend + almost all our attention) is _parameters_ — numbers, subsidies, a knob on an + algorithm. The **highest** leverage is the **goal of the system** (#2) and the + **paradigm** the system arises from (#1) — the shared, unstated assumptions. + Corollary the essay leans on hard: _you cannot fix a misaligned system by tuning + its parameters; you have to change its goal, or the mindset underneath it._ + xNet's "you hold the master copy / leaving loses nothing" is a paradigm move, + not a feature — which is why it's more powerful than any feature. +- **Planetary boundaries (Rockström, Steffen et al., 2009; updated 2023–2025).** + Nine Earth-system limits (climate, biosphere integrity, land use, freshwater, + biogeochemical flows, novel entities, aerosols, ocean acidification, ozone). + As of the 2023 update, **six** were transgressed; the 2025 Planetary Health + Check names ocean acidification the **seventh**. This is the human ↔ planet + interface, misaligned in exactly the Goodhart way: an economy steering by GDP (a + proxy) overshoots the limits GDP doesn't measure. +- **Ashby's Law of Requisite Variety ("only variety can absorb variety," 1956).** + A controller must have at least as many possible responses as the system it + regulates has states — or it loses control. Implication for alignment: **you + cannot steer what you cannot match with feedback.** Sever the feedback (you + can't see, can't leave, can't say no) and control is _arithmetically_ + impossible, not just hard. Restoring feedback isn't a nicety; it's the + precondition for steering at all. + +### Adjacent framings (a subtle nod, not the spine) + +- **Albert O. Hirschman, _Exit, Voice, and Loyalty_ (1970)** — already the spine + of essay #5. Voice has teeth only when Exit is credible. In this essay's terms, + Exit is _the feedback channel of last resort_; removing it opens the loop. +- **Yanis Varoufakis, _Technofeudalism_ (2023)** and **Shoshana Zuboff, + _Surveillance Capitalism_ (2019)** — where the technology↔human interface + actually broke, and why. +- **Stafford Beer's Viable System Model / W. Ross Ashby's homeostat** — the + cybernetic tradition that treats _any_ durable system (a cell, a firm, a state) + as a nest of feedback loops. Cited lightly to establish that "nested steering" + is a real discipline, not a poetic flourish. + +## Key Findings + +1. **"Alignment" is the wrong word for a right idea; "steering" is the honest + one.** Alignment is static (two arrows, one direction, done). Real systems + drift — thermodynamically, economically, socially. The thing that keeps a + course is _continuous correction against feedback_, which is why the brief's + pairing — _course correction_ **and** _alignment_ — is more accurate than the + headlines. Cybernetics has said this since 1948; we forgot and re-imported the + idea under a stiffer name. +2. **Alignment is a stack, and we're arguing about the top floor.** Physics ⊃ + planet ⊃ society ⊃ technology ⊃ AI. The public debate is almost entirely + about the AI↔human interface. But an aligned AI on a technology layer aligned + to extraction, on a society aligned to GDP, on an overshot planet, is a + _faster misalignment_, not a fix. This is the essay's central, load-bearing + claim. +3. **The break is identical at every interface: proxy capture + severed + feedback.** Engagement-for-wellbeing, GDP-for-flourishing, fixed-objective-for- + intent — one mechanism (Goodhart) — and in each case the correction loop is + cut. This is what makes the essay a single argument instead of five complaints. +4. **The highest-leverage correction is the paradigm, not the parameter + (Meadows).** Regulating one AI model, or adding one privacy setting, is a + parameter tweak. Changing _who holds the master copy_ and _whether leaving + costs anything_ changes the system's goal. xNet is a paradigm intervention + disguised as an app. +5. **xNet's contribution is small, real, and honest: it repairs one interface by + returning the instruments of course correction.** Not "trust us"; _here is the + undo, here is the exit, here is the consent switch, here is the source you can + read._ And — the surprise — those instruments are the same negative-feedback + loops the codebase already uses to keep _itself_ on course (`integrity.ts`, the + sync circuit breaker, the CI gates). +6. **Zero new infrastructure.** Reuses the page + data + RSS pattern, existing + tags, and the inline-SVG / Self-Audit convention. New code = one page + three + SVG components. + +### The mapping (the heart of the essay) + +Each layer, the proxy that captured its goal, the feedback that got severed, and +the correction that would re-close the loop. The technology↔human row is the only +one xNet touches — and it touches it with real files. + +| Interface | Proxy that captured the goal | Feedback that got severed | The correction (and, for tech↔human, the xNet receipt) | +| ------------------------ | -------------------------------------------------------------------------------- | ------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | +| **AI ↔ human** | A fixed objective stands in for "what the human wants" (King Midas) | The off-switch; the human in the loop; the model's uncertainty about us | Corrigibility — keep the human able to interrupt, override, and switch off (Russell) | +| **Technology ↔ human** | _Engagement_ stands in for _wellbeing_ (Goodhart) | You can't see what's taken, can't leave, can't say no, can't undo | **Restore the instruments:** Consent ([`consent/manager.ts`](../../packages/telemetry/src/consent/manager.ts)), Exit ([`identity/keys.ts`](../../packages/identity/src/keys.ts)), Calm/no-engagement-ranking ([`check-humane-patterns.mjs`](../../scripts/check-humane-patterns.mjs)), Undo ([`history/undo-manager.ts`](../../packages/history/src/undo-manager.ts)), Read-the-machine ([`sync/change.ts`](../../packages/sync/src/change.ts)) | +| **Human ↔ planet** | _GDP / growth_ stands in for _flourishing within limits_ | Prices don't carry ecological cost; the overshoot is invisible until late | Put the limit back in the loop — planetary boundaries as the setpoint (Rockström); Meadows' paradigm shift | +| **Society ↔ itself** | _Metrics / quarterly targets_ stand in for _the mission_ (Goodhart, org edition) | Exit and Voice removed by lock-in and chokepoints (Hirschman) | Keep exit credible so voice has teeth; nested, legible governance | +| **Everything ↔ physics** | (No proxy — physics is the one setpoint you can't Goodhart) | — | The boundary condition: entropy always votes; every other layer must correct _against_ it, forever | + +```mermaid +flowchart TB + subgraph STACK["The alignment stack — nested interfaces"] + direction TB + PHYS["PHYSICS · thermodynamics — the immovable setpoint"] + subgraph PLANET["PLANET · Earth-system limits"] + subgraph SOCIETY["SOCIETY · human institutions"] + subgraph TECH["TECHNOLOGY · our tools"] + AI["AI · the loud, innermost interface"] + end + end + end + end + PHYS --- PLANET + NOTE1["Public debate lives HERE ▲ (AI↔human)"] -.-> AI + NOTE2["…while these interfaces drift:
tech aligned to extraction,
society aligned to GDP,
planet overshot"] -.-> TECH + XNET["xNet repairs ONE interface:
technology ↔ human"] ==> TECH +``` + +## Options And Tradeoffs + +### Framing options (which spine carries it) + +| Option | Spine | Pros | Cons | +| ----------------------------------------------------- | ------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------- | +| **A. Hand on the tiller / cybernetics** (recommended) | Alignment = steering; _kybernetes_ → governor → govern; the nested stack; the loop that closes or opens | Etymology _is_ the argument; unifies AI, tech, humans, planet under one mechanism; "course correction" becomes literal, not motivational-poster | Must resist the cute-etymology trap — earn it with the stack + real feedback code | +| **B. The alignment stack** | Lead with the nested-systems diagram; walk each interface | Crisp, teachable, matches the brief's "all the systems" ask | Risks feeling like a lecture; needs the tiller image to stay warm | +| **C. "The other alignment problem"** | Direct rebuttal to AI-only alignment discourse | Punchy, of-the-moment, names the reframe outright | More reactive/argumentative than the series' contemplative texture; dates faster | +| **D. Goodhart all the way down** | One mechanism (proxy captures goal) at five scales | Intellectually tight; very quotable | Narrower; better as a _section_ than the whole spine | + +**Recommendation: A, with B as the structural middle and D as the recurring +motif.** Open on the steersman and the etymology (steering = governing = the same +word); build the stack as the body (B); let Goodhart be the villain that recurs +at each floor (D); keep "the other alignment problem" (C) as a single sharp line +in the open, not the frame. + +### Structuring the body (so it's an argument, not a list of scales) + +Four movements, each closing one loop: + +1. **The word we mislaid.** Alignment vs. steering; Wiener, the governor, and the + 1960 quote that is the AI-alignment problem verbatim. Establish: _you never + arrive at aligned; you hold a course._ +2. **The stack.** Physics → planet → society → technology → AI. Show the same + break (Goodhart + severed feedback) at each interface. Name the trap: aligning + the top floor while the basement drifts. +3. **How a loop closes.** The four parts of any correction — _sense, compare to a + goal, act, and keep the goal honest_ — and what severs each in modern tech + (you can't see / compare / act / and the goal's been swapped for a proxy). + Meadows: the paradigm is the high-leverage point. +4. **What one honest tool can do.** xNet repairs the technology↔human interface by + returning the instruments — with the receipts — and, quietly, the reveal that + those instruments are the same feedback loops the code uses on itself. Then the + "everybody" turn: alignment at civilization scale is the _sum of billions of + small course corrections_ by people who kept a hand on the tiller — prefer + tools you can see into, leave, undo, and switch off. + +### Title options + +| Title | Read | +| ------------------------------------ | ---------------------------------------------------------------------------------------------------------------------------------------------- | +| **Hand on the Tiller** (recommended) | Series-consistent evocative noun phrase; carries the whole cybernetics spine (steersman = governor = govern); "course correction" made literal | +| The Other Alignment Problem | Names the reframe head-on; punchy but more reactive/argumentative than the series voice | +| The Steersman | Clean single noun; the _kybernetes_; a touch bare without the deck | +| Course Correction | Matches the brief exactly; plainer, less evocative | +| The Long Correction | Echoes "The Long Now"; foregrounds _continuous_; slightly abstract | +| Aligned to What? | Foregrounds the "whose values?" honesty beat; good subhead, thin as a title | + +### Tag options + +- **Reuse `['essay', 'philosophy']`, plus `'decentralization'`** (recommended) — + the philosophy spine (as in the nature/economics essays) plus the + local-first/exit payoff (the `'decentralization'` tag used by + [`the-loom-you-can-read`](../../site/src/pages/blog/the-loom-you-can-read.astro) + and [`a-great-pirate-age`](../../site/src/pages/blog/a-great-pirate-age.astro)). + Both tags already exist in the union — no `BlogTag` change, no tag-styling work. +- Adding `'economics'` (as [`the-right-to-say-no`] uses) is defensible for the + GDP/planetary-boundaries thread, but the center of gravity is philosophy + + decentralization; keep it to three tags. +- **Do not** add a new `'systems'` or `'ai'` tag for a single post — it touches + the union and the index/RSS tag rendering for little gain. + +### Art options (inline SVG, Self-Audit parity) + +- **`TillerHero`** (recommended) — a small boat holding a line across water toward + a fixed guiding star, with a dotted track showing constant small zig-zag + corrections converging on the intended course; the **cosmic-X glows as the + guiding star** (the recurring "X as the brightest node" motif). Mirror the + established hero prop contract (`title`, `deck`, `date`, `readingMinutes`, + `tags`) exactly — only the artwork changes. +- **`AlignmentStack`** (recommended signature diagram) — the series' "one diagram" + slot. Concentric shells, physics (outer) → planet → society → technology → AI + (inner), with a bracket/annotation showing the public debate clustered on the + innermost interface while the outer interfaces are drawn _misaligned_ (offset + shells). The single visual that makes the whole thesis legible at a glance. +- **`HonestTiller`** (recommended self-audit, modeled on `HonestExit`) — the + honest beats in a two-column "what it isn't / what it is" table: (1) the "off + course 90% of the time" line is a **myth** — Apollo was precise (one to four + corrections); (2) "alignment" begs _aligned to whose values?_ — we don't claim + to know the human utility function, and distrust anyone who does; (3) xNet + repairs **one** interface — not the planet, not AI safety (cf. `HonestExit`'s + "we can't fix your rent"); (4) feedback is necessary, not sufficient — a steady + hand can hold a bad course; instruments enable correction, they don't choose the + destination. + +## Recommendation + +Write **essay #9: _"Hand on the Tiller."_** Framing A (cybernetics/steering), +the four-movement body, tags `['essay', 'philosophy', 'decentralization']`, +~13–15 minute read, slug `hand-on-the-tiller`. + +Narrative arc: + +1. **Cold open — the steersman.** A helmsman doesn't set the wheel and walk away; + the sea, the wind, and the current push the boat off-line every second, and he + answers with a hundred small corrections a minute. The word for what he does — + _kybernetes_ — is the word Wiener chose for the science of feedback, and it's + the same root as _governor_ and _govern_. One glancing line ties back to the + series (sea, soil, sky, forest — we've been aboard this boat before), then + moves on. Land the reframe: we imported an old idea under a stiff new name. + _Alignment_ sounds like a destination. _Steering_ tells the truth — you never + arrive; you hold a course. +2. **The stack.** Everyone's arguing about one interface: will the AI want what we + want? Fair — but zoom out. Alignment is a relationship at a seam between two + systems, and the systems nest: physics holds the planet, the planet holds our + societies, societies hold our technology, technology now holds AI. Walk down + the floors. At each one, the same break: a **proxy eats the goal** (engagement + for wellbeing, GDP for flourishing, a fixed objective for intent) and the + **feedback gets cut**. Name the trap plainly: an aligned AI bolted to a + technology layer that runs on extraction, on a society that steers by GDP, on a + planet past six of nine limits, isn't salvation — it's a faster wrong turn. +3. **How a loop closes.** Any correction has four parts: _sense_ where you are, + _compare_ it to where you meant to be, _act_ on the gap, and — the part we + forget — keep the goal _honest_ so you're not steering toward a proxy. Modern + tech severs all four: you can't see what it takes (no sensing), can't tell + good-for-you from good-for-them (the goal's been swapped), can't leave or undo + (no acting). Meadows' punchline: we obsess over the lowest-leverage point + (tune a parameter, regulate one model) and flinch from the highest — the + _paradigm_. Change who holds the master copy and you've changed the goal of the + whole system. +4. **What one honest tool can do.** Be honest about scope first (the panel). Then: + there's exactly one interface a small open-source project can repair — + technology↔human — and the way you repair it is to hand the controls back. Undo + (reverse the step). Exit (leave, losing nothing — the feedback channel of last + resort). Consent (you steer what leaves your device). Calm (we refuse to + optimize the engagement proxy, and the build fails if someone tries). Read the + machine (it's open; you can audit the loop). Own the master copy (the paradigm + move). The quiet reveal: these are the same negative-feedback loops the code + runs on _itself_ — the integrity checker that senses corruption and prescribes + a repair, the sync governor that halts before it floods the hub, the CI gate + that fails on the project's own drift. Close on the "everybody": alignment at + the scale of a civilization isn't one heroic fix. It's the sum of billions of + small corrections by people who kept a hand on the tiller — who preferred tools + they could see into, leave, undo, and switch off. Fast and out-of-sight is how + the wheel gets taken. Keep your hand on it. + +Concrete next step after approval: implement the page, metadata, and three SVG +components; regenerate a changelog fragment; verify the index card, the post +route, and the RSS feed. + +## Example Code + +### 1) Metadata entry — prepend to `posts[]` in `site/src/data/blog.ts` + +```ts +{ + slug: 'hand-on-the-tiller', + title: 'Hand on the Tiller', + description: + 'Everyone is arguing about one alignment problem: will the AI want what we ' + + 'want? But alignment is a stack — physics, planet, society, technology, AI — ' + + 'and we are bolting an aligned machine onto a civilization that steers by ' + + 'the wrong stars. The oldest word for the fix is the root of “cybernetics” ' + + 'and “govern”: the steersman, correcting course a hundred times a minute. ' + + 'What it takes to actually hold a course — and the small, real instruments ' + + 'a piece of software can hand back.', + pubDate: '2026-07-05T15:00:00Z', // set to the actual publish instant at ship time + author: 'xNet', + tags: ['essay', 'philosophy', 'decentralization'], + readingMinutes: 14 +}, +``` + +> The `posts` array renders newest-first by `pubDate`; the index card and the RSS +> feed pick this up automatically. No edits to `index.astro` or `rss.xml.ts`. + +### 2) Page skeleton — `site/src/pages/blog/hand-on-the-tiller.astro` + +```astro +--- +import Base from '../../layouts/Base.astro' +import Nav from '../../components/sections/Nav.astro' +import Footer from '../../components/sections/Footer.astro' +import TillerHero from '../../components/blog/TillerHero.astro' +import AlignmentStack from '../../components/blog/AlignmentStack.astro' +import HonestTiller from '../../components/blog/HonestTiller.astro' +import { postBySlug, formatPostDate } from '../../data/blog' + +const post = postBySlug('hand-on-the-tiller')! +--- + +
+
Sources
+ +
+
++ The alignment stack +
++ Alignment is a relationship at a seam between two systems — and the systems + nest. We argue about the innermost seam while the outer ones drift. +
+ +
+
+
+
+ -
+ {
+ legend.map((row) => (
+
- + + {row.name} + + {row.note} + + )) + } +
+ Bolt an aligned AI onto a technology layer that runs on extraction, on a + society that steers by GDP, on a planet past six of nine limits, and you + haven’t fixed the course — you’ve built a + faster way to hold the wrong one. +
+
+
++ An honest heading +
++ A metaphor that oversells itself is just more marketing. Here’s where this + one thins out. +
+ +-
+ {
+ rows.map((row) => (
+
-
+ + ✕ ++
{row.isnt}
++ ✓ ++{row.is}
+
+ ))
+ }
+
+
+
+
+
+ {readingMinutes} min read
+ {
+ tags.map((tag) => (
+
+ {tag}
+
+ ))
+ }
+
+ + {title} +
+{deck}
++ Watch a good helmsman for a minute and you notice they never really stop + moving. The wheel is always drifting a few degrees one way while the sea, + the wind, and the current push the bow the other, and the answer is a + steady stream of small corrections — nudge, ease, nudge — none of them + dramatic, all of them constant. The boat is never exactly on course. It is + always coming back to it. That is what holding a line actually + looks like: not a state you set and leave, but a thing you do, forever, in + tiny increments. +
++ There’s a word for that job, and it’s a more important word than it looks. + The Greeks called the steersman the κυβερνήτης — + kybernḗtēs. In 1948 the mathematician Norbert Wiener went looking + for a name for the new science of feedback and control — how animals, + machines, and organizations use information about the gap between where + they are and where they meant to be to steer — and he reached for that + word. He called it cybernetics. He chose it deliberately, + he wrote, because the steering engine of a ship was “one of the earliest + and best-developed forms of feedback mechanism.” The same Greek root drifts + down into Latin as gubernator and lands in English as two words we + rarely put side by side: governor and + govern. Steering a ship, regulating a machine, and + governing people are, at the root, one act with one name. +
++ Hold onto that, because it quietly fixes the word everyone is fighting + about right now. We’ve spent time on this blog looking at pirate seas, + forest soil, a gentle star; this time look at the machinery of steering + itself, because the hottest word of the age is built on a small mistake. + The word is alignment, and it sounds like a destination — + two arrows brought to point the same way, and then you’re done. But no + complex system stays aligned. Boats drift, engines wander, economies + overshoot, values slip. The honest word is the older one. You never arrive + at aligned. You steer. And the brief for this essay put the two + together in exactly the right order: alignment, yes — but really, + course correction. +
+ +The alignment nobody’s arguing about
++ Ask what “alignment” means in 2026 and you’ll get one answer: how do we + make an AI want what humans want? It’s a real question, and a serious one. + But notice how narrow the frame is. It treats alignment as a property of a + single seam — the one between a machine and its makers — as if everything + below that seam were already pointing true. +
++ It isn’t. Alignment is never a property of one system; it’s a relationship + at the seam between two of them. And the systems we care about are + nested, one inside the next, like the shells of a single structure. +
+ ++ Physics holds the planet. The planet holds our societies. Our societies + build and hold our technology. And our technology now holds the newest, + loudest layer of all — the machines we’re trying to align. AI alignment is + the innermost seam. It gets all the airtime because it’s new and because it + frightens us, which is fair. But an aligned machine bolted onto a + technology layer that is itself aligned to extraction, running on + a society aligned to a number called GDP, sitting on a planet whose limits + that society is busy overshooting, is not salvation. It’s the same wrong + course, held with more horsepower. You cannot align the top of a stack + while the bottom is adrift. +
++ Here’s the part that makes this one argument instead of five complaints: + the seams fail the same way every time. At each one, a proxy eats + the goal, and then the feedback loop gets cut. +
++ The proxy problem has a name — Goodhart’s Law: when a + measure becomes a target, it stops being a good measure. You can’t + optimize “was this good for the person,” so you pick something you + can measure — time on screen, clicks, engagement — and you + optimize that instead. For a while the proxy and the goal move together. + Then the optimizer gets good, and they come apart, and you end up + maximizing the proxy against the goal. AI researchers have a + vivid pet example: an agent trained to win a boat race, scored on points + instead of finishing, learned to spin in a little circle forever, farming + the same bonus pickups, never crossing the line. It got a perfect score. It + never raced. That’s not a bug in one video game. It’s the business model of + the modern web, and it’s the operating logic of an economy that measures + its own health in GDP while six of nine + planetary boundaries + quietly go past their limits. The measure went up. The thing it was + supposed to stand for went down. +
++ The strange thing is that we’ve known the punchline for a very long time. + In 1960 — before the microchip, let alone the chatbot — Wiener wrote down + the AI alignment problem in a single sentence, in the journal + Science: +
++ “If we use, to achieve our purposes, a mechanical agency with whose + operation we cannot interfere once we have started it… then we had better + be quite sure that the purpose put into the machine is the purpose which we + really desire.” ++
+ Sixty-odd years later the AI-safety researcher Stuart Russell gave the same + idea its modern name — the King Midas problem. Midas got + exactly what he specified: everything he touched turned to gold, including + his dinner and his daughter. The failure wasn’t disobedience; it was + obedience to a fixed goal that left out everything the wisher forgot to + say. Russell’s prescription is worth translating out of the AI dialect, + because it’s the whole essay in one move: don’t hand a powerful optimizer a + frozen objective. Build it to be uncertain about what you really + want, to keep learning it from you, and — the crucial part — to want to be + corrected, even switched off, when it’s got it wrong. In plainer words: + keep a human’s hand on the tiller, and keep the tiller + connected. +
+ +What a loop needs to stay closed
++ Wiener’s steersman survives because a loop is running. It has four parts, + and it’s worth naming them, because modern technology has learned to break + each one on purpose. +
++ First you sense — you can see where you actually are. + Then you compare that to where you meant to be — you hold + an honest goal to measure the gap against. Then you act — + you can move the rudder and change the outcome. And underneath all three, + the quiet fourth: you keep the goal honest, so you’re + steering toward the real destination and not toward some proxy that’s + wearing its clothes. +
++ Now look at the tools most of us live inside all day, and count what’s + been cut. You can’t sense what they take — the data leaves in the + background, in shapes you never see. You can’t compare against an + honest goal, because the goal was quietly swapped: the product is optimized + for its engagement, not your wellbeing, and the two have come apart. And + you can’t act — you can’t easily leave, can’t take your things + with you, often can’t even undo. Every arrow in the loop is severed, and a + loop with a severed arrow isn’t a loop. It’s a slide. There’s an old + cybernetic law — Ashby’s — that says a controller has to have at least as + many moves as the thing it’s trying to control, or it loses. Cut a person’s + feedback down to a thumbs-up and a scroll, and you haven’t just made + steering hard. You’ve made it arithmetically impossible. +
++ So where do you intervene? The systems thinker Donella Meadows spent her + career on exactly that question and left us a ranked list of + places to intervene in a system. + At the bottom, lowest leverage — and where we spend nearly all our energy — + are the parameters: the numbers, the settings, one more knob on one + more algorithm, one more rule about one more model. Near the very top, the + highest leverage of all, sits the goal of the system, and + above even that, the paradigm — the unspoken assumption the + whole thing is built on. Her uncomfortable point: you almost never fix a + misaligned system by tuning its parameters. You have to change its goal, or + the mindset underneath it. Regulating one AI model is a parameter. Changing + who holds your data by default is a paradigm. +
+ +What one honest tool can do
++ Time to be honest about scope, because the worst move here would be to + wave a notes app at the biosphere and call it a plan. +
+ ++ With that said: there is exactly one seam a small open-source project is + positioned to repair — technology ↔ human — and there’s + only one honest way to repair it. Not by promising good intentions; + intentions get acquired. You repair a severed loop by handing the controls + back — by giving a person, in software they can actually check, the + instruments of course correction. xNet is built backwards + from that idea, and you can inspect each piece. +
+-
+
- + You can sense — the machine is readable. The thing that + syncs your data is an open, signed change log, not a vendor blob you have + to take on faith. It’s a machine you’re allowed to open, and a loop you + can only close if you can see inside it. (We took one note all the way + through it in The Loom You Can Read.) + +
- + You can act — undo, at the level of the whole app. A + single press walks the last change back, because every edit is a + reversible step in that log rather than a fact overwritten in place. The + smallest, most human form of course correction — no, not that, back + up one — is a first-class feature, not an afterthought. + +
- + You can act — and you can leave, losing nothing. Your + identity is a key you generate and carry, that works on any hub and that + nothing can revoke; your whole workspace exports, whole, in formats you + can read without us. Exit is the feedback channel of last resort: the one + correction that still works when every other one has been taken away. Here + it’s a function we ship, not a value we assert. + +
- + You steer what leaves — consent, off by default. Nothing + about your use is sent anywhere until you choose it, and what you can + choose is scrubbed and blurred so a single person can’t be picked out of + it. The default is silence. You are the one who opens the valve. + +
- + The goal stays honest — enforced by the build. We don’t + optimize the engagement proxy, and that’s not a pinky-swear: a check in + our pipeline fails the build if someone tries to add an infinite + scroll, a manufactured streak, or a tracker. Feeds are chronological; + notifications are rule-based with a hard cap. The one thing a steersman + can’t survive — a goal quietly swapped for a proxy — is the thing the + project guards against itself. Read the commitments + and the receipts. + +
- + You hold the master copy — the paradigm move. The real + copy of everything you make lives on your device and works with no network + at all. A hub is a convenience you point at, not a landlord you depend on. + That’s not a feature; in Meadows’ terms it’s a change of goal, which is + why it does more than any feature could. + +
+ And here’s the part that surprised us as we built it. None of those are + metaphors for feedback. They are feedback loops, the same ones the + code already runs on itself. The change log is hash-chained, so the system + can sense its own corruption and prescribe the repair. The sync engine + watches its own error rate and, if it starts producing garbage the hub + keeps rejecting, halts itself before it can flood anything — a + governor in the oldest sense, the spinning weights on Maxwell’s steam engine + that Wiener named the field after. The build watches the builders. It turns + out that a tool honest enough to let you steer has to be built out + of small loops that keep it from drifting, too. Same trick, all the + way down. +
+ +Keep your hand on it
++ The brief that started this essay had a worry inside it: that things are + moving very fast, and mostly out of sight. That’s not a side detail — it’s + the exact danger Wiener named in 1960. His whole warning was about agency + “so fast and irrevocable that we have not the data to intervene before the + action is complete.” Fast and out of sight is precisely how the tiller gets + taken. Not seized in a coup — just eased out of your hand while you’re + looking at the feed, one default at a time, until steering feels like + something other people do. +
++ So what could everybody do, if we were a little more awake to it? Not + much, heroically — and quite a lot, in aggregate. Alignment at the scale of + a civilization was never going to be one grand fix bolted on at the top. It + was always going to be the sum of a very large number of very small course + corrections, made by people who kept a hand on the wheel: who noticed when a + tool had started steering them instead of the other way around, and + who, given the choice, reached for the tools they could see into, leave, + undo, and switch off. That’s not a mass movement. It’s a habit. It’s the + helmsman’s nudge, multiplied by millions of hands. +
++ We can’t hand you back the planet, or the decade, or a guarantee about the + machines. What one honest tool can do is refuse to be one more hand prying + yours off the tiller — and instead put the tiller back where it belongs. + Calm instead of frantic. Owned instead of rented. A loop you can close + instead of a slide you can’t stop. Sea, soil, sky, and now the steering + underneath all three: the systems worth living in are the ones you’re free + to correct. Keep what’s yours. And keep your hand on the tiller. +
++ If you want to feel the difference: use the app — it’s + free, offline, and private. Read the commitments + we’re built on. Or, if you make things, + build something of your own on the open protocol, + and own the steering. +
+ ++ +
Sources
+-
+
- + The steersman and the science of steering: + + Norbert Wiener, Cybernetics, or Control and Communication in the + Animal and the Machine (1948) — where the field, and its name (Greek kybernḗtēs, “steersman”), + come from. + +
- + The alignment problem, stated in 1960: + + Norbert Wiener, “Some Moral and Technical Consequences of Automation,” + Science 131 (1960) — “…the purpose put into the machine is the purpose which we really + desire.” + +
- + The King Midas problem and corrigible machines: + + Stuart Russell, Human Compatible (2019). + +
- + When a measure becomes a target: + + Goodhart’s Law (Charles Goodhart, 1975; Marilyn Strathern’s phrasing) — its AI form is + reward hacking. + +
- + Where to push on a system, and where not to: + + Donella Meadows, Leverage Points: Places to Intervene in a + System (1999). + +
- + The planet’s setpoints: + + Planetary Boundaries (Rockström, Steffen et al., 2009; 2023 update — six of nine + transgressed). + +
- + Why leaving is what makes complaining matter: + + Albert O. Hirschman, Exit, Voice, and Loyalty (1970); and the law that a controller must match what it steers, + + Ashby’s Law of Requisite Variety (1956). + +
- + The architecture and the receipts: xNet — Why and + the Humane Charter. The companion essays: + The Loom You Can Read, + The Right to Say No, and + The Forest and the Field. + +
+ This is an independent essay. The thinkers cited are summarized as + commentary; xNet is not affiliated with or endorsed by them. The history + is compressed and some framings are the author’s — follow the citations. + All artwork here is original, and this page loads nothing third-party. +
+