The Decision Gate

When should a promising idea advance, change, or stop?

Five gate checks, a new worksheet, and six program cases worked through under different constraints. Drawn from Paul Ross's dissertation and the Digital Divide Projects and Initiatives archive.

Preface

The moment before money moves

Every digital inclusion effort reaches a point where an idea has momentum, funding is in sight, and someone has to decide whether to go ahead. The dissertation calls that point the strategic decision gate and names it the model's core control point.

A companion volume, From Barrier to Solution, followed a problem from its visible symptom to its cause. This book picks up where that one ends. Once a cause is known and options are on the table, how does a team decide which option deserves to proceed, which needs rework, and which should not go forward at all?

The gate sits in the middle of the dissertation's ten-phase conceptual model. Six phases of diagnosis come before it. Four phases of design, measurement, results, and improvement come after it. Everything the first six phases learn is gathered into a gate evidence packet, and the gate's job is to judge each option against that evidence.

The model is iterative: technologies, eligibility rules, funding programs, threats, and community needs change faster than a static intervention can keep up with, so outcomes loop back into diagnosis.

What is new in this book

The dissertation defines the gate and its tests. This book adds two things for practitioners: a gate worksheet that turns the five checks into a working session, and six program cases that run real project designs from the archive through the gate under different constraints: coverage, building ownership, missing electricity, low confidence, expiring money, and unsettled safeguards.

The worksheet's scoring rules and the case ratings are this book's own applications of the dissertation's model. They illustrate the reasoning; they are not evaluations of the named projects, whose archive records describe documented activities rather than measured results.

Chapter 1

Why a gate at all

Ideas in digital inclusion rarely fail for lack of enthusiasm. They fail because they were chosen before the problem was understood.

The dissertation puts it bluntly: a technology may be fundable, available, or popular, yet not causally appropriate. Its examples are familiar to anyone who has worked in the field. Tablets do not solve an affordability problem if families cannot pay for broadband. A broadband build does not solve a skills problem if users get no support. An online platform does not solve an educational problem if it is inaccessible, culturally misaligned, or too complex for the people meant to use it.

The gate exists to resist what the dissertation calls solutionism: selecting a response because it is ready to hand rather than because it fits the cause. It is placed after diagnosis and before design so that the evidence gathered in the first six phases actually governs the choice.

Three pressures that push ideas through too early

Funding windows. Grants arrive with deadlines and eligible-use lists. The dissertation notes that disparate funding streams lead organizations to build stand-alone projects that meet short-term grant requirements without adding up to a sustainable strategy.
Visible outputs. Devices handed out and miles of fiber laid are easy to count and photograph. Interview participants warned that programs often measure process outputs instead of service outcomes.
Emergencies. The pandemic forced fast decisions, many of them right. Emergency logic, though, tends to outlive the emergency, and short-term subsidies connect families only temporarily.

The gate is also an ethical checkpoint

The dissertation asks the gate to examine privacy, accessibility, trust, community engagement, benefits, burdens, and differential outcomes. Drawing on Alfalah et al. (2025), it frames the core ethical question sharply: does the proposed intervention reduce burden, or merely shift it onto people who already experience exclusion? That question matters more as inclusion work adopts artificial intelligence, learning analytics, automated advising, and platform services.

"And you need to understand what's causing it so you can address those issues."Interview participant QK, on choosing an intervention

What the gate is not

The gate is not a veto meeting and not a formality. Its most common output, in practice, is a better version of the idea. An option that enters the gate as "hand out hotspots" often leaves as "hotspots where coverage tests pass, paired with laptops, setup guides, and an after-hours help line, with a different plan for addresses on the edge of the signal." That is the gate doing its job.

Chapter 2

Before the gate

A gate can only judge what is put in front of it. The dissertation's model includes a strategic pre-gate transition that turns six phases of diagnosis into options worth judging.

The pre-gate transition, as outlined in the dissertation's detailed model description (Appendix AC), moves through eleven steps. They form a sequence: each one depends on the step before it.

Confirm the decision problem and priority gap. State the one gap this decision is about.
Translate evidence into solution requirements. What must any acceptable option achieve, for whom, and where?
Develop alternative solution concepts. At least two real options, not one favorite and a straw man.
Establish the causal pathway for each option. Trace how each option would move people along the pathway from funding to infrastructure, access, skills, adoption, and development.
Develop the preliminary architecture. A first sketch of the technology and services each option needs.
Estimate initial costs and total cost of ownership. Capital and operating costs, including what continues after launch.
Define expected benefits. Outcomes, not only outputs.
Document funding assumptions. Which money pays for what, and when it ends.
Assess risks and safeguards. Privacy, accessibility, trust, security, and burden.
Validate options with stakeholders. Including the people who will live with the result, who the dissertation notes are often the least formally powerful.
Confirm evidence, assumptions, and decision readiness. Name what is still unknown.

The gate evidence packet

What arrives at the gate should be a short, honest packet. Drawing on the pre-gate steps, it should contain:

The validated root cause, with the evidence behind it, and the population it affects, disaggregated where possible. One interview participant urged going below ZIP codes to census blocks where provider data exist.
Each option's causal pathway and the links it leaves unsupported.
Cost estimates that include operating costs after launch.
Funding assumptions with end dates.
A risk and safeguard scan.
What stakeholders said, including people affected by the decision.
A plain list of unknowns.

A packet with a long list of unknowns is not a failure. It tells the gate that the right decision today may be to return for evidence rather than to rule on the option.

Chapter 3

The five gate checks

The dissertation's gate asks five questions. Each looks at the same option from a different side, and a strong answer to one cannot cover a weak answer to another.

Causal fit

Does the option address the established root cause?

This is the check the rest of the gate leans on. The dissertation's evidence is clear that effective policy depends on correctly identifying whether exclusion comes mainly from availability, affordability, devices, skills, accessibility, trust, or a combination. It also ties causal fit to data quality: a wrong diagnosis can aim a technically excellent intervention at the wrong problem.

Evidence to bring

The validated cause and how it was tested
Where the option sits on the causal pathway
Which neighboring links it leaves unsupported

Red flags

Infrastructure money aimed at an area where a network exists but uptake is low
Devices offered where the gap is home service
Training offered where there is no connection to practice on

Opportunity fit

Is there a priority need, and room for meaningful impact?

An option can fit the cause and still be the wrong priority. Opportunity fit asks whether this gap matters most for this population now, and whether the option can change something people will feel. The dissertation also connects opportunity to strategic alignment: does the option serve community and institutional priorities rather than only a funder's list?

Evidence to bring

Which populations carry the heaviest burden of this barrier
What outcomes would change, and for whom
How the option fits local plans and priorities

Red flags

The need is real but already met by another program
Benefits flow mainly to people who were already connected
No one can say what would be different in a year

Feasibility fit

Are there enough resources, partners, systems, and time, now and after launch?

Feasibility covers delivery and, in this book, durability. The dissertation's detailed model description lists Business Value and Sustainability Fit as a further gate test, and interview participants pressed hard on the point: sustainability should be a design criterion from the start, because operating costs persist after installation. To keep five checks, this book folds sustainability into feasibility, and the worksheet gives it its own line.

"So there are op-ex costs that need to be encouraged to maintain it."Interview participant QR1

Evidence to bring

Signed or likely partners, with their roles
Staff capacity to deliver and support
Who pays in year two and after

Red flags

Recurring costs funded by one-time money
No repair, replacement, or help-desk budget
A critical partner or property owner not yet on board

Technology fit

Is the technology appropriate, interoperable, sustainable, and measurable in this setting?

The same technology can succeed in one place and fail in another. The dissertation's example: hotspot lending may work in a city with reliable cellular coverage but not in a rural community with spotty signal. Fiber may be the best long-term choice but costly in sparse areas, so fixed wireless, satellite, public Wi-Fi, community networks, and hybrids all have a place. Interview participants added usability to this check: a person with a connection and a device can still be excluded by an interface that is too complex.

Evidence to bring

Site conditions: coverage, terrain, power, building constraints
How the option will be maintained and measured
Usability and accessibility testing with real users

Red flags

A design that works in good weather but not in storms
Devices too limited for the learning they are meant to support
A platform tested only with experienced users

Safeguard review

Have risks been mitigated, and will benefits and burdens be shared equitably?

The safeguard review is where the gate acts as an ethical checkpoint. The dissertation asks it to cover privacy, accessibility, trust, community engagement, benefits, burdens, and differential outcomes, and it notes that a subsidized plan affordable only until a grant expires would not pass the model's safeguard and sustainability review.

Evidence to bring

Data collected, who can see it, and for how long
Accessibility and language plans
Who might be worse off, and how that will be watched

Red flags

Eligibility rules that stigmatize participants
Automated decisions no one can explain
Service that ends abruptly when funding ends

Chapter 4

Three exits and how to choose one

The dissertation gives the gate three outcomes: an option is advanced, modified, or not implemented. The skill lies in reading the five checks together.

Advance

The option fits the cause and nothing is weak. It moves to Phase 7, which begins by confirming the gate decision and its conditions of approval.

Every partial check becomes a condition that design must satisfy.

Modify

The idea is worth keeping, but not as proposed. It returns to the pre-gate step to be reshaped: paired with missing supports, narrowed to where it fits, or rephased.

Then it comes back through the gate.

Stop

The option is not implemented as proposed. Usually the cause is poor causal fit, or a weakness that cannot be fixed without changing what the idea is.

Stopping an option is not abandoning the need.

The cross-fit reading

The dissertation's detailed model includes a cross-fit analysis step before dispositions. The principle is that checks interact. Strong opportunity fit cannot rescue weak causal fit, because a priority need met with the wrong response is still unmet. Strong causal fit cannot carry weak feasibility, because the right response that collapses in year two leaves people where they started, sometimes worse off.

For the worksheet in the next chapter, this book proposes five working rules. They are a practical reading of the model, not rules stated in the dissertation, and teams should feel free to adapt them.

Any check not yet known → not ready. Return for evidence before ruling on the option.
Causal fit weak → stop as proposed. Go back to the diagnosis and the options.
Any other check weak → modify. If fixing it would change what the idea fundamentally is, treat it as stop and redesign.
Causal fit partial → modify. Bundle or narrow the option until it meets the cause.
Causal fit strong and nothing weak → advance. Each partial check becomes a condition of approval.

Stopping well

A stopped option still leaves a need behind. A good stop decision names the cause that remains, records what the gate learned, and points to the redesign. Several of the cases that follow show an idea stopped in its first form and advanced in its second.

After the gate

The gate's decision travels forward. Phase 7 turns conditions of approval into design requirements, procurement terms, safeguards, and readiness thresholds. Phase 8 sets indicators that match the causal logic. Phases 9 and 10 check whether outputs became outcomes, and for whom. When they did not, the evidence goes back to diagnosis, and the next idea meets the gate better informed.

Chapter 5

The gate worksheet

This worksheet turns the five checks into a working session. Fill it in with your team, rate each check honestly, and the worksheet will suggest an exit using the rules from Chapter 4.

Your entries are kept in this browser only, so you can come back to them. Use "Load a sample" to see a completed worksheet from Case 1.

Part A: The idea

The promising idea, in one line The decision problem and priority gap Who is affected The validated root cause, and how you know Other options considered

Part B: The five checks

Rate each check as Strong, Partial, Weak, or Not yet known, and note the evidence behind the rating.

Part C: The gate's reading

Rate all five checks

The suggested exit appears here once every check has a rating.

Conditions of approval, or the changes required What would change this decision The team's decision

The team's decision can differ from the suggestion. If it does, record why in the conditions box.

Plain-text summary to copy into minutes or a proposal

Case 1

The homework gap on a long bus route

Constraint: coverage and uptake

Connectivity that rides the bus reaches students during the trip. At home, on the fringe of cellular coverage, a loaned hotspot may find no usable signal.

The setting

Picture a rural district, a composite drawn from the dissertation's hotspot use case, where many students ride long bus routes from farming communities and live on the edge of cellular coverage. Teachers have quietly stopped assigning online homework because they know not every student can finish it. Families' home situations vary: some have no service at all, some share a single phone, and some have broadband they cannot afford to keep.

The promising idea

Lend a mobile hotspot to every student who reports no home internet. It is fast to procure, easy to explain to the board, and there is precedent. The idea arrives at the gate with strong support from teachers.

What the archive shows

Harris County Public Library's ECF Hotspot Lending Program bought and lent Wi-Fi hotspots to patrons who needed internet away from library buildings, funded through the Emergency Connectivity Fund. DD-04141
The Coachella Valley School Bus Wi-Fi Initiative put Wi-Fi routers on school buses serving rural farming communities with long rides and limited home broadband, implemented before July 2019. Its record describes turning travel time into connected learning time. DD-04098

The dissertation's own hotspot use case sets the diagnostic bar. Lack of home internet is the obvious barrier, but some students are held back by cost, some by service too slow for video, some by phones unsuited to writing and research, and some by coverage gaps, building interference, or data caps. It adds a telling scenario: a district distributes 300 hotspots and finds many unused. The model treats that as evidence to investigate, not a failure to ignore.

Running the gate

PartialCausal fit. For students in good coverage with no home service, a hotspot addresses the cause. For students on the fringe, the cause is the signal itself, and a hotspot cannot fix it. For students without a suitable computer, connectivity alone leaves the device barrier in place.
StrongOpportunity fit. The homework gap is documented and teachers are already changing instruction around it. Closing it would show up quickly in assignment completion and platform use.
PartialFeasibility fit. Devices are easy to buy. Data plans are a recurring cost, and the dissertation lists device loss, damage, battery limits, inventory, and sustainability funding as known downsides.
PartialTechnology fit. Strong where coverage is good, weak where it is not. The dissertation says hotspots in poorly covered rural areas may need to be replaced or supplemented with fixed wireless, satellite, or community anchor solutions.
PartialSafeguard review. Eligibility procedures can stigmatize; content filtering and privacy need settling; families need multilingual setup help.

The cross-fit reading shows something a single rating hides. Technology fit splits the student population in two, and causal fit splits along the same line. The option is right for part of the population and wrong for the rest.

Modify

Reshape the option by coverage and need. Test coverage at each student's address before checkout. Where the test passes, lend a hotspot with a district laptop if needed, multilingual setup guides, and after-hours help. Where it fails, pursue bus Wi-Fi on the longest routes and a fixed wireless or community anchor option for the fringe addresses.

Measure the right thing. Track active use and assignment completion, not devices issued, and investigate idle hotspots.

What would change the decision

If household-level coverage data showed usable signal at nearly every address, technology fit and causal fit would both rise, and the hotspot program could advance with conditions on support and eligibility.

Case 2

Wiring an older apartment complex

Constraint: owner control and ongoing cost

A new riser reaches every unit, but two things sit outside the wiring: who controls access to the building, and what residents pay once service is live.

The setting

The dissertation's multiple dwelling unit use case describes a 100-apartment building from the 1970s. Broadband is "available" on paper, yet residents experience slow, unreliable, expensive service. The building has old copper or coaxial pathways, no structured cabling, and materials that interfere with signal. One provider may control access, and the owner controls risers, rooftops, utility rooms, conduit, budgets, and construction schedules. Residents, the main beneficiaries, are the least formally powerful people in the decision.

The promising idea

Upgrade the building with fiber and a managed in-building network, and provide residents with service at no cost for the funded period. This is already a bundled idea: it pairs infrastructure with affordability, which is why it arrives at the gate in better shape than a wiring-only proposal.

What the archive shows

California's CASF Broadband Public Housing Account approved up to $595,911 on August 13, 2026, for four affordable housing developments with 315 units, installing infrastructure of at least 100/20 Mbps with service to residents at no cost. DD-04866
The Detroit Last Mile Initiative commits $3.5 million of a $4.5 million Rocket Community Fund gift, with DigitalC and the Detroit Housing Commission, to reach up to 2,400 households and about 6,000 residents, building on a 2025 pilot. DD-05149
Maryland's Home Stretch for Public Housing Broadband Program funded last-mile and in-building construction in affordable housing through the ARPA Capital Projects Fund. DD-04151

The dissertation compares four design options for the building: fiber to every unit, fiber to the building with managed Wi-Fi, a neutral-host or open-access model shared by several providers, and a temporary wireless bridge during internal upgrades. It recommends ranking them on resident need, total cost of ownership, speed and reliability, accessibility, privacy, maintainability, and affordability, and it lists the adverse conditions to expect: aging electrical systems, asbestos or construction limits, lack of riser space, displacement fears, language barriers, privacy concerns, and distrust of management.

Running the gate

StrongCausal fit. Barrier mapping found two causes, outdated building infrastructure and unaffordable service. The option answers both.
StrongOpportunity fit. Residents rely on mobile-only access or public Wi-Fi for school, work, telehealth, and public services. The expected outcomes are concrete.
PartialFeasibility fit. Right-of-entry is not yet signed, riser space is unconfirmed, and the no-cost service has an end date. What happens after the funded period is the year-two question.
StrongTechnology fit. Options were compared against the building's conditions; the managed network is maintainable and measurable by unit.
PartialSafeguard review. A managed building network raises privacy questions, and residents' distrust of management could suppress sign-ups. Multilingual onboarding and support are not yet designed.

Advance with conditions

Conditions of approval: a signed right-of-entry agreement before construction; a written plan for affordable service after the no-cost period; a privacy notice for the managed network; multilingual onboarding and a support path for devices, passwords, and installation.

Measure outcomes, not units passed. The dissertation is explicit that a building can be wired while the divide persists. Track subscriptions, reliability, and support tickets by unit type and language group.

What would change the decision

If the owner declines to sign access terms, feasibility turns weak and the option must be modified, perhaps toward a neutral-host model or a different building. If residents are not subscribing after launch, the dissertation points to price, distrust, missing devices, confusing sign-up, or weak outreach as the causes to investigate.

Case 3

A school beyond the grid

Constraint: no power, no network

Where there is no grid and no telecom, connectivity has to bring its own power. The school becomes the site of both.

The setting

A primary school in a remote district has no grid electricity and no telecom service. Pupils have never used a computer at school. Teenagers who have left school have nowhere nearby to go online. A donor has offered help.

First idea: send laptops

The donor's first offer is a shipment of laptops. It is generous, concrete, and quick. It goes to the gate first.

WeakCausal fit. The root causes are missing electricity and missing connectivity. Laptops address neither; without power they cannot even be charged reliably.
StrongOpportunity fit. The need is severe and the population is clearly identified.
PartialFeasibility fit. Delivery is easy; use is not.
WeakTechnology fit. Devices without power or network do not match the setting.
PartialSafeguard review. No obvious harms, but little benefit either.

Stop as proposed

Causal fit is weak, so the option stops. The need does not. The gate sends the team back to the options with a clear finding: power and connectivity have to come first, or at least together.

Second idea: power, connection, and a place to learn

What the archive shows

Cameroon's Mokolo Primary School Connect My School e-container, inaugurated in 2024 by UNICEF Cameroon, the Ministry of Basic Education, and IHS Cameroon, combines connectivity, solar power, and digital learning equipment, and serves pupils, teachers, and out-of-school adolescents. DD-00120
Bolivia's WiFi Comunitario via Túpac Katari, run by the Agencia Boliviana Espacial, creates shared Wi-Fi zones in communities with no telecom service, using satellite backhaul and solar power where electricity is absent. DD-00579
Australia's AARNet project, funded under the Telecommunications Disaster Resilience Innovation Program, uses wind, solar, and battery systems to keep regional connectivity running. DD-00869

The dissertation's text-network analysis surfaced reliable power as a basic infrastructural condition that an access-centered analysis can miss, and it flagged the cyclical relationship between electricity and broadband in rural materials. The redesigned option follows that finding: a solar-powered learning unit with satellite connectivity, devices, and after-hours community access.

StrongCausal fit. Addresses power, connectivity, and devices together.
StrongOpportunity fit. Serves pupils, teachers, and young people outside school from one site.
PartialFeasibility fit. Needs a trained local caretaker, a battery replacement budget, and an agreement on who pays for satellite service after the donor period.
StrongTechnology fit. Solar and satellite match the setting; the unit can be monitored remotely.
PartialSafeguard review. Community-access hours, supervision, and content filtering need rules that the community helps set.

Advance with conditions

Conditions of approval: a named local caretaker with training; a maintenance and battery budget; a service-continuity agreement; access rules agreed with the community.

The same need, the same donor, a different option. That is the gate's most common success.

Case 4

Tablets for older adults

Constraint: skills and confidence

Ownership and confidence are different barriers. The dissertation's heat map puts digital literacy at the top for older adults.

The setting

A city wants older residents to be able to book appointments, join telehealth visits, and reach public services online. Senior centers report that many visitors do not own a suitable device. Staff also report that visitors who do own one often avoid using it.

The promising idea

Buy tablets and distribute them to older residents through parks and senior programs.

What the archive shows

D.C. Senior Tablet Distribution, run by the Department of Parks and Recreation, funded tablet procurement and distribution for older residents at $340,500 for FY2024. DD-03967
D.C.'s Department of Aging and Community Living, with Tech Together D.C., paired iPad distribution and 100 Chromebooks with classes at six senior wellness centers in Wards 1, 4, 5, 6, 7, and 8. DD-03992
Luxembourg's Seniors-for-Seniors Intergenerational Digital Mediation, planned for 2026 to 2030, builds peer digital support into the model. Luxembourg also runs introductory AI workshops for older people. DD-02335, DD-02339

Both D.C. records come from the same state digital equity planning source, so the tablet program may well sit alongside the training program in practice. The case uses them to show two designs, not to judge either program.

The dissertation's Impact Heat Map rates digital literacy as the most severe barrier for older adults. Its rural use case adds that when older adults are excluded from telehealth, training and support may be as important as broadband service. Its trust category adds fear of scams and data misuse.

Running the gate

PartialCausal fit. Device ownership is a real cause for some residents. For many others, the dominant cause is skills and confidence, which a tablet does not address.
StrongOpportunity fit. Telehealth and online public services make the need urgent.
StrongFeasibility fit. Senior centers, procurement channels, and partners exist.
PartialTechnology fit. Tablets are a sensible form factor, but home connectivity has not been checked; a tablet without affordable service at home works only at the center.
PartialSafeguard review. Scam exposure and privacy are live concerns for new users; accessibility settings need attention.

Modify

Pair every device with a way to learn. Hold classes at the centers, recruit peer mentors, add scam-awareness and privacy basics, and check each recipient's home service, offering low-cost plan enrollment where needed.

Measure confidence and independent use, not tablets issued: the dissertation recommends skills, confidence, independent task completion, and support needs as indicators when literacy is the cause.

What would change the decision

A short intake survey could settle causal fit. If most intended recipients are confident users who simply lack a device, the original idea could advance with conditions on home connectivity and safeguards.

Case 5

When the emergency money ends

Constraint: time-limited funding

One-time money can connect families quickly. Unless something durable replaces it, the connection ends when the money does.

The setting

A composite district expanded device and hotspot lending during the pandemic using federal emergency funds. The program worked: students got online and teachers rebuilt assignments around it. The emergency money is now gone, and the district is covering data plans out of a shrinking contingency line.

The promising idea

Keep the program alive by applying for a new one-time grant each year, whichever program is open.

What the archive shows

The federal Emergency Connectivity Fund, established in 2021 with $7.2 billion, funded devices and off-campus broadband for students, schools, and libraries during the pandemic. Its archive record is marked historical. DD-03744
The D.C. BEAD Low-Cost Broadband Service Option requires deployment subgrantees to offer eligible households service at $30 or less per month, including taxes and fees, at 100/20 Mbps where feasible, with latency at or below 100 ms and no data caps. DD-04002

The dissertation is direct about this pattern. Short-term subsidy or emergency programs can connect families only temporarily. Grants may buy devices and launch programs without funding maintenance, staffing, subscriptions, evaluation, cybersecurity, training, or replacement. Interview participants described what a realistic yearly package costs:

"So I think our recommendation was about $1,000 to $1,200, which would include internet access for the year, a device, and then, of course, labor costs, right, for the digital literacy training that gets provided, and then the ongoing support for that."Interview participant QR2

Running the gate

StrongCausal fit. The cause, students without affordable home service, is real and still present.
StrongOpportunity fit. Instruction now depends on home connectivity.
WeakFeasibility fit. Recurring costs rest on one-time money that may not appear. There is no replacement budget, and each application year consumes staff time.
PartialTechnology fit. The devices are aging; the hotspots carry the coverage limits described in Case 1.
PartialSafeguard review. An abrupt lapse would cut families off mid-year, the burden-shifting the dissertation warns about.

By the worksheet's rules, one weak check other than causal fit means modify. Here, though, the fix for feasibility, stable recurring funding, is exactly what the idea lacks by design. Fixing it changes what the idea is.

Stop as proposed, then redesign

Stop the grant-renewal model. Keep the need on the table and redesign around durable affordability: enroll families in low-cost plans where providers offer them, pursue standing requirements like D.C.'s through state broadband programs, fold device replacement into the regular budget, and partner with libraries for lending that can outlast any single grant.

Protect families during the transition with a published end date for loaned service and help moving to a lasting plan.

What would change the decision

A multi-year funding commitment from a stable source would move feasibility to partial, and the program could return to the gate as a modify decision rather than a stop.

Case 6

An AI tutor for every student

Constraint: safeguards and uneven readiness

The same tool meets students with very different access. Without safeguards and access checks, an AI tutor can widen the gap it was meant to close.

The setting

A district's leadership has seen demonstrations of AI tutoring and wants every student to have one. A vendor offers a platform on a per-student subscription. The district's own access survey shows a mix of students with laptops and home broadband, students with only a phone, and students with no home service.

The promising idea

License an AI tutoring platform for all students, available at school and at home.

What the archive shows

Andorra's Ministry of Education is integrating digital competencies, computational thinking, media literacy, and AI literacy into the basic education curriculum, from 2025 onward. DD-01302
Intel's AI training programs in Costa Rican schools build AI skills and creator capacity for students and educators as part of broader efforts to avoid an AI-driven divide. DD-00023

The dissertation treats AI as a new driver of inequality. AI tools could support tutoring, translation, accessibility, advising, and feedback. If access requires paid subscriptions, advanced devices, reliable broadband, data literacy, or institutional computing capacity, the divide could widen, and bias, opacity, privacy risks, and uneven AI literacy can create new disparities. It cites U.S. Department of Education guidance (2023) that AI in education should be human-centered and attend to equity, safety, effectiveness, transparency, trust, privacy, and local context.

Running the gate

PartialCausal fit. Limited access to tutoring is a real gap. For students without home service or a capable device, though, the platform does not reach the cause of their exclusion.
PartialOpportunity fit. Potential impact is high, but the benefit would flow first to students who are already best connected.
PartialFeasibility fit. Subscription costs recur; teachers have not been trained to use or supervise the tool.
PartialTechnology fit. Works well on laptops with good bandwidth; untested on phones and with assistive technology.
WeakSafeguard review. The data agreement is unsigned, the vendor has not explained how responses are generated or what student data is retained, and no accessibility review has been done.

Modify

Split the idea. Advance AI literacy now for students and teachers, following the curriculum-based approach in the archive. Hold the platform until a safeguard review passes: a signed data agreement, transparency about how the tool works, an accessibility test, and teacher training.

Then pilot where access is confirmed, compare outcomes across student groups, and pair any wider rollout with the connectivity and device work that reaches students on the far side of the gap.

What would change the decision

A completed safeguard review and a plan to close home access gaps for the students who lack them would lift the weak check and could bring the platform back to the gate as advance with conditions.

Closing

What the six cases teach

Different constraints, one discipline: judge the option against the cause, then read the checks together.

Case	Constraint	Causal	Opportunity	Feasibility	Technology	Safeguards	Exit

Most good ideas leave the gate changed. Five of the seven gate runs ended in modify or stop, and the stopped laptop shipment in Case 3 came back as a solar learning unit that advanced.
Causal fit is the hinge. Every stop in these cases traced to a weak causal fit or a weakness that could not be fixed without changing the idea.
Year two belongs at the gate. Feasibility and safeguards both turned on what happens after launch money ends.
Conditions carry the gate forward. An advance decision is only as good as the conditions attached to it, and the indicators that check them.

Back matter

Sources and records

Ross, P. (2026). Dissertation, committee draft dated September 23, 2026: Chapter 4 (substantive interpretation of the conceptual model, RQ1 barrier results, model simulations and use cases, interview findings) and the outline of Appendix AC (pre-gate transition, gate tests, cross-fit analysis, gate dispositions). The appendix text itself was not in the file provided; the book uses its section headings only.

Digital Divide Projects and Initiatives master file, updated September 24, 2026, digests 1 to 10 (6,295 records), All Projects sheet.

Record	Project	Place	Used in
DD-04141	Harris County Public Library ECF Hotspot Lending Program	Texas, US	Case 1
DD-04098	Coachella Valley School Bus Wi-Fi Initiative	California, US	Case 1
DD-04866	CASF Broadband Public Housing Account portfolio	California, US	Case 2
DD-05149	Detroit Last Mile Initiative	Michigan, US	Case 2
DD-04151	Maryland Home Stretch for Public Housing Broadband	Maryland, US	Case 2
DD-00120	Mokolo Primary School Connect My School e-container	Cameroon	Case 3
DD-00579	WiFi Comunitario via Túpac Katari	Bolivia	Case 3
DD-00869	AARNet Wind, Battery and Solar Telecommunications Resilience Project	Australia	Case 3
DD-03967	D.C. Senior Tablet Distribution	District of Columbia, US	Case 4
DD-03992	DACL Senior Wellness Center Training and Device Distribution	District of Columbia, US	Case 4
DD-02335	Seniors-for-Seniors Intergenerational Digital Mediation	Luxembourg	Case 4
DD-02339	Introductory AI workshops for older people	Luxembourg	Case 4
DD-03744	Emergency Connectivity Fund	United States	Case 5
DD-04002	D.C. BEAD Low-Cost Broadband Service Option	District of Columbia, US	Case 5
DD-01302	Digital and AI Literacy Curriculum Integration	Andorra	Case 6
DD-00023	Intel AI Training Programs in Schools	Costa Rica	Case 6

Settings described as composites are illustrative scenarios built from the dissertation's use cases. Gate ratings are this book's applications of the model, not evaluations of the named projects.