Courts & Constitutional LawMarkets & FinanceSoftware & Platforms

Meta's Platforms Ruled a Public Nuisance: What Wasn't Fixed

A New Mexico court found the design harmful, then named Section 230 among the reasons it could not order that design changed.

Josh LynwoodFounder
Published
Read time
15 min
Share
Where Things Stand

One Test Applied Twice, to the Claim and Then the Remedy

The order runs 1,829 lines. A jury had found 75,000 violations of the state's Unfair Practices Act. At paragraph 79 the court found Meta's platforms "a cause of and substantial contributing factor to the youth mental health crisis in New Mexico". What follows does two things that are easy to read as one. At paragraph 107 the court concludes that "47 U.S.C. § 230(c)(1), which provides that an interactive computer service shall not be treated as the publisher or speaker of information provided by another information content provider. The New Mexico court applied the three-element test its Court of Appeals adopted in Woodhull v. Meinel, together with the duty-focused formulation from Barnes v. Yahoo!: the question is whether the duty the plaintiff seeks to enforce derives from the defendant's status or conduct as a publisher or speaker. does not protect Meta from the State's An unreasonable interference with a right common to the general public. The State brought the claim under both NMSA 1978, § 30-8-1 and the common law. Whether the doctrine reaches products rather than land use is the live question in this litigation nationally, and it is where the Oklahoma Supreme Court set aside a $465 million opioid judgment in 2021. claim," because the claim "seeks to impose liability on Meta for the impact of various platform designs and features, rather than third-party content". Fifty-six paragraphs later the same statute reappears on the other side of the case. At paragraph 163, autoplay, infinite scroll and algorithmic content recommendations are "most closely tied to content presentation" and are therefore "features with clear Section 230 and First Amendment implications." At paragraph 165, push notifications and like counts are "least connected with platform content." The first group was left alone. The second was ordered restricted. Nothing about that sequence is a reversal, and the order supplies the reason itself: the test the court adopted at paragraph 105 asks whether the duty being enforced derives from the defendant's status as a publisher, and it was applied twice, to two different objects, first to the State's claim, then to the State's requested commands.

  • What Was Ordered. Three problematic-use mandates, all keyed to under-18 accounts. Push notifications eliminated from 10:00 PM to 7:00 AM daily and from 8:00 AM to 3:00 PM on academic-year weekdays, with carve-outs for connected-user messaging and urgent security alerts (paragraph 168). Like counts hidden by default, overridable only with a parent or guardian's permission (paragraph 169). Usage is also restricted to "not more than 90 hours of use per month cumulatively across Facebook and Instagram" (paragraph 170). Alongside them: a nine-part age-assurance program, connection and messaging restrictions, a prohibition on sexualized chatbot interactions with minors, and semiannual compliance reports for five years.
  • What Was Not. "The Court grants no abatement relief relative to the design and implementation of Meta's algorithms" (paragraph 173). Autoplay and infinite scroll were not touched. Robust age verification, which the court had just called "the key to making Meta's platforms safe for adolescents" - was foreclosed not by Section 230 but by The Children's Online Privacy Protection Act, 15 U.S.C. §§ 6501-06, and the COPPA Rule. The Rule defines collection broadly enough to include passive tracking of a child online, which is why the court concluded it could not order Meta to collect the data that more accurate age verification would require, the remedy it had just called the key to the problem.: the court "cannot order Meta to request children to submit personal data or be passively tracked online, even for age-verification purposes" (paragraph 135). A Child Safety Monitor was refused as "extremely vague and, often, aspirational" (paragraph 196).
  • Two Constraints, Not One. Paragraph 166 states the remedial test conjunctively: the court could find no equitable abatement method "(a) that impacts only Meta, rather than imposing restrictions industry-wide; and, (b) that respects the protections of the First Amendment and Section 230". Limb (a) does separate work. Paragraph 164 records that autoplay and infinite scroll "are used widely in the industry" and that, "given the absence of Meta's competitors in this litigation," restricting them "could harm the viability of Meta and its platforms." Section 230 set the boundary; the single-defendant problem widened the no-go zone on the near side of it.
  • The Register Difference. At the liability stage the court quotes the statute, applies the three-element test New Mexico adopted in Woodhull v. Meinel, works through three out-of-state authorities and concludes. At the remedial stage there is no Section 230 analysis at all, no statutory text, no test applied, no authority cited, and no First Amendment analysis of any kind. The verbs change with it: "mindful" (paragraph 82), "could violate" (paragraph 150), "possibly running afoul" (paragraph 160), "implications" (paragraph 163), "would likely violate" (paragraph 172). Read most naturally, the court declined to decide a constitutional question it did not need to reach, having already identified an independent equitable ground at paragraph 171. It is not a second holding, and nobody should report it as one.
  • The Counter-Case, Named and Dated. The defense-side argument is that the order is half right, and the remedial half is the right half. Cory Andrews of the Washington Legal Foundation argued in a July 8, 2026 paper that "[h]ow a feed is composed, sequenced, and served is the very editorial activity the First Amendment protects," and that relabeling editorial judgment as product design is wordplay. On that view the error, if any, is at paragraph 107. In Doe v. Grindr, decided February 2025, the circuit that decided Lemmon v. Snap held Section 230 did bar design-defect claims, because they "necessarily implicated Grindr's role as a publisher of third-party content". WLF files on the defense side of these cases.
Sources3See all 24
Which features could the court actually reach?

The Order Reached the Features Furthest From Content

Three problematic-use mandates were entered, all keyed to under-18 accounts. Under paragraph 168, push notifications are eliminated from 10:00 PM to 7:00 AM daily and from 8:00 AM to 3:00 PM on academic-year weekdays, with carve-outs for connected-user messaging and urgent security alerts. Under paragraph 169, like counts are hidden by default and overridable only with a parent or guardian's permission. Paragraph 170 sets usage at "not more than 90 hours of use per month cumulatively across Facebook and Instagram." Alongside those sit a nine-part age-assurance program, connection and messaging restrictions, a prohibition on sexualized chatbot interactions with minors, and semiannual compliance reports across five years.

Several requested measures were not entered. Paragraph 173 records that the court "grants no abatement relief relative to the design and implementation of Meta's algorithms." Autoplay and infinite scroll are not among the ordered measures. Paragraph 196 declined a Child Safety Monitor as "extremely vague and, often, aspirational." Paragraph 135 states that the court "cannot order Meta to request children to submit personal data or be passively tracked online, even for age-verification purposes," attributing that limit to COPPA. The order's own sorting line sits at paragraphs 163 and 165: autoplay, infinite scroll and algorithmic content recommendations are described as "most closely tied to content presentation," while push notifications and like counts are "least connected with platform content."

A second constraint sits alongside the first. Paragraph 166 phrases the remedial test conjunctively, requiring a method "(a) that impacts only Meta, rather than imposing restrictions industry-wide; and, (b) that respects the protections of the First Amendment and Section 230." Paragraph 164 records that autoplay and infinite scroll "are used widely in the industry" and that, "given the absence of Meta's competitors in this litigation," restricting them "could harm the viability of Meta and its platforms." Paragraph 160 adds that regulating such industry-wide features would "require legislative or executive branch action." On the order's own account, the reach of a single-defendant case is what set the perimeter.

Sources4See all 24
Intersections

The same recommender landed on both sides of the line

Software & Platforms. Algorithmic content recommendation drew no relief at all (paragraph 173). Algorithmic people recommendation was constrained hard: "in no circumstances may Meta recommend an account belonging to a user under 18 years of age to an unconnected account belonging to an adult" (paragraph 143). The court's own taxonomy at paragraph 146 separates "algorithms to recommend content and people." In engineering terms it constrained who may enter whose candidate set and refused to specify the ranking objective, a retrieval-layer eligibility filter, not a change to the model. WhatsApp is the control case: paragraph 148 excludes it from the entire decree because it "does not recommend content or people."

Courts & Constitutional Law. The order's own precedents make the same cut. It relies on Anderson v. TikTok, where the Third Circuit found no immunity because the algorithm was "the platform's own expressive activity" - a premise that, if accepted, strengthens First Amendment protection for that same activity when a court considers ordering it changed. The order is structurally consistent with that, declining to reach the algorithm partly on First Amendment grounds at paragraph 172. The court also recorded at paragraph 106 that it "has not found binding authority that is analogous" - everything at the liability stage is persuasive out-of-state authority, none of it binding on review.

What would make this wrong

Meta implements the under-18 connection limits by changing the ranking model rather than by filtering who enters a candidate set. That would break the retrieval-layer reading.

Open question

On the engineering side: connection eligibility was orderable and content ranking was not, so where does retrieval sit, the layer that feeds the ranking model?

Sources1See all 24
The Weave

The Weave maps a single development across domains and across time. Each row follows one domain from where things stand now through the next eighteen months, and expands for the reasoning behind that trajectory.

Wiiver
SECTOR / DOMAINclick a domain to expand
As It Standsthe current status
Immediate0–6 months
Near-Term6–18 months
Government + Policy
Section 230 does not bar the nuisance claim. The same order names it among the limits on relief, with no analysis.
One Test, Two Objects
Because it was applied to two different objects: the State's claim, and the State's requested commands.
Three Courts, Four Documents, One Split
Three courts reviewing the same four documents split on crime-fraud; the District is the outlier.
Business + Markets
A $567 million abatement fund is ordered prefunded. The market-share figure behind its reduction is nowhere in the order.
The Method Published, the Inputs Withheld
It published the method and withheld the inputs, including the market-share discount percentage.
Two Instruments, Not One Number
No. One is a jury penalty for past conduct; the other is a mandatory injunction to prefund a program.
The court adopts Meta's proposals in large measure and freezes the litigation-era changes against rollback for five years.
Privilege Pierced in One Forum of Three
Privilege over the same four communications has been decided three ways in three forums.
The Perimeter Was the Court's
In substantial part the defendant's own, with a perimeter the court set and terms neither party proposed.
Technology + Engineering
No abatement relief reaches the content recommender. The decree limits who is recommended to whom, not how content ranks.
Three Paragraphs, Three New Services
Three short paragraphs imply three systems that do not exist today. None is a configuration change.
One Recommender, Both Sides of the Line
Content recommendation drew no relief; connection recommendation drew binary eligibility limits.
wiiver.co · 4 impacted domains shownWiiverv1 · August 10, 2026
Looking Forward

A Bond Could Stop the Clock Before the First Report Comes Due

Five markers sit in the forward window. Each resolves against a document that will be filed or will not be, and nothing here predicts which way any of them goes.

  • The Bond and the Clock. Meta has said it will appeal, and under paragraph 131 the five-year Abatement Period "shall be tolled during the pendency of any appeals, provided that Meta pays a Security a party posts to stay enforcement of a judgment while it appeals. Under paragraph 131 of this order the bond does more than that: it tolls the five-year Abatement Period, so the compliance clock stops rather than running through the appeal. The order sets no amount.". The bond buys time on the compliance clock rather than merely staying collection; a defendant that does not post one faces the clock running through the appeal. The order sets no amount. As of the morning of August 10 no notice of appeal was confirmed filed on any source reached, a limit on what could be retrieved, not a statement about the docket.
  • Oakland, This Week. Jury selection begins Wednesday in the trial brought by California, Colorado, Kentucky and New Jersey before Judge Yvonne Gonzalez Rogers in Oakland, with opening statements on August 18. The jury's role there is advisory; Gonzalez Rogers decides any penalty herself, and the trial is expected to run six weeks. The second court to reach this theory will reach the remedy question the way New Mexico did, with a judge deciding what can be ordered.
  • The Baseline Measurement. The order's only near-term hard deadlines fall thirty days out. Paragraph 191 requires Meta to communicate with the National Center for Missing & Exploited Children about improving its CyberTip reports, and paragraph 194 requires it to report its current rate of new-CSAM detection, the Baseline Measurement against which a 5 percent improvement is later measured. That number makes the child-safety mandates auditable, and the order assigns the party under the obligation to supply it.
  • The First Compliance Report. Paragraph 132 requires written reports to the court and the State by June 30 and December 31 of each year of the Abatement Period. The first would be due December 31, 2026, and would arguably toll with the Abatement Period if a bond is posted. It is the first document to show how Meta defines "use" for a 90-hour monthly budget, what counts as "the academic year," and which messages qualify as "urgent, targeted." Those definitions are written by the party they bind.
  • The District's Own Case. Washington's case against Meta is a Consumer Protection Procedures Act case filed in October 2023, part of the 42-attorney-general coalition, and it is still in discovery. On July 16, 2026 the D.C. Court of Appeals denied Meta's An extraordinary writ directing a lower court to act or refrain from acting. A petitioner must show a clear and indisputable right to the writ, which is a far higher bar than showing the lower court was wrong, which is why denying a petition decides nothing about whether the underlying ruling was correct. petition seeking to vacate two orders compelling production of four internal communications under the A doctrine under which attorney-client privilege is forfeited where there is probable cause to believe the communications were intended to advance, and did further, a crime or fraud. It is an evidentiary determination made for the purpose of deciding whether documents must be produced. It is not a finding of fraud and not an adjudication of liability. to privilege, holding that Meta had not shown a "clear and indisputable" right to the writ, and stating expressly that it did "not decide whether the trial court's crime-fraud ruling was legally and factually correct". Two other courts reviewing the same four documents reached the opposite result.
Sources8See all 24

Every issue

  1. 01Intersection of the week
  2. 02Impact of the week
  3. 03The week in review

Wiiver Weekly

One free email, Saturdays at 7:00 AM ET.

Unsubscribe anytime.

Sources and Verification
16 of the 24 sources cited here are primaryfilings, opinions, statutes and agency releases read directly
Primary sources16
Secondary sources, by sector8
Government + Policy3
Business + Markets4
Technology + Engineering1
v2 · Reviewed by Josh Lynwood · August 10, 2026
Back to top