Public methodologyVersion rr-posture-v1Last reviewed May 6, 2026

From public evidence to a posture band.

This is the reading aid for every product page in the reference library. It shows which signals were looked for, how those signals were weighted, and how much confidence Rooted Reality has in the evidence base — for one platform, on one review date.

See a worked example →Read the rubric

01 · Source

Public disclosure

02 · Pattern

Named evidence pattern

03 · Axis

Score 0–100

04 · Composite

Weighted blend

05 · Band

Early · Emerging · Capable · Independent

What the rubric measures.

Four axes turn public evidence into a single composite score. Youth protection carries the largest weight because the library exists for parent decision-making — every other axis matters, but not equally.

Moderation

20%

How clearly a product explains and enforces rules for user-generated content and community behavior.

Signals include

Codified content rules
Recent transparency report
Recent moderation audit

Privacy

20%

How clearly a product explains data practices, privacy settings, and protections for younger users.

Signals include

Detailed privacy policy
Recent privacy audit
Minor accounts default private

Security

15%

How clearly a product describes account security, reporting paths, and protections against misuse.

Signals include

Two-factor authentication
Encryption in transit
Bug bounty program

Youth protection

45%

How clearly a product accounts for age, teen experience, family controls, and youth-specific safeguards.

Signals include

Teen account defaults
Family supervision tools
Regulatory youth-harm finding

Why the heavier hand: a strong story on the other three axes cannot, on its own, make a platform appropriate for a younger child. The library is built for parents of kids and teens — the weighting reflects that.

The weighted composite, set out in plain sight.

Each axis is scored from 0 to 100 against the evidence patterns above. Those four scores combine with fixed weights — printed here, not hidden in a function — to produce a single composite.

Composite score · 0–100

composite=M×0.20+P×0.20+S×0.15+Y×0.45

Where M moderation, P privacy, S security, Y youth protection — each in 0–100.

The composite is the only number the rubric uses to decide a band. Per-axis scores still appear on every product page, so a reader can see where the composite came from, not just what it is.

Four bands, with the door clearly marked.

The composite maps to one of four posture bands using fixed thresholds. The band is a parental decision-support signal, read against everything else a family knows about their child. A higher composite earns a younger-friendly band because more public evidence supports a younger reader.

Band thresholdscomposite 0 ← … → 100

Age 15+

Independent

composite < 60

Age 12+

Capable

60 ≤ composite < 75

Age 9+

Emerging

75 ≤ composite < 85

Age 6+

Early

composite ≥ 85

Bands are deterministic. The same scores produce the same band every run — there is no per-platform override and no human re-read at this stage. If a band surprises a reviewer, the answer is to edit the rubric or the axis criteria, not to override the output.

A floor on youth protection.

One gate sits in front of the band logic. If the youth-protection score is below 50, the final band is capped at Independent regardless of how strong the other three axes look — a thin youth-protection evidence base cannot be carried by moderation, privacy, or security.

Gate · Youth-protection floor

Y < 50 ⇒ band ≤ Independent.

The gate fires before the composite is mapped to a band. When it fires, the composite is still printed in the trace — a reader sees what the score would have produced without the floor, and why the rubric chose otherwise. The floor cannot be overridden by a higher composite.

From a public source to a published band.

Every published score has a chain behind it. A public source fires one or more named evidence patterns; those patterns contribute to an axis score; the four axis scores combine into a composite; the composite — checked against the youth floor — maps to a band.

01 · Source

Cited public source

Vendor release note, policy page, transparency report, audit, regulator finding.

official_policy · third_party_report

02 · Pattern

Evidence pattern fired

A named pattern, defined in the rubric, that this source supports.

TRANSPARENCY_REPORT_RECENT

03 · Axis

Axis score

Patterns contribute to one or more axes; ceilings keep thin coverage honest.

moderation: 72 · medium

04 · Composite

Weighted composite

M × 0.20 + P × 0.20 + S × 0.15 + Y × 0.45 — checked against the youth floor.

composite: 58.0 · gated

05 · Band

Derived band

Deterministic threshold map — same inputs always produce the same band.

band: Independent (15+)

Evidence patterns — the bridge between source and score.

Patterns are named because the logic should be inspectable. A source that contains a relevant statement fires one or more patterns; the per-axis files define how combinations map to score bands and ceilings.

Moderation patternsM

Codified content rules

boost

Required to land above the lowest moderation tier.

Recent transparency report

boost

Combined with codified rules, lifts the moderation ceiling above 75.

Stale transparency report

neutral

Holds the moderation ceiling around 65.

Recent moderation audit

boost

Stacked with recent transparency reporting, lifts the moderation ceiling near 90.

Regulatory moderation finding

drag

Lowers the moderation ceiling to the 40-59 band.

Documented moderation failure

neutral

Lowers the moderation ceiling; stacks with regulatory findings.

Confidence is about our evidence base — not the product.

HIGH, MEDIUM, and LOW describe how complete, current, and directly usable the public evidence was when the rubric ran. They do not describe whether a platform is good or bad.

HIGH

Multiple relevant public sources are available, the evidence is current enough for review, and the patterns behind the score are clear.

MEDIUM

Enough public evidence exists to apply the rubric, but one or more areas have thinner coverage, older sources, or less direct evidence.

LOW

The evidence base is limited, unresolved, or still being reviewed. DRAFT entries remain held at LOW confidence until reviewed for publication.

On DRAFT:A DRAFT entry is not a negative product signal. It means the entry is held for review, kept at LOW confidence until the evidence base is checked. We would rather show a DRAFT than ship an undercooked summary.

A worked example, from sources to band.

One trace of how a public review actually moves through the rubric. Open any step to see the inputs the rubric used and the values it produced.

Variant

Instagram· meta/instagram

Confidence MEDIUM

01Sources cited13 sources›

OFFICIAL_TOSInstagram Terms of Use
Instagram publishes terms of use and community guidelines that spell out what content is allowed. · help.instagram.com
OFFICIAL_TRANSPARENCY_REPORTMeta Community Standards Enforcement Report
Meta publishes a quarterly transparency report covering enforcement actions on Instagram. · transparency.meta.com
OFFICIAL_PRIVACYInstagram Privacy Policy
Instagram publishes a privacy policy describing what personal information it collects and how it is used. · privacycenter.instagram.com
OFFICIAL_HELP_CENTERInstagram two-factor authentication setup
Instagram offers two-factor authentication for all accounts and encrypts data in transit. · help.instagram.com
OFFICIAL_SECURITYMeta Bug Bounty Program
Meta runs a bug bounty program that pays outside security researchers to find vulnerabilities affecting Instagram. · bugbounty.meta.com
OFFICIAL_HELP_CENTERIntroducing Instagram Teen Accounts (2024)
Instagram's Teen Accounts default to private, with stricter content filtering and limits on who can contact the account. · about.instagram.com
OFFICIAL_HELP_CENTERMeta Family Center — supervision tools
Meta Family Center gives parents supervision tools for their teen's Instagram account. · familycenter.meta.com
INTERNAL_LEAKWSJ — The Facebook Files (2021 series)
Internal Meta documents published by the Wall Street Journal in 2021 showed the company was aware of teen mental-health harm patterns on Instagram before publicly addressing them. · wsj.com
REGULATORY_ADVISORYUS Surgeon General Advisory — Social Media and Youth Mental Health
The U.S. Surgeon General issued an advisory in 2023 citing peer-reviewed research linking Instagram to youth mental-health concerns. · hhs.gov
REGULATORY_ACTION41-state AG bipartisan lawsuit against Meta — youth mental-health harm allegations
A group of 41 state attorneys general filed suit against Meta in 2023, naming Instagram in youth-harm allegations. · oag.ca.gov
THIRD_PARTY_AUDITMeta Oversight Board — 2024 Annual Report (Aug 2025)
The Meta Oversight Board's 2024 annual report says Instagram accounted for 22% of the user appeals tracked across Meta's surfaces that year. · oversightboard.com
REGULATORY_ACTIONEuropean Commission — preliminary DSA findings against Meta (Oct 2025)
The same October 2025 European Commission DSA finding cited for Facebook applies to Instagram: users do not have an effective path to challenge content-moderation decisions. · digital-strategy.ec.europa.eu
INVESTIGATIVE_JOURNALISMReuters — Instagram failed to curtail hate speech against women politicians (Aug 2024)
An August 2024 Reuters report on a Center for Countering Digital Hate study said Instagram left 93% of abusive comments aimed at women politicians up after they were flagged. · reuters.com

02Patterns fired16 patterns›

MCodified content rules
+
MRecent transparency report
+
MRecent moderation audit
+
MRegulatory moderation finding
−
MDocumented moderation failure
−
PDetailed privacy policy
+
PMinor accounts default private
+
STwo-factor authentication
+
SEncryption in transit
+
SBug bounty program
+
YTeen account defaults
+
YFamily supervision tools
+
YInternal-document leak (youth harm)
−
YSurgeon General advisory
−
YPeer-reviewed harm finding
−
YRegulatory youth-harm finding
−

03Axis scoresM 30 · P 80 · S 65 · Y 25›

M · 20%Moderation
30
Documented moderation failure or regulatory finding lowered ceiling.
P · 20%Privacy
80
Detailed policy plus audit or minor-default-private; no breach history.
S · 15%Security
65
2FA + in-transit encryption documented; at-rest or audit missing.
Y · 45%Youth protection
25
Internal documentation of awareness combined with active regulatory finding.

04Weighted composite43.0›

Each axis contributes its weighted share to the composite:

M: 30 × 0.20 = 6.0P: 80 × 0.20 = 16.0S: 65 × 0.15 = 9.8Y: 25 × 0.45 = 11.3

Composite = 6.0 + 16.0 + 9.8 + 11.3 = 43.0.

05Floor checkY = 25 < 50 · fired›

The youth-protection floor sits at 50. Y = 25 is below the floor, so the gate fires before the composite is mapped — the band is set by the floor, not by the composite alone.

Floor fired. Composite 43.0 is held back; the band is set by the floor, not by the composite. Reader sees both numbers in the trace.

06Derived bandIndependent · 15+›

The rubric publishes Independent (15+) for this variant, with MEDIUM confidence — 13 public sources, scored against the rubric on this review date.

Youth-protection score 25 is below the floor of 50, ceiling band at "independent" regardless of composite 43.0.

Evidence can move. The trail shouldn't.

Public sources are messy. They get moved, edited, or quietly retired. Two practices keep the review trail durable across hundreds of variants and sources.

Source metadata, captured on ingest.

Every source carries title, vendor, publication date, source kind, and the claim it supports — saved at the time the rubric ran, not refetched on every read.

Archival capture, where supported.

When the source's host allows it, the page is sent to the Wayback Machine via Save Page Now so a future reviewer can recover the evidence as it appeared on the review date.

What this page is not.

The rubric explains a method. It does not make a family's decision, and it does not promise that a platform will behave the same way in every household.

Drawn lines

The rubric stops at the threshold.

Posture is a signal a parent can read, weighted against everything else they know about their child. The library does not replace that judgment — it equips it.

It does not tell a family what to choose for their child.
It does not rank one product above another.
It does not promise future behavior on any platform.
Confidence is about our evidence base, not product quality.
A change in public disclosures can change posture on the next review.
Per-account, per-region, and per-version behavior can differ from the public posture.

Independence and conflicts.

Rooted Reality does not sell placement in the Apps Reference Library. Reviewed entities do not approve or edit posture summaries before publication.

If Rooted Reality has a material relationship with an entity covered in the library, that relationship is disclosed on the relevant page. The public library is designed to be privacy-respecting — no third-party analytics run on these surfaces.

Get methodology updates

Notified when our scoring rubric or evidence rules change. One email per change.