Part IV
Open questions
What the specification does not resolve, and what has to be checked before anything here is filed or funded.
This page is the part of the document that argues against it. It is reproduced from Part IV of the source specification with nothing softened, because an unresolved question that has been rewritten as a milestone has stopped being a question.
Scope. This is a design document - design and research, not legal advice. The constitutional analysis on this site summarizes reported decisions and identifies where courts disagree; it does not predict how any particular court would rule on any particular program. Before drafting legislation or committing capital, this needs review by counsel admitted in the target jurisdiction with practice experience in this area.
Unresolved design and policy questions
Five questions the design does not answer. The first is the one that decides whether any of the rest matter.
Anchor coverage economics
The system is only as useful as its anchor density, and anchors require a per-site relationship with school districts and parks departments - one at a time, negotiated, maintained, and renewed. No coverage model exists yet. This is the largest unmodeled risk in the concept, and it is a risk of the kind that does not appear in a pilot, because a pilot buys its own anchors.
Who pays
Three funding models, each with a defect that is already documented somewhere else. Offender-paid fees drive non-compliance among indigent registrants and are widely litigated. Agency-paid requires an appropriation, which means it competes against everything else the agency does. Site-paid requires school districts to fund it, out of budgets that are not built for a hardware program with a maintenance tail. Unresolved.
Online grooming is not addressed
The scenario this concept is most often described with - a groomed twelve- to seventeen-year-old - is now largely a digital-contact problem. 1.4 million online enticement reports were recorded by NCMEC in 2025. [37] A proximity system does not touch it. Not partially, not indirectly: an anchored zone is a fact about physical space, and the contact this scenario turns on does not happen in physical space until it is already over.
NCMEC CyberTipline data (2025)
NCMEC attributes much of the rise to expanded mandatory reporting under the 2024 REPORT Act. This is report volume, not measured incidence.
Measurable effectiveness
No design on this site proves the system reduces offending, and nothing here should be read as claiming it does. The honest expectation, from the evidence base this document rests on, is that measurable effects appear in compliance and absconding rather than in new offenses - which is exactly the pattern the supervision evaluations already show: 2.1% against 9.9% was the absconding rate for GPS-supervised against comparison cases, and failure to register ran 3.2% against 13.2% - while new sex crime convictions were 1.1% in both groups. [24]
Turner, Chamberlain, Jannetta & Hess (2015), Victims & Offenders
A small matched design, 94 against 91, over one year. It is the study that most directly tested new offending, and it found no difference.
The evaluation must be designed before deployment or it cannot be run afterwards. Comparison group, outcome definitions, and the pre-registered analysis all have to exist before the first device is issued. A program that decides after the fact what would have counted as success has not measured anything.
Adults opting in
The concept proposes that any adult who wishes may carry a receiving device. Under the anchor model this is technically straightforward - the receiver is passive and the anchors do not care who is listening. But it expands the population who receive alerts, and therefore the vigilante exposure set out in the threat model. It needs its own analysis, and that analysis has not been done.
To verify before filing or funding
Research for the source document was current through August 2026, and its author flagged the following as unconfirmed. Each should be checked against a primary source before any of this is relied on in a filing or a pitch. They are reproduced here rather than resolved, because the alternative - dropping them, or restating them as settled - is how a research document becomes a brochure.
| Item | What is unconfirmed, and why it matters | Where to check |
|---|---|---|
| Does v. Whitmer, Nos. 25-1413, 25-1414 (6th Cir.) | Argued 21 July 2026. Outcome unknown at the time of writing. The appeal runs directly at the reasoning this site's retroactivity analysis is built on [8], and the analysis on legal architecture should be re-read against the opinion the day it issues. | Sixth Circuit opinions and the docket for both numbers. |
| People v. Lymon (Mich. 2024) · Park v. State (Ga. 2019) · State v. Hilton / Strudwick (N.C. 2021) | Official reporter citations. Each of these decisions is discussed in the research behind this document, and none of the three could be pinned to an official reporter volume and page. They are therefore not in the source list and no figure on this site rests on them. | The official reporter for each state, not a secondary database summary. |
| Ellingburg, applied to registration or monitoring | Whether any court has yet applied the text-and-structure method to a registration or a monitoring scheme. As of writing, none had. If one has since, it may matter more than anything else on this page, because it would change the method a court uses to decide whether a scheme like this one is punishment. | Subsequent-history and citing-references searches, repeated - this is the item most likely to change first. |
Closed since
A revision of the source specification answered two of the items that used to sit in the table above. They are moved out of it rather than deleted, with the answer beside the question. A list of open questions that only ever grows is a list nobody is working through, and one that quietly loses its entries cannot be audited at all - the useful version shows both the question and what happened to it.
| Item | What was unconfirmed | How it was resolved |
|---|---|---|
| 28 C.F.R. pt. 72 Closed | Whether the part was substantively amended in 2026. eCFR reports "last amended 7/31/2026", and no corresponding Federal Register rule could be located - leaving it unclear whether the federal registration baseline described here was out of date. | It was not substantively amended. The eCFR date is a title-wide currency stamp rather than a statement about this part, which is precisely why no matching Federal Register rule exists. The absence of the rule was the answer, not a failed search, and the baseline described on this site stands as written. |
| Levenson & Zgoba (2016), Florida prior-conviction share Closed | The figure was obtained secondhand rather than from the paper, and this site printed the authors in the wrong order - as Zgoba & Levenson [5]. A secondhand statistic is a statistic whose method nobody here has read. | The paper is Levenson & Zgoba (2016), the author order is corrected, and the figure is taken from it rather than from a restatement: 6.5% of Florida sex crime arrests involved someone with a prior sex crime conviction - that is 93.5% first-time. It now carries the New York first-time-offender finding's replication on the case. |
Scope of this document
This is a design and research document, not legal advice. The constitutional analysis summarizes reported decisions and identifies where courts disagree; it does not predict how any particular court would rule on any particular program. Before drafting legislation or committing capital, this needs review by counsel admitted in the target jurisdiction with practice experience in this area.
What this site still owes
The two lists above belong to the source specification. This one belongs to the site: things asserted here without a source behind them, or without a check that runs. The back-reference table is where most of it is visible. It is written down because a reader has no way to find them otherwise, and because a build that gates most of a page and says nothing about the rest is quietly claiming the rest is fine.
21 of the 49 figures in the facts table cite a source in prose rather than by key into the source list.
They are the case-law and victimization figures carried over from this site's first draft. Each one names its source in the line beneath it, so nothing is unattributed - but a prose citation is not checked by the cite-resolves gate, because there is no anchor for it to resolve to. Until they are keyed, that part of the sourcing is checked by reading, not by the build.
The registry population figure has two different provenances on this site.
It is attributed to a national survey by the National Center for Missing & Exploited Children, and the source list records that the same organization discontinued that survey and that the current number comes from a private compilation. Both statements cannot be the best available description of where the figure came from. One of them has to go.
Several sources in the appendix are not cited by any page yet.
They are listed as not yet cited on the sources page rather than hidden, and that list is rebuilt from the page modules every time the site is generated. A source nobody cites is not a problem in itself; a reference list that conceals which half of it is load-bearing is.
The claim that no frame of footage on this site shows an identifiable child is asserted, not measured.
Every clip was reviewed as a still before selection and the generated clips were prompted against people entirely, which is a process, not a check. The build measures the footage for contrast, for seam continuity and for decodable length; it cannot measure what is in the frame. That claim rests on human review and is written down here so it is not mistaken for a gated one.
The coverage and cost models on this site run on assumptions with no citation behind them.
Anchor density, encounter rates and dispatch minutes per call are engineering estimates chosen so the instruments have something to compute. They are labeled in the instruments themselves and every one of them is a knob the reader can move, which is the honest form for an unsourced number - but it is still an unsourced number, and the anchor coverage question at the top of this page is the reason none of them can be closed yet.
No page states what evidence would count as this system failing.
The measurable-effectiveness question below says the evaluation has to be designed before deployment. That design does not exist on this site. Until it does, the honest reading of every claim here is that it describes what the system is meant to do, not what it has been shown to do.
Everything in this list is visible from the build itself: the facts table records which figures cite by key and which cite in prose, the sources page rebuilds its own list of uncited sources every time the site is generated, and the gate suite prints the names of the checks it ran. Nothing here required inside knowledge, which is the point - if it did, it would not belong on a page about what cannot be verified.
How this works
1:04 · narrated · captioned · nothing here is sent anywhere when you press play. Prefer to read it? The transcript is below.
Transcript
| At | Section | What it says |
|---|---|---|
| 0:02 | What this page is | This is the part of the document that argues against it. It reproduces the source specification's open questions with nothing softened. |
| 0:10 | Why nothing is softened | An unresolved question rewritten as a milestone has stopped being a question. So none of these is restated as a phase, a next step, or a thing already in hand. |
| 0:21 | The largest one | The largest unmodelled risk is anchor coverage. Anchors need a negotiated relationship with each site, and a pilot hides that, because a pilot buys its own. |
| 0:30 | Flagged as unconfirmed | A second list holds what the research itself could not confirm. A pending appeal, decisions that could not be pinned to an official reporter, and where to check. |
| 0:40 | Closed, and kept | Two items have since been answered. They stay on the page, marked closed, with the answer beside the question. A list that quietly loses entries cannot be audited. |
| 0:50 | What the site owes | A third list belongs to the site rather than the source. Figures cited in prose, assumptions with no citation, and one claim that rests on review, not on a gate. |