Assessment · Part A
What it solves
The case for the system existing, made in the terms an informed reviewer will use - with the four claims it cannot support printed on the same page as the five it can.
Most of this site argues against the concept, because that is what keeps it standing. This page is the other half of the same discipline: the argument that it should exist at all, scoped to what the evidence will carry.
Scope. This is a design document. Nothing described here has been built or deployed, and what follows is an argument for building it rather than a report of it working.
A.1 · It fixes a documented failure that has already killed someone
Electronic monitoring in the United States does not fail because the technology cannot locate people. It fails because it locates them so noisily that nobody looks.
| Where | What the field record shows |
|---|---|
| Cook County, Illinois | 95.50% of monthly GPS monitoring alerts in Cook County were false, up from 57.79% under radio frequency. [17] |
| Tennessee | 80% of alerts were never checked by officers in a Tennessee audit. [29] |
| California | 1% of roughly 63,000 monthly California alerts were restricted-zone entries. Restricted-zone entry is the actual event the system exists to catch. [30] |
| Los Angeles County | One overnight shift logged 135 alerts across 65 offenders and produced zero officer responses. [30] |
Hwang, Maniar, Smith & Nivens (2026), "False Alerts Related to GPS Electronic Monitoring in Illinois Cook County Sheriff's Office," Criminal Justice Review; urban GNSS error, Satellite Navigation (2025)
Driven by GPS drift; the worst zip codes were distinguished by tall buildings.
Karsten & West, Brookings (2017)
AP / NBC News (2011), reporting CDCR data for Jan. 2009 - Dec. 2010; Govtech (2014), Los Angeles County alert volumes
GAO-16-10 records where that ends [31]. An offender generated repeated tamper alerts, each of them shorter than the five-minute threshold his regional office used to decide what was worth escalating, so no single alert ever crossed the bar. His officer was never notified. He removed the device, and he committed rape and murder.
The core claim
Every deployed system in this category treats alerting as a notification problem. It is a signal-to-noise problem, and it has never been engineered as one.
Cross-modality corroboration, dwell thresholds and zone classification are ordinary engineering answers, and they are not applied here - not because they are hard, but because the industry has not treated this as the kind of problem they answer. The gap is real, it is documented in published audits, it has a body count, and closing it does not require inventing anything.
This is a stronger position than the framing the concept started with, and a more defensible one. The claim is not that a device stops a person from choosing to offend. The claim is that an alerting system whose output gets discarded protects nobody, that this is an engineering failure, and that it has an engineering answer built here. The mechanism is set out on alert tiers and the modality arithmetic behind it on sensing.
A.2 · Two-sided alerting exists nowhere
Every fielded monitoring system answers one question for one audience: where is this person, told to a supervising agency, on a delay. The people physically present - a parent at a playground, a school office, the adult standing nearest the child - are told nothing. Nothing happens at the moment of the encounter, which is the only moment at which anything protective could happen.
That gap is the original insight of this concept, and it is the one thing in it that survived every research finding. The coverage arithmetic on coverage cut the reach of the claim down to a fraction of the problem. The evidence review on the case removed the recidivism claim outright. The constitutional analysis on legal architecture narrowed the population it can lawfully cover. None of that touched the requirement itself.
The requirement that survived
The alert should reach the people who are actually there, at the time it matters.
Every fielded system in this category fails that requirement by design rather than by accident, because each was built to answer a supervising agency rather than a bystander. The requirement is legitimate, it is unmet, and it is the reason the concept is not simply a rearrangement of what already exists.
A.3 · The political position almost nothing in this category can occupy
Proposals in child protection normally force a choice between safety and civil liberties, and then lose whichever constituency they did not choose. The safety framing loses the courts and the advocacy organizations; the liberties framing loses the committee that controls the appropriation. Most proposals in this space are killed by the constituency they did not write for.
This design is more protective of the monitored person's privacy than the system it replaces
Current GPS monitoring builds a continuous, queryable, breachable movement history of every monitored person, held by a supervising agency or a private vendor and browsable by insiders. That history is the product. It is subpoenaed, it is breached, and it is read by people with no business reading it.
This design holds none of it. Routine operation produces no location history at all. Identity resolution requires a court order together with an independent key custodian, and writes an audit record when it happens. The isolation and data flow pages set out what is and is not retained at each step.
That is not a talking point. It is a structural property: the record that would be browsed is never written, so there is nothing for a policy to protect, nothing for an insider to reach, and nothing for a breach to take.
The consequence is commercial and legislative rather than technical. The same proposal can be defended to a child-safety committee and to a civil-liberties organization without changing the pitch - the identical document, in both rooms, with no line that has to be left out of one of them. Very little in this space can do that, and it materially improves the odds of a bill surviving the room it is drafted in.
The position is only available while the design keeps its side of it. A program that quietly begins retaining routine location data has not weakened a feature; it has forfeited the argument that made it fundable. That is why retention is written as a constraint on the system rather than as a setting in it.
A.4 · The architecture on its merits
Set aside the political argument entirely. Five properties of the design would be worth having whoever was buying it.
Anchored zones move the hard work off the body
Every constraint that makes a capable wearable impossible - power budget, antenna count, calibration, knowledge of the local environment - is relieved by putting the intelligence in a mains-powered fixed device. The wearable becomes the simplest object the problem allows: a thing that broadcasts. The reasoning is set out on anchored zones and the wearable that results on the bracelet. This is the right shape for the problem, and it is the shape the category has not taken.
Split custody makes the privacy guarantee structural
The operator holds observations without keys; the custodian holds keys without observations. An insider at either organization, acting alone, cannot produce an identity. That is a guarantee which survives a dishonest employee, and a policy promise is not. The split is specified on isolation.
The database that does not exist cannot be breached
The most reliable security control available is the absence of the asset. It also removes the single most common way systems in this category generate a scandal: not a failure of the alerting, but a disclosure of the history the alerting happened to accumulate.
The child-side device is designed against its own category
The children's tracker market has a nine-year unbroken record of catastrophic security failure: default passwords, plaintext traffic, silent microphone activation, and shared backends behind millions of units.
- About 600,000 children's tracking devices were found using the default password 123456, with all traffic in plaintext HTTP.Avast / Martin Hron, Shenzhen i365 tracker research (2019) [34]
- 70+ tested children's tracking devices - tens of millions of units - funnel through just three shared backends.Stykas & Solferini, Black Hat (Aug. 7, 2026)
An attacker could locate a watch, alter emergency contacts, falsify location, and activate microphones and cameras with no visible indication. [35]
A receive-only device with no microphone, no camera and no location history is not vulnerable to any of it, because it does not hold the capability being exploited. That is not hardening; it is the removal of the attack surface. The device is specified on devices.
Graduated response protects the monitored person too
A system that requires two sensing modalities to agree, plus dwell time, plus zone class, plus a human decision, before anything public happens is a system that produces far fewer false accusations than one that fires on a single noisy reading. Every threshold on alert tiers that makes the system quieter also makes it slower to accuse.
That is a benefit to the person wearing the device, and it is worth saying out loud in a courtroom. A scheme that can show it was engineered to avoid falsely marking the person it monitors is arguing about its own character on far better ground than one that cannot.
A.5 · Commercial and legal strengths
The buyer exists and already has a budget line
$35.96 per day was the cost of GPS supervision against $27.45 for traditional supervision.Gies et al., NIJ (2012), California GPS supervision evaluation
This is not a new category requiring a new appropriation. It is a replacement for something a supervising agency is already paying for, sold against a line that already exists.
The compliance effects are real and replicated
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]
Those are large effects on outcomes agencies are measured on, and they can be claimed without overclaiming on recidivism - which, in the same study, did not move.
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.
Anchor deployment is a moat
Per-site relationships with school districts and parks departments are slow to build and slow to displace. Whoever builds that footprint first is difficult to dislodge - which is also why the unmodeled cost of building it is the first entry on unsolved.
Scoping to supervised populations makes it buildable now
No post-discharge constitutional question, no retroactivity fight, no waiting on a court. Within a term of supervision the authority already exists in most states, so a pilot does not require new law to begin. The reasoning is on legal architecture.
Honesty is a competitive advantage here
This is a category in which vendors state that their tamper technology "virtually eliminates false alerts" while the published field data shows 95.50% of monthly GPS monitoring alerts in Cook County were false, up from 57.79% under radio frequency.[17] A specification that publishes its own false-alert rate, by tier, is differentiated by the act of publishing it.
A.6 · What it actually solves, stated precisely
Table A.6 of the assessment, reproduced whole. The verdict column is the part that matters: it is the difference between a specification and a brochure, and every row below the fifth is one an informed critic will raise.
Supportable Plausible Not demonstrated Not claimed
| Claim | Supportable |
|---|---|
| Makes zone alerts actionable instead of ignored | Yes - core claim |
| Notifies people physically present at the moment of an encounter | Yes - novel |
| Eliminates the standing location database current monitoring requires | Yes - structural |
| Reduces absconding and registration violations | Yes - replicated |
| Reduces false accusations against monitored people | Yes - by design |
| Deters approach to protected places | Plausible, unproven |
| Reduces new sexual offending | Not demonstrated in any GPS study |
| Protects children from sexual abuse generally | No - see B-5 |
| Protects children from abuse by family or household members | No - the system is blind to it |
| Addresses online grooming | No |
How to read the table
The first five rows are a real product. The last four are what an informed critic will attack, and the only defense is to have said them first.
The sixth row sits between the two and is the one to watch. Deterrence is the claim a reader is most likely to supply on the system's behalf and the claim a pilot is least likely to establish, so it is written here as plausible and unproven, which is what it is. The four rows beneath it are not modesty; they are the boundary of the design, and each one has a numbered entry on unsolved explaining why no version of this architecture reaches it.
How this works
1:05 · 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 | The other half | Most of this site argues against the concept, because that is what keeps it standing. This page is the other half of the same discipline: the case for it existing at all. |
| 0:13 | The documented failure | Monitoring does not fail because it cannot locate people. It fails because it locates them so noisily that nobody looks. In Cook County, ninety five percent of alerts were false. |
| 0:25 | The rare position | This design is more protective of the monitored person's privacy than the system it replaces, because routine operation writes no location history at all. |
| 0:34 | Why that matters | That is structural rather than rhetorical. The same proposal can be defended to a child safety committee and to a civil liberties organization without changing the pitch. |
| 0:44 | Table A point six | Table A point six draws the line. Five claims are supportable, one is plausible and unproven, and four are not claimed at all. |
| 0:52 | The four it will not claim | It does not reduce new sexual offending, does not protect children generally, is blind to abuse inside a household, and has nothing to say about online grooming. |