Opinionated
Editorial judgment
Grades are editorial judgments under this published rubric, not neutral facts, legal findings, certifications, ESG ratings, or investment advice.
Scoring
How scores are made.
Good Companies Directory uses an opinionated rubric. It states a moral view about power, extraction, accountability, and harm, then applies that view as consistently as possible through source-backed claims.
Position
Not neutral. Accountable.
Sources required
Evidence standard
Public claims need source links, timestamps, confidence, status, and revision history. Primary records and credible independent reporting carry more confidence than commentary or self-description.
Checked, not endorsed
Verification limit
Verification checks identity, sources, ownership, and score application. It does not guarantee a good grade, remove criticism, or certify that an organization is good.
Evidence only
No paid score
Payment for verification never buys score changes. Scores change only for factual errors, missing evidence, outdated claims, calculation mistakes, or miscalibration.
Who controls the institution, who receives the surplus, and who bears the costs?
Only for documented practices that go beyond ordinary decency.
Weak ownership, weak governance, and crisis-line active-rescue risk limit how high the score can go.
Violence, capture, toxicity, surveillance, opacity, and other harms come last.
Output
Grade labels
Good
50+ · Structurally constrained goodness.
Qualified
42-49 · Materially good, but not structurally locked across all relevant power.
Performative
35-41 · Good behavior without sufficient structural support.
Extractive
0-34 · Systemic extraction without sufficient constraint.
Malignant
Below 0 · Severe harm and a negative total score.
Calculation
Final score
Final score
min(Material + Bonus, Cap) +
Penalties Final score = min(Material + Bonus, Cap) +
Reversibility + Human Harm + Coercive Violence + Ecological Harm +
Policy Capture + Debt Peonage + High-Carbon Products + Toxic Products +
Animal Testing + Youth Capture + Subscription Capture + Shrinkflation +
Price Manipulation + Unilateral Terms + Accountability Opacity +
Identity Capture + Surveillance Capture + Ideological Disavowal
Material axes are positive. Bonuses are small. Caps prevent weak structures and unsafe crisis-line active-rescue practices from floating upward. Penalties can drive the total below zero.
Classification
Size, status, and badges
SMALL / MEDIUM / LARGE / INSTITUTIONAL
Company Size
Size is an operational classification, not a moral score. It tells readers whether the ordinary rubric is being applied at institutional scale or with the small-business calibration.
Headcount can inform the call, but control, operating footprint, public-company status, national-chain scale, and institutional power matter more than a single employee-count cutoff.
Company Size Scale
- SMALLIndependent, limited-scale, and not a public company, national chain, government, university, or large institution.
- MEDIUMBroader than a local small business, but not a dominant national platform, public-company-scale firm, or major institution.
- LARGENational, public-company-scale, dominant regional, conglomerate, or otherwise large enough that ordinary customers and workers face institutional power.
- INSTITUTIONALGovernment, military, university, agency, church system, union, foundation, movement organization, or similar non-company institution.
Ordinary company policy
Size does not excuse structure.
- Large, medium, and institutional entries are judged directly under the ordinary rubric.
- Public-company scale, national-chain operations, private-equity platforms, catalog rollups, and institutional power do not receive small-business treatment.
Small-business calibration
Small means person-scale reading.
- Independent, limited-scale businesses can receive the small-business calibration when public-company, national-chain, platform, or hidden-parent control is not evident.
- The classification changes how evidence is read; it is not a moral bonus or a substitute for proof.
Cap can fall to 49 / 34
Crisis-Line Active Rescue
Suicide, crisis, or peer-support lines that reserve nonconsensual police, 911, emergency-service, or involuntary psychiatric escalation are capped below A. Deceptive promises or frequent escalation can cap at D even before ordinary penalties. Consent-based no-active-rescue support is not capped by this rule.
Disclose, do not hide
Parent Ownership
Subsidiaries are not mechanically capped at the parent score. Parent control is shown and scored through ownership, governance, reversibility, policy capture, disclosure, and related axes.
Grey + asterisk
Draft Scores
Grey asterisked scores are source-backed scaffolding, not final human review.
Auto-computed
Confidence
Computed from claim confidence, source quality, direct axis support, disputes, recency, reviewer status, and evidence-bearing component coverage.
Sketch to Exhaustive
Thoroughness
A non-scoring research-completeness label. “(AI)” means the thoroughness label itself has not been human-reviewed.
Bounded role snapshots
Public Leaders
Individual public-leader entries are marked separately and judged by a bounded public role, administration, or movement-leadership record rather than private character or a whole biography.
Relevant evidence includes policy, administration, ideology promoted, constitutional norms, tolerance for dissent, corruption or extraction, identity scapegoating, and the use or rejection of coercive violence. Product, subscription, and consumer-market axes should be left neutral or not applicable unless the role actually used those mechanisms.
Non-scoring flags
G / L badges
G marks private-government power. L stands for Lobbying and marks severe Policy Capture through lobbying or public-policy pressure. The score effects still appear through the ordinary axes.
Browsing marks
Vegan / Vegetarian badges
Dietary badges are evidence-backed and non-scoring. Product consequences are scored through High-Carbon Products, Ecological Harm, Toxic Products, and other axes.
Positive base
Material axes
These are the main positive points. Higher means power is more accountable and the product or institution is more useful without extraction.
0-10
Ownership
Control rights: shareholder-dominated at 0, worker cooperative control at 10.
0-10
Governance
Binding decision authority: centralized control at 0, democratic stakeholder control at 10.
Ordinary company policy
Weak structure caps hard.
- Ownership <= 2 and Governance <= 2 cap the score at 35 before penalties.
- Ownership <= 4 and Governance <= 4 cap the score at 41 before penalties.
- Nonprofit status, founder intent, B Corp status, community language, or public-interest branding do not substitute for accountable power.
Small-business calibration
Person-scale firms can reach B.
- Independent small businesses with weak formal ownership or governance can reach B / 49 when person-scale evidence supports the ordinary axes.
- A still requires durable shared power, public accountability, or another structure that survives the current owner.
- The calibration does not relax harm, toxicity, surveillance, labor, coercion, opacity, or hidden-parent penalties.
0-10
Extraction
Surplus allocation, wage share, CEO pay ratio, margins, and structured extraction judgment.
Ordinary company policy
Surplus patterns carry the score.
- Margins, wage share, executive pay, shareholder extraction, and structured surplus allocation are read institutionally.
- Owner, investor, or executive surplus weighs against the score when it comes from worker, contractor, customer, or community dependence.
Small-business calibration
Owner pay is not automatically extraction.
- A small proprietor paying themself living wages, retirement wages, or risk-bearing compensation is not penalized by itself.
- Owner-over-worker surplus, opaque contractor cuts, unpaid family labor, dependency pricing, or avoidable underpayment still count heavily.
0-7
Labor Sovereignty
Worker power: coercive conditions at 0, co-determination or ownership at 7. Employee dissatisfaction matters only when source-backed evidence shows concrete limits on worker agency, such as coercive scheduling, retaliation, wage theft, harassment, unsafe conditions, suppression of worker voice, or extreme turnover.
Ordinary company policy
Systems and policy set the baseline.
- Scheduling, retaliation, wage theft, harassment, safety, turnover, suppression of worker voice, and bargaining power are read through institutional practice.
- A standard procedure matters because it controls many workers even when no single owner personally handles every case.
Small-business calibration
Interpersonal treatment weighs more.
- In a small shop, the owner or manager's direct conduct toward workers, contractors, and applicants carries more evidentiary weight.
- Fair dealing, transparent cuts, non-retaliation, humane scheduling, and meaningful recourse can matter more than formal HR machinery.
0-7
Solidarity with the Unemployed
Treatment of exits and nonworkers, including severance, redeployment, and non-competes.
0-7
Loss-Bearing Fidelity
Willingness to absorb costs to preserve values, workers, users, and public obligations.
0-5
Market Conduct
Pricing fairness, switching costs, lock-in, and rent extraction.
Ordinary company policy
Market power is the main concern.
- Pricing, lock-in, switching costs, hidden fees, and rent extraction are judged by bargaining power, dependency, opacity, and scale.
- Large firms get less benefit of the doubt because customers often cannot meaningfully negotiate or exit.
Small-business calibration
Higher prices can be legitimate.
- Living-wage, retirement-wage, craft, safety, or quality-preserving prices are not extraction by themselves.
- Urgency pressure, opaque add-ons, personalized pressure, dependency pricing, and manipulative advertising still count against small businesses.
0-5
Product Integrity
Preservation of quality rather than degradation for monetization.
0-5
Scale Integrity
Whether growth improves or degrades fairness and accountability.
Rare credits
Bonus axes
Bonuses cannot bypass the ownership and governance caps.
+0 to +3
Openness to Dissent
Credit for tolerating internal, user, customer, worker, and public dissent without retaliation, capture, or viewpoint laundering.
Ordinary company policy
Dissent tolerance must be procedural.
- Credit requires evidence that workers, customers, users, or the public can disagree without retaliation or capture.
- Public relations language, generic civility, or founder accessibility does not earn the bonus by itself.
Small-business calibration
Owner response is direct evidence.
- Because the owner often is the institution, how they handle criticism, disputes, refunds, worker voice, and public pushback matters more.
- Personal warmth is not enough; the evidence must show non-retaliation, repair, and real recourse.
+0 to +3
Constitutional Spirit
Credit for respecting constitutional rights and civil-liberties norms even where private law does not strictly require it.
+0 to +3
Good Deal
Credit for unusually fair value: durable quality, fair pricing, low lock-in, and clear customer surplus.
+0 to +3
Cost Transparency
Credit for clear posted prices, all-in fees, unit costs, public rate cards, margin/cost visibility, or surplus-allocation transparency, especially in markets where opaque quotes, hidden fees, or individualized pricing are normal.
+0 to +3
Anti-Scapegoating Culture
Credit for documented non-punitive practices that prevent scapegoating, mobbing, and isolate-and-expel dynamics. Ordinary civility, peer support, or anti-bullying branding is not enough; the practice must also avoid public shaming, punitive pile-ons, and scapegoating of alleged scapegoaters.
Subtractions
Penalty axes
Penalties are negative numbers. The severe scales are embedded below so the scoring rule and the scale stay in the same place.
-5 to 0
Reversibility
Applied when positive conduct depends on current living leaders, founders, family owners, or other person-contingent governance rather than durable structure.
This is about person-contingent governance, not subscription lock-in.
-20 to 0
Human Harm
Severity, scale, culpability, vulnerable targets, willful reoffense, reputation laundering, pattern escalation, and time decay.
Older harms decay only when conduct stops, repair is visible, and recurrence is structurally less likely.
Human Harm Scale
- 0No accepted human-harm claim.
- -1 to -4Bounded, indirect, remediated, or low-scale harm.
- -5 to -9Serious repeated labor, consumer, privacy, discrimination, or community harm.
- -10 to -14Widespread severe harm, coercive exposure, preventable injury, or major civil-rights harm.
- -15 to -19Mass harm with strong culpability, denial, retaliation, or repeated conduct after notice.
- -20Worst-actor tier: knowing contribution to mass death, forced labor, terror financing, catastrophic industrial harm, or equivalent atrocity.
-30 to 0
Coercive Violence
Direct or institutionally enabled killing, torture, detention abuse, armed repression, forced displacement, violent domination, or nonconsensual crisis-line emergency escalation that exposes callers to confinement or force.
Covers killing, torture, detention abuse, armed repression, forced displacement, policing brutality, terror attacks, violent domination, and crisis-line active-rescue practices that can trigger police, emergency-service, or involuntary psychiatric escalation without caller consent.
-30 to 0
Ecological Harm
Footprint, trajectory, irreversibility, culpability, and scale, including whether ecological harm is central to the model and continued after credible notice.
General ecological damage belongs here; the scale now reaches -30 for ecocide-level conduct. Carbon-intensive products remain visible separately when the product itself is the problem.
Ecological Harm Scale
- 0No accepted ecological-harm claim.
- -1 to -4Bounded manufactured-product, packaging, campus, retail, logistics, or local footprint harms.
- -5 to -9Material ecological harm: high-emissions operations, serious pollution, war-related environmental destruction, pesticide exposure, or repeated damage after notice.
- -10 to -14Severe ecological harm: industrial agriculture, aviation, major infrastructure damage, climate obstruction, or large-scale damage that is central but not worst-tier.
- -15 to -22Very severe ecological harm: chemical contamination, fossil-fuel dependency, mass pollution, large destructive land or water impacts, or repeated denial after credible notice.
- -23 to -30Worst-actor tier: ecocide-scale conduct, fossil extraction or climate obstruction at global scale, catastrophic irreversible contamination, or ecological destruction comparable in gravity to severe institutional violence.
-15 to 0
Policy Capture
Private-interest attempts to bend law, regulation, subsidies, taxes, enforcement, trade, labor, safety, competition, environmental, healthcare, housing, surveillance, civil-liberties, or consumer policy against workers, customers, citizens, affected communities, or ecological life. Public-interest advocacy is not penalized merely because it is lobbying.
Public-interest advocacy is not penalized merely because it involves law or government.
Policy Capture Scale
- 0No accepted private-interest policy-capture claim, or advocacy is clearly public-interest, rights-protective, consumer-protective, worker-protective, ecological, or remedial.
- -1 to -3Limited lobbying or regulatory influence for ordinary institutional advantage without stronger evidence of public harm.
- -4 to -7Material private-interest lobbying or policy pressure on taxes, labor, competition, safety, privacy, environment, healthcare, housing, transportation, or platform rules.
- -8 to -11Major lobbying operation defending market power, weak regulation, harmful subsidies, liability shields, privatized enforcement, anti-worker rules, anti-consumer rules, or anti-citizen rules.
- -12 to -15Worst-actor tier: policy capture is central to preserving coercive or extractive institutional power.
-12 to 0
Debt Peonage
Coercive or structurally inescapable debt systems, including bankruptcy-resistant student debt, compelled servicing relationships, wage or benefit garnishment leverage, distressed-debt buying, debt collection, or refinancing practices that turn education, medical care, housing, household necessity, or public obligation into long-term debtor control. Ordinary transparent lending is not penalized by itself.
Ordinary transparent lending is not enough; the issue is debtor control and realistic inability to exit.
Debt Peonage Scale
- 0No accepted debt-peonage claim, or ordinary transparent credit without debtor-control evidence.
- -1 to -3Limited exposure: opt-in refinancing, tuition or aid-channel participation, ordinary lending, or debt-adjacent services.
- -4 to -6Material exposure: private student lending, legacy student-loan servicing, medical or necessity debt collection, or trap-prone debt products.
- -7 to -9Core debt-control model: compelled student-loan servicing, default-collection leverage, major debt buying, or collection systems the debtor did not meaningfully choose.
- -10 to -12Worst-actor tier: repeated borrower abuse, predatory student lending or servicing, or business models built around people who cannot realistically exit.
-10 to 0
High-Carbon Products
Core products, services, financing, or supply chains that materially depend on fossil-fuel combustion, high-emissions transport, industrial animal agriculture, meat, dairy, or other unusually carbon-intensive activity. This is scored separately from general ecological harm so ordinary consumers can see carbon-intensive product exposure directly.
This keeps carbon-intensive consumer choices visible instead of burying them inside general ecological harm.
High-Carbon Products Scale
- 0No accepted high-carbon-product claim.
- -1 to -2Incidental high-carbon exposure in mixed retail, grocery, restaurant, shipping, or supply chains.
- -3 to -5Material exposure: meat, dairy, gasoline-adjacent retail, combustion transport, conventional vehicles, or high-carbon logistics.
- -6 to -8Core high-carbon model: aviation, combustion fleets, fossil finance, meatpacking, or industrial animal agriculture.
- -9 to -10Worst-actor tier: fossil extraction, petroleum refining/marketing, coal/oil/gas advocacy, or another model dependent on large-scale fossil-carbon release.
-5 to 0
Toxic Products
Current sale of products whose ordinary use, exposure, ingredients, residues, packaging, or formulation creates toxic, addictive, or consumer-safety risk. Food products receive at least light scrutiny for artificial colors, artificial flavors, petro-derived additives, pesticide residues, contaminants, endocrine-disrupting packaging, and ultra-processed formulation. Ordinary adult nightlife or alcohol service is not penalized by itself without evidence of predatory marketing, youth targeting, addiction-extractive design, or unusual product-safety misconduct.
legality is not treated as proof of harmlessness; avoidable industrial ingredients and exposure risks count when evidenced. Ordinary adult nightlife is not a penalty by itself.
Toxic Products Scale
- 0No accepted toxic-product claim, no material additive, residue, contaminant, addictive-design, or exposure concern identified, or only ordinary adult nightlife/alcohol service without predatory or unsafe conduct.
- -1Light ordinary-exposure concern: artificial colors, artificial flavors, petro-derived additives, avoidable processing chemicals, or similar formulation choices.
- -2 to -3Material exposure: routine synthetic additives, pesticide residues, endocrine-active packaging, contamination history, addictive formulation, or hazardous product categories.
- -4Severe or core-product toxic exposure, or credible notice has not changed conduct.
- -5Worst-actor tier: ordinary use or production creates grave toxic, addictive, or consumer-safety harm at scale.
-8 to 0
Animal Testing
Direct animal testing, contract animal testing, animal-testing-dependent product development, or business lines where current safety/regulatory approval normally relies on animal experiments. Stronger penalties apply where animal testing is core, repeated, non-optional, or avoidable; documented cruelty-free or non-animal methods reduce or eliminate the penalty.
A credible cruelty-free or non-animal-testing record can reduce or eliminate the penalty.
Animal Testing Scale
- 0No accepted animal-testing claim, or credible cruelty-free / non-animal-testing evidence.
- -1 to -2Sector exposure or legacy/raw-material testing risk without stronger current company-specific evidence.
- -3 to -5Material product line or institution likely depends on current animal-testing systems.
- -6 to -8Animal testing is structurally embedded in the business or institution.
-5 to 0
Youth Capture
Manipulative youth-directed marketing, youth-data or attention monetization, collectible/add-on traps, parasocial or identity capture, surprise mechanics, or treating young people as lower-disclosure consumers. Making useful toys or welcoming children is not itself penalized.
Child-oriented usefulness is not a penalty. The penalty is for manipulating young people through data extraction, attention capture, collectible traps, parasocial/IP pressure, surprise mechanics, or lower disclosure.
-5 to 0
Subscription Capture
Manipulative recurring-payment, automatic-renewal, cancellation-friction, bundling, trial-conversion, or refund designs that profit from inertia or confusion.
This is the lock-in axis for recurring-payment traps and cancellation friction.
-5 to 0
Shrinkflation
Reducing product quantity, count, usable volume, or package contents without a proportionate price reduction, especially when the change is not clearly announced at the shelf, package front, or purchase interface. Openly labeled size changes are less severe than quiet downsizing.
A clear front-of-package or shelf notice can reduce severity; quiet downsizing is the core offense.
-5 to 0
Price Manipulation
Opaque, personalized, dynamic, urgency-based, algorithmic, hidden-fee, or market-power pricing that prevents customers from knowing the real price or exploits dependency, scarcity, or information asymmetry. Transparent posted pricing and genuinely public rate cards are not penalized here.
Ordinary price changes are not enough; the penalty requires opacity, manipulation, algorithmic coordination, hidden fees, dependency, or unfair information asymmetry.
-5 to 0
Unilateral Terms
Non-negotiable clickthrough or adhesion terms that turn ordinary customer, user, subscriber, patient, tenant, or public-facing access into a one-sided contract, especially forced arbitration, class-action or mass-action waivers, unilateral modification, venue restrictions, broad indemnity, account termination discretion, or essential-service access conditioned on unreadable terms.
Employment and contractor terms are scored under Labor Sovereignty or Solidarity with the Unemployed unless the same terms also govern ordinary customer access.
-2 to 0
Accountability Opacity
Material opacity, reputation laundering, or hidden accountability structures that prevent public accountability.
Opacity, laundering, or whitewashing can block accountability even when the underlying conduct is also scored elsewhere. Crisis services must clearly disclose active-rescue, police, emergency-service, and confidentiality exceptions.
-3 to 0
Identity Capture
Customer pressure, employee pressure, and pervasive identity saturation.
-1 each for customer pressure, employee pressure, and pervasive identity saturation.
-5 to 0
Surveillance Capture
Invasive surveillance, unreasonably non-optional tracking, facial recognition, biometric identification, or AI behavior scanning of customers, workers, bystanders, or the public.
Includes customer, worker, bystander, and public surveillance when tracking is invasive or not reasonably optional.
-3 to 0
Ideological Disavowal
Concealed ideology presented as neutrality, expertise, professional necessity, public-service administration, market inevitability, or non-ideological common sense while exercising power.
Penalizes neutrality theater, not viewpoint. Openly stated values score better than concealed ideology presented as non-ideology.
Procedure
How evidence becomes a score
- Collect claims.Each claim has a statement, axis, source, confidence, timestamp, status, and revision log.
- Reject unusable submissions.Unsupported, duplicate, abusive, irrelevant, or unmappable claims do not score.
- Use verified inputs.Unverified and disputed claims can be visible, but cannot determine a score.
- Score each axis.Every nonzero axis value should trace to accepted claim IDs.
- Calibrate comparatively.Sorting any column should produce a defensible spectrum with no unexplained inversions.
- Compute confidence.The public confidence label comes from the evidence trail, not from a manual vibe check.
- Publish the trail.Profiles show label, score, confidence, rubric version, scoring judgments, sources, and review date.
Calibration
Comparative Calibration
Axis values are ordinal. A score is defensible only if the entity belongs at that position relative to the others in the same column. Sorting any column should produce a defensible spectrum with no unexplained inversions.
Evidence
Source quality
Source quality affects confidence, not the score itself.
Claims should be checkable and bounded. Strong sources can still be incomplete or misleading; weak sources can still point to real issues. Reviewers evaluate the source, the claim, the axis fit, and the comparative placement separately.
| Source class | Weight | Use |
|---|---|---|
| Primary public record | 100% | Court, regulator, government, military, parliamentary, public-health, or other official public records. |
| Primary institutional record | 90% | Annual reports, audited financials, SEC filings, proxy statements, official policy documents, and organizational bylaws. |
| Independent reporting | 82% | Wire services, established newspapers, public-interest investigative journalism, and other independent newsrooms. |
| Official self-description | 68% | Official websites, about pages, impact pages, sustainability pages, press releases, and other self-published institutional claims. |
| General web source | 58% | Ordinary web sources that are usable but not independently high-authority for scoring. |
| Social or commentary source | 40% | Social platforms, video platforms, newsletters, blogs, and similar sources that usually need corroboration. |
Review
Roles and authority
Bring evidence
Contributors
Contributors submit source-backed claims or challenges. Submissions need structure, a source link, and an axis.
Check evidence
Verifiers
Verifiers inspect sources and decide whether claims are usable, disputed, duplicative, or out of scope.
Synthesize scores
Reviewers
Reviewers connect verified claims to axis values, rationales, confidence, and comparative calibration.
Draft only
LLM Assistance
LLMs may extract, normalize, deduplicate, and summarize. They do not make final scoring decisions.
Specific challenges
Appeals
Appeals should target a claim, source, axis value, calculation, or comparative placement. Decisions are logged publicly.
2036 listed
Launch Scope
The directory currently includes {scored_count} scored entries.
Premise
Corporate privilege
Corporate charters, limited liability, and market privileges are public legal constructs. This rubric asks who controls an institution, who bears losses, who receives surplus, and who is harmed when it scales.
The word company is a plain-language default. The same scoring method can also cover cooperatives, nonprofits, open-source projects, movements, agencies, governments, churches, public leaders, and institutions when people buy from them, work for them, rely on them, fund them, or live under their power.