When Scrutiny Becomes Cheap
🔗 **Source:** [When Scrutiny Becomes Cheap](https://www.jonoalderson.com/conjecture/when-scrutiny-becomes-cheap/)
> [!note]+ Document Notes
> tags: [ai-economics, scrutiny-abundance, seo-evolution, brand-strategy, organizational-drift, due-diligence]
> author: Jono Alderson
> source: jonoalderson.com
> reference: https://www.jonoalderson.com/conjecture/when-scrutiny-becomes-cheap/
> date created: {{date:YYYY-MM-DD}} {{time:HH:mm}}
>
> # 🏷️ Title: When scrutiny becomes cheap
>
<!-- deep-read:sections start -->
## 🧭 Table of Contents
- [📝 Summary](#-summary)
- [📖 Main Content](#-main-content)
- [🗒️ Notes](#️-notes)
- [📚 References](#-references)
- [🗣️ Paraphrase](#️-paraphrase)
- [📰 Newsletter](#-newsletter)
- [✅ Action Items](#-action-items)
- [🧪 Hypotheses](#-hypotheses)
- [📊 Methodology](#-methodology)
- [❓ Questions](#-questions)
- [🔮 Future Research](#-future-research)
- [💡 Implications](#-implications)
- [🧠 Learning Styles Assessment](#-learning-styles-assessment)
- [📊 Accessibility Assessment](#-accessibility-assessment)
- [🌍 DEIB Assessment](#-deib-assessment)
- [✅ Logical Validity](#-logical-validity)
- [🛡️ DARVO Analysis](#-darvo-analysis)
- [🎯 Conclusion](#-conclusion)
### 🔗 References
- [When scrutiny becomes cheap — Jono Alderson](https://www.jonoalderson.com/conjecture/when-scrutiny-becomes-cheap/)
## 📝 Summary
- An AI security tool called Mythos found lots of vulnerabilities and caused a panic. Alderson argues the vulnerabilities weren't the story; the collapse in the *cost* of finding them was.
- It matters because security is just the first domain where this cost collapse shows up. The same economic shift is coming for marketing, HR, procurement, investor relations, and SEO.
- Key stakeholders: brands, publishers, SEOs, procurement teams, investors, employees, journalists, and the AI agents now doing the inspecting on everyone's behalf.
- Risk and tradeoff: organizations that are "mostly made of signals" — brands, rankings, testimonials, certifications — lose their protective moat when anyone can cheaply check the signal against the substance behind it.
- Bottom line: the internet made information abundant; AI makes scrutiny abundant. The new competitive advantage belongs to organizations whose story requires the least explanation.
## 📖 Main Content
- The core claim is an economics-of-attention argument. Every organization runs under constraints, so it accumulates compromises — "good enough" processes, stale messaging, policies that made sense once. Over time these create countless small gaps between how a company presents itself and how it actually operates. Alderson calls the slow version of this "drift."
- That gap used to be survivable for one reason: inspection was expensive. No buyer could read every support ticket, benchmark every claim, and check five competitors. So markets invented shortcuts — brands, awards, rankings, testimonials, "trusted by" logo strips — to compress complexity into something a tired person can act on.
- AI changes the economics. Not because it's magic or always right, but because it makes certain kinds of scrutiny cheap enough to become routine. Point a model at thousands of pages, reviews, transcripts, and complaints, then ask one question: what doesn't add up? That question used to be expensive; now it's casual.
- The consequence reframes SEO. For twenty years the web rewarded discoverability — can you be found, can you rank. In a scrutiny-abundant world, visibility becomes an *invitation to be inspected*. A search for "best CRM software" stops being a treasure hunt and becomes due diligence: an agent compares options, analyzes complaints, finds hidden risks, and recommends the one least likely to waste six months.
- Machines don't respect departmental boundaries. They notice when the homepage promise contradicts the reviews, when the help center contradicts the sales deck, when the "people-first culture" produces a suspicious number of public resignations. Companies aren't usually sitting on one big scandal; they're sitting on thousands of small inconsistencies — individually manageable, collectively revealing.
- Terms worth defining: *scrutiny* here means the active work of evaluation (auditing, fact-checking, cross-referencing), distinct from *discovery* (being found). *Drift* is the normal, non-malicious divergence of claims from reality over time.
- Contested or uncertain: this is a conjecture (the URL even lives under `/conjecture/`). Alderson offers no data on how reliably models actually surface real inconsistencies versus hallucinated ones, and he concedes branding isn't dead — only that signal-heavy brands have a problem. The reliability of AI as an inspector is the load-bearing assumption left untested.
### 🗒️ Notes
- The Mythos panic is the hook, not the subject. The reusable insight: when a cost collapses, the interesting effect is the behavior change downstream, not the first visible symptom.
- Compromises aren't lies. Alderson is careful: drift "means the organisation is alive, messy and full of humans." Decisions made under pressure by different teams at different times with different incentives.
- The shortcut economy exists *because* proper investigation is too expensive. Brands, certifications, analyst reports — all compression artifacts for a world where due diligence on everything is impossible "unless you have no job, no children, no inbox."
- The pivot question — "What doesn't add up?" — is the whole essay in five words. It's cheap to ask now, and it scales across customers, employees, investors, procurement, competitors, and journalists simultaneously.
- The SEO reframe is the sharpest practical takeaway: search is "moving from discovery towards evaluation." You can't fix evaluation with schema markup or a content calendar. You reduce the gap between what you say and what you are.
- The closing inversion is the thesis in miniature: when scrutiny is cheap, optimize the thing the signal points to, not the signal.
## 📚 References
- [When scrutiny becomes cheap — Jono Alderson](https://www.jonoalderson.com/conjecture/when-scrutiny-becomes-cheap/) — the source essay.
- Mythos AI security tool (referenced as "the recent Mythos panic") — context the essay assumes; no primary link provided in the text.
## 🗣️ Paraphrase
- The Mythos story wasn't surprising because an AI tool found bugs — all software is fragile. It was surprising because finding the bugs suddenly got cheap, and security is only the first place that cost collapse will bite.
- Companies survive on a gap between their story and their substance, a gap that was safe only while inspecting it was expensive. AI removes that expense, so brands built mostly on signals — rather than on the substance the signals point to — are now exposed, and SEO shifts from helping you get discovered to determining what conclusions an inspector reaches about you.
## 📰 Newsletter
- Here's a thought to chew on between plugin updates: the web spent two decades teaching everyone to win at discoverability — rank, get found, show up. Jono Alderson's "When scrutiny becomes cheap" argues the game is changing under our feet. When an AI agent can read every review, ticket, and policy and ask "what doesn't add up?", visibility stops being a prize and becomes an invitation to be inspected.
- For the WordPress crowd, the takeaway is delightfully un-technical: you can't schema-markup your way out of a contradiction between your homepage and your reviews. The fix is boring and real — better products, cleaner operations, fewer gaps between the story and the substance. Worth a read before your next "brand refresh."
## ✅ Action Items
#### DevRel
- Developer advocacy: when demoing AI-agent search behavior, show the "what doesn't add up?" query against a real site to make the shift concrete.
- Education: build a short workshop on "inspectability" as the successor to "discoverability" for content teams.
- Developer experience: document how structured data should now back *verifiable* claims, not decorative ones.
- Developer marketing: audit your own funnel for claim-vs-evidence gaps an agent would surface.
- Community management: survey practitioners on whether their orgs are measuring drift at all.
- Internal/external/cross-project: align marketing, ops, and support narratives so cross-channel inspection finds consistency.
#### OSPO
- Policy & Governance: treat public-claim accuracy as a governance concern, with owners across marketing and ops.
- License Compliance: ensure stated license posture matches actual dependency reality (a classic drift surface).
- Security Management: the Mythos lesson — assume vuln-finding cost is near zero; prioritize remediation throughput over obscurity.
- Software Inventory: keep SBOMs fresh; an agent comparing your security page to your dependency tree will notice the gap.
- Developer Support: offer consultation hours on closing claim-vs-substance gaps in public docs.
- Community Engagement: coordinate disclosure norms now that cheap scrutiny finds issues faster than embargo cycles assume.
- Strategic Alignment: add "inspectability" as a tracked KPI alongside visibility metrics.
#### Community
- Maintainers: run a claim-audit clinic — does the README promise match the issue tracker reality?
- Contributors: create a "first issue" rotation for fixing documented-vs-actual behavior mismatches.
- Sponsors: publish transparent budgets and milestones; cheap scrutiny rewards orgs whose numbers reconcile.
- Ecosystem: run a cross-project drill where agents inspect several projects for contradictions and report patterns.
#### WordPress
- Core: treat changelog accuracy and deprecation honesty as inspectability assets.
- Core Performance: publish real perf data; gaps between claimed and measured speed are now cheap to expose.
- Core AI: build evaluation sets for AI-summarized comparisons of WordPress vs. alternatives.
- **Core Program Team:** add a cross-team retrospective question — where does our public story outrun our substance?
- Marketing: shift narrative spend toward substance proof points, not signal polishing.
- Documentation: close DevHub/HelpHub gaps that contradict actual behavior; those are prime "doesn't add up" hits.
- Accessibility: ensure accessibility claims match audit reality before an agent does the audit for you.
- Support: knowledge-base contradictions with the docs are a measurable drift surface — triage them.
#### The WPCC
- Programs: fund a "drift audit" fellowship that helps small projects reconcile claims with operations.
- Projects: run governance clinics on owning public-claim accuracy across teams.
- Funding experiments: transparent ledgers as a competitive advantage in a scrutiny-abundant world.
### FAIR
- Technical independence: local, verifiable metadata and provenance make claims inspectable by design.
- Policies and covenants: funding transparency directly reduces the claim-vs-reality gap agents probe.
- Productized governance artifacts: scorecards that reconcile stated and actual governance survive inspection.
## 🧪 Hypotheses
- If AI-agent-mediated evaluation rises, then organizations with high claim-vs-substance gaps will see measurably worse agent-recommendation rates than signal-matched peers.
- If "search moves from discovery to evaluation," then traditional SEO levers (keywords, backlinks, schema) will show declining correlation with agent-driven conversions over time.
- Brands "mostly made of signals" will lose market share faster under agent-mediated procurement than brands with verifiable substance.
- The number of small inconsistencies ("deltas between story and reality"), not the presence of one scandal, predicts reputational exposure when scrutiny is cheap.
- Organizations that proactively reduce their say-do gap will spend less on reactive PR than those that optimize signals.
- AI inspectors will produce a meaningful false-positive rate (flagging hallucinated contradictions), creating a new category of "defending against bad scrutiny."
## 📊 Methodology
- Test hypothesis 1 with a controlled study: feed matched company pairs (high vs. low say-do gap) to evaluation agents and measure recommendation outcomes.
- Test the SEO shift with a longitudinal cohort: track agent-citation and agent-recommendation rates against classic ranking signals across 12–24 months.
- Measure "drift" by sampling public claims (homepage, careers, security page) and scoring them against operational evidence (reviews, tickets, filings) — a reproducible delta score.
- Use procurement records and win/loss data to test whether substance-heavy vendors win agent-assisted RFPs more often.
- Assess false-positive rates by having human auditors verify a sample of agent-surfaced "contradictions."
## ❓ Questions
- How reliably do today's models distinguish real inconsistencies from hallucinated ones, and does that reliability gate the whole thesis?
- Who is liable when a cheap-scrutiny agent confidently surfaces a false contradiction that damages a brand?
- Does cheap scrutiny pressure organizations toward genuine integrity, or toward better-coordinated signal management?
- Will defensive "anti-scrutiny" tactics (claim laundering, evidence suppression) emerge as a counter-industry?
- What's the right metric for "inspectability," and can it be gamed the way discoverability was?
## 🔮 Future Research
- Build and validate a "say-do gap" or drift index that organizations can track over time.
- Study the failure modes of AI inspectors — hallucinated contradictions, adversarial content, poisoned sources — as a discipline parallel to SEO spam fighting.
- Investigate whether cheap scrutiny actually improves outcomes for buyers or just shifts the manipulation surface.
- Explore the maintainer-burden angle: who bears the cost when open-source projects must keep claims and reality reconciled under constant agent inspection?
## 💡 Implications
- **Practice:** SEO and content strategy reorient from "get found" to "survive being examined together." The whole footprint — product, careers page, ex-employees, suppliers, support — becomes one evaluable surface. Teams that managed marketing, ops, HR, and legal "separately" lose that luxury because agents don't respect departmental boundaries.
- **Policy:** Public-claim accuracy becomes a governance responsibility with named owners, not a marketing afterthought. Transparency stops being a virtue signal and becomes a defensive asset, since the cheapest thing to verify is whether your stated numbers reconcile.
- **Security:** The Mythos lesson generalizes — assume the cost of finding your weaknesses has collapsed. Obscurity is worthless; remediation throughput and honest disclosure are the moat. This connects to the open-source-maintainer-burden theme: cheap scrutiny aims a firehose of findings at projects whose maintainers were already underwater, so triage and funding models have to scale with the new finding-rate.
- **Education:** "Inspectability" needs to be taught alongside discoverability. The skill shifts from crafting signals to reducing the gap between signal and substance.
- **Funding models:** This is an economics-of-attention and AI-as-leverage story. AI lowers the cost of a previously expensive activity (due diligence) the way the internet lowered the cost of distribution — and just as cheap distribution reshaped media economics, cheap scrutiny reshapes trust economics. It also rhymes with the AI-bubble theme: a lot of brand value is "made of signals," and a cost collapse can reprice that value quickly. Sponsors, grantmakers, and investors get the same cheap tools, so transparent ledgers and reconcilable milestones move from nice-to-have to table stakes.
## 🧠 Learning Styles Assessment
- Reading level estimate: roughly Flesch–Kincaid Grade 9–11. Conversational but argument-dense, with long enumerative sentences ("Customers can inspect… Employees can compare… Investors can test…"). Flesch Reading Ease likely in the 50s.
#### 👀 Visual learners
- Present: none — pure prose, no diagrams or charts.
- Improve: a before/after diagram of "discovery web" vs. "evaluation web"; a drift-over-time line chart; a signal-vs-substance two-column visual.
#### 🔊 Aural learners
- Present: none.
- Improve: a 3-minute audio summary; the essay reads well aloud given its rhythmic, list-driven cadence.
#### 🗣️ Verbal learners
- Present: strong — vivid analogies ("haunted house of dependencies," "treasure hunt" vs. "due diligence"), repeated definitions of shortcuts.
- Improve: a short glossary (scrutiny, drift, inspectability, signal).
#### 🧩 Kinesthetic learners
- Present: none.
- Improve: a hands-on exercise — run "what doesn't add up?" against your own public footprint.
#### 🧠 Logical learners
- Present: clear cause→effect chain (cost collapse → behavior change → SEO reframe).
- Improve: a decision tree for "is your brand mostly signal or substance?"; a KPI table for inspectability.
#### 🧑🤝🧑 Social learners
- Present: none.
- Improve: discussion prompts for marketing/ops teams to find their own say-do gaps together.
#### 🧘♀️ Solitary learners
- Present: reflective, essayistic tone suits solo reading.
- Improve: a self-audit worksheet scoring your claim-vs-substance gaps.
#### 🌿 Naturalistic learners
- Present: implicit — "the organisation is alive, messy" frames the firm as an ecosystem.
- Improve: lifecycle framing of drift as natural erosion needing tending.
## 📊 Accessibility Assessment
#### Screen reader semantics
- Present: as a single-source essay, the analysis uses a clear heading hierarchy.
- Improve: ensure the published article uses real `<h2>`/`<h3>` landmarks, not styled paragraphs.
#### Keyboard navigation
- Present: not applicable to the prose itself.
- Improve: confirm the source page has a skip link and logical focus order.
#### Contrast and readability
- Present: not assessable from text alone.
- Improve: verify 4.5:1 body-text contrast on the published page; the long sentences benefit from generous line-height.
#### Text resizing and reflow
- Present: not assessable from text alone.
- Improve: confirm reflow at 200% zoom without clipping.
#### Media and captions
- Present: no media in the source, so nothing to caption.
- Improve: if an audio version is added, ship a transcript.
#### Forms and interactions
- Present: none.
- Improve: n/a.
#### Standards reference
- Map any published-page gaps to WCAG 2.2 AA (1.4.3 contrast, 1.4.10 reflow, 2.4.1 bypass blocks).
## 🌍 DEIB Assessment
#### Cultural breadth
- Present: examples are generic-corporate (CRM, careers pages, LinkedIn), broadly legible across Western business contexts.
- Improve: global case studies — drift and cheap scrutiny look different under different regulatory and language regimes.
#### Inclusive language
- Present: gender-neutral, plain, jargon defined in passing ("A brand is a shortcut").
- Improve: a one-line plain-language summary up top for non-native readers.
#### Neurodiversity support
- Present: consistent short paragraphs and rhythmic structure aid scanning.
- Improve: more subheadings; the essay runs as one long unbroken argument.
#### Representation
- Present: abstract actors (customers, employees, investors) rather than named individuals.
- Improve: balance the "glassy-eyed LinkedIn posts" snark, which leans on a specific cultural in-joke.
#### Psychological safety
- Present: careful, non-accusatory framing — "None of this necessarily means anybody is lying."
- Improve: note that cheap scrutiny can also harm via false positives, so the burden falls unevenly on smaller orgs.
## ✅ Logical Validity
#### Strong reasoning
- **Causal (whole essay):** the cost-collapse → behavior-change chain is the spine and it's sound — lower the price of an activity and you get more of it. Strong because it generalizes a real security event into a domain-independent economic claim.
- **Analogical (¶29–30):** "The internet made information abundant. AI may make scrutiny abundant." A tight, load-bearing analogy; both are cost-collapse stories, which makes the parallel apt rather than decorative.
- **Inductive (¶24–25):** from many small observed inconsistencies (homepage vs. reviews, help center vs. sales deck) to the generalization that "most companies are… sitting on thousands of small inconsistencies." Reasonable, though the sample is rhetorical, not measured.
- Clarify further: state the key assumption explicitly — that AI inspection is reliable enough to matter — since the whole chain hangs on it.
#### Fallacies
- **Hasty generalization (¶27):** "It exposes drift" in every organization presumes universality from vivid examples; fix by bounding the claim to signal-heavy orgs, which the essay nearly does.
- **Appeal to novelty (implicit, ¶13–15):** "That question used to be expensive. Now it is becoming casual" treats the shift as inevitable; fix by acknowledging adoption friction and inspector unreliability.
- **Suppressed evidence (throughout):** no engagement with AI false positives / hallucinated contradictions, which would complicate the thesis; fix by naming the failure mode.
#### Tone of voice
- Wry, confident, world-weary-expert. Examples: "Jira tickets from 2018," "no job, no children, no inbox and a suspiciously large amount of free time," "a worse reporting dashboard." The humor lands the argument without softening it.
## 🛡️ DARVO Analysis
- Not applicable to this source — it's a forward-looking opinion essay, not a defense against an accusation, so there's no denial, attack, or victim/offender reversal to analyze. The closest adjacent move is Alderson preemptively crediting organizations ("None of this necessarily means anybody is lying"), which is the opposite of DARVO: it removes blame rather than deflecting it.
## 🎯 Conclusion
- Bottom line: the cheap-scrutiny shift means the safest thing to optimize is the substance, not the signal that points to it. The internet rewarded being visible; AI rewards being inspectable, and those aren't the same problem.
- Practical next steps for readers: run "what doesn't add up?" against your own public footprint before someone else does; treat claim-vs-substance gaps as a tracked metric with named owners; stop investing in signals you can't back with evidence; and, if you do SEO, start asking what conclusions an agent reaches when it examines everything around your brand together.
## 🔄 Refactor Appendix — When scrutiny becomes cheap
> Density: 2–5 Literature atoms (H3), 0–3 Permanent candidates (H4).
> Defaults: 3 Literature, 1 Permanent.
### 📚 Literature Atoms (H3)
#### [H1] Cheap scrutiny, not vulnerability count, is the real shift
> [!quote] ID:rw_h1
> The interesting thing was that the cost of finding those problems collapsed.
**Context & use**:
- Opening of the essay (¶3), reframing the Mythos panic.
- The reusable insight: watch the cost collapse, not the first visible symptom.
**Tags**: #xref #aphorisms
<!-- atomicity: 1 idea ✓ -->
#### [H2] Brands exist because investigation is too expensive
> [!quote] ID:rw_h2
> They compress complexity into something a tired person can act on.
**Context & use**:
- ¶11, on why brands, rankings, and certifications exist.
- Explains why a cost collapse in scrutiny threatens signal-based value specifically.
**Tags**: #xref
<!-- atomicity: 1 idea ✓ -->
#### [H3] Drift, not scandal, is what cheap scrutiny exposes
> [!quote] ID:rw_h3
> They are sitting on thousands of small inconsistencies. Tiny deltas between story and reality.
**Context & use**:
- ¶25, the essay's diagnostic core.
- Most reputational exposure is cumulative and mundane, not a single big secret.
**Tags**: #xref #articleseed
<!-- atomicity: 1 idea ✓ -->
#### [H4] Visibility flips from prize to inspection invitation
> [!quote] ID:rw_h4
> In a scrutiny-abundant world, the challenge is being inspectable.
**Context & use**:
- ¶30, the SEO reframe.
- The successor goal to discoverability; can't be solved with schema or content calendars.
**Tags**: #xref #addendum
<!-- atomicity: 1 idea ✓ -->
### 🧠 Permanent Candidates (H4)
#### [P1] When an inspection cost collapses, signal-based value reprices
**Claim** (one sentence, Courtney's voice): When the cost of checking a claim collapses, value built on unverifiable signals erodes and value built on verifiable substance compounds.
**Evidence pointer**: [[#H1]]
**Scope & conditions**: Holds when the cheap inspector is reliable enough to be trusted; breaks if false-positive rates stay high enough that scrutiny output is discounted.
**Links**:
- [[Economics of attention]] — because extends
- [[AI as leverage]] — because applies-in cost-collapse dynamics
- [[Open-source maintainer burden]] — because contrasts on who absorbs the new finding-rate
<!-- atomicity: 1 idea ✓ -->
### ✏️ Fleeting Queue
- [F1] What is the false-positive rate of AI "what doesn't add up?" inspection, and who's liable for a hallucinated contradiction? — https://www.jonoalderson.com/conjecture/when-scrutiny-becomes-cheap/ #fleeting #inbox
- [F2] Draft a "say-do gap / drift index" an org could actually track. #fleeting #inbox
### → Things 3 (max 2 actions)
- [ ] Run a "what doesn't add up?" pass on courtneyr.dev's own public footprint — #things
- [ ] Add an "inspectability" angle to the next SEO/agent-readiness writeup — #things
<!-- deep-read:sections end -->
📄 **Full text:** [[Full Document Contents/Articles/When Scrutiny Becomes Cheap.md|When Scrutiny Becomes Cheap by jonoalderson.com]]
## Highlights
- The interesting thing was that *the cost of finding those problems collapsed*. ([View Highlight](https://read.readwise.io/read/01kw01aw0wexqsh05btmkw0eh2))
- Note: The real shift in the Mythos panic isn't the vulnerability count — it's that the cost of finding problems collapsed.
- Tags: atom/literature/h1, scrutiny-abundance
- Brands, awards, certifications, rankings, testimonials, review scores, analyst reports and glossy websites all exist, at least in part, because proper investigation is too expensive. They compress complexity into something a tired person can act on. ([View Highlight](https://read.readwise.io/read/01kw01ayr8gs6s2sxwasvkrrk7))
- Note: Why trust shortcuts exist — and what's threatened when scrutiny gets cheap.
- Tags: atom/literature/h2, brand-strategy
- This is where the real problem starts. Most companies are not sitting on one big scandal. They are sitting on thousands of small inconsistencies. Tiny deltas between story and reality. Individually manageable. Collectively revealing. ([View Highlight](https://read.readwise.io/read/01kw01b1j0sfrnxt2e5a595chd))
- Note: Cheap scrutiny exposes cumulative drift — thousands of small inconsistencies — not a single scandal.
- Tags: atom/literature/h3, organizational-drift
- In an information-abundant world, the challenge was being *visible*. In a scrutiny-abundant world, the challenge is *being inspectable.* ([View Highlight](https://read.readwise.io/read/01kw01b2z057txbs07g773w9ez))
- Note: The SEO reframe: inspectability succeeds discoverability as the goal.
- Tags: atom/literature/h4, seo-evolution