Field Notes

question

Does age verification reduce harms other measures cannot?

StatusWorkingUpdatedStands on4 sourcesSupportsnotes

The wiki’s age-verification notes document weaknesses: X-rated compliance theater shows deployed systems bypassable at low cost, Age-verification circumvention services maps the workaround economy, and Virtual private networks and child protection EPRS records the VPN displacement problem. What the vault lacks is evidence for the other side of its own proportionality test: whether mandatory verification reduces any identified harm by more than the less intrusive measures could, after its privacy and access costs are counted.

Exposure is not that harm. The vault’s evidence treats exposure as heterogeneous in effect, so counting blocked encounters answers a proxy question. Exposure change still matters as the mediating link: a gate that does not change what minors actually see cannot change any harm that runs through what they see.

Why it matters

Alternatives to age verification demands that a gate’s proponents “estimate the reduction the gate can actually achieve.” The demand cuts both ways: the vault cannot currently supply that estimate either, so its skepticism rests on bypassability and cost, not on measured outcomes. A leaky gate could still shift first exposure to later ages or reduce casual and accidental encounters, and whether that residue matters depends on which harms are age- and exposure-mediated at all. The comparison the proportionality test demands needs effectiveness estimates on both sides; Does pornography literacy improve adolescent outcomes queues the other half.

What is already known

  • Deployed systems in four European countries were heterogeneous and frequently bypassable under low-cost attacks.
  • Circumvention is commercialized and VPN use spikes when national age-gating takes effect.
  • Compliant major platforms report large traffic drops after enforcement begins, but traffic is not exposure, and exposure is not harm: Displacement to less accountable sites could leave exposure similar while worsening the material encountered. A traffic drop measured only at the gated platforms is a measurement at the controlled point, which is precisely the observation that cannot distinguish a reduction from a relocation.
  • The first post-enforcement UK snapshot exists. In the Internet Matters Online Safety Act report, two months after the Online Safety Act’s children’s codes took effect, 53% of children had recently been asked to verify, 32% had bypassed a check, and 49% reported a harmful experience in the past month, including explicit sexual content for 8% of those 13 and over. It is cross-sectional self-report with no pre-Act baseline, so it bounds the Act’s early effect without measuring an exposure change.
  • No evidence in the vault connects an enforcement event to a change in any measured harm.
  • Children’s circumvention behaviours online, the UK government’s commissioned study, documents the displacement and post-bypass harm pattern at scale but not the causal chain: 53% of children chose platforms to avoid checks, 51% of bypassers encountered harmful content after bypassing, and explicit sexual content was the harm type most likely to be seen only after bypassing. Most harmful-content encounters were passive scrolling rather than active search. It remains cross-sectional self-report, so it maps where harms sit relative to the gate without measuring whether the gate changed them.

What would settle it

The causal chain has three links, each needing its own evidence:

  1. Does enforcement change minors’ actual exposure — age at first exposure, frequency, and content mix — not just compliant-site traffic?
  2. Does any exposure change translate into a change in an identified harm: the aggression pathway tracked in Does pornography use cause adolescent sexual aggression, use experienced as compulsive or distressing, or script and attitude effects?

  3. Is that harm reduction larger than what literacy, design, or education achieve without the privacy and access costs of identity infrastructure?

  4. Find natural-experiment studies of national enforcement events (UK Online Safety Act enforcement from 2025, United States state laws, the French and German regimes) measuring minors’ self-reported exposure and its content mix.

  5. Find whether any regulator publishes outcome data on minors’ exposure or wellbeing rather than deployment compliance.
  6. Represent the strongest empirical pro-verification case at full strength before EU age verification framework 2026 obligations begin to bind.

Built on 4 sources (4 external).

Working out connections…