Field report
The Portal Assumes a Desk
The form fits on a phone. The documents, passwords, deadlines, and rest of a person’s day may not.
“Apply online” is often presented as the absence of a barrier. No trip. No paper packet. No line at a counter. The portal is open all day, and the form shrinks politely to fit the screen in a person’s hand.
The task may still assume a desk.
It may assume a second device holding an email code, a stable connection, a charger, a current password, a quiet place to read, files already scanned in the right format, and enough uninterrupted time to recover when the session expires. None of those items appears in the button label. They appear when a person tries to finish.
The point is not that digital service is a mistake. It is that channel availability and completed access are different measures.
Sixteen percent is a group, not a diagnosis
In surveys conducted in 2025, Pew Research Center found that 16 percent of U.S. adults were “smartphone-only.” Pew surveyed 5,022 adults between February 5 and June 18, 2025 and published the result in November of that year.
The figure is not a count of people who lack the internet. It is not a portal-failure rate. It is a national survey estimate describing a particular access arrangement.
Pew defined smartphone-only adults as people who owned a smartphone but did not subscribe to home broadband. That definition does not reveal whether someone also used a work computer, library connection, tablet, public Wi-Fi, or another person’s service. It does not measure data caps, connection quality, computer ownership, device age, accessibility tools, or skill with a particular task.
Keeping the definition intact is important because loose language can turn “phone without home broadband” into “offline,” “poor,” or “unable.” The source does not support those substitutions. It supports a narrower observation: a substantial share of adults may depend on a phone and mobile service as their primary personal route to the internet.
Survey method also places a seam in the trend. Pew changed its data-collection mode in 2023. Comparisons that cross that change require caution because a shift in how people are reached and answer can affect the estimate. The 2025 figure is useful as a current bounded snapshot. It should not be turned into a smooth historical story without accounting for the break.
That is not bureaucratic fussiness. A good access argument cannot demand careful service design while being careless with its own denominator.
A phone can work beautifully
The Pew record does not show that smartphone-only adults fail online forms more often. The practical concerns—screen size, connection stability, switching between applications, document handling, data limits, and session persistence—are plausible mechanisms, not measured outcomes in that survey.
Some phone-based tasks are easier than their desktop versions. A camera can capture a document. A text can deliver a reminder. Autofill can prevent typing errors. Biometric sign-in may be more usable than a remembered password. A responsive form with saved progress can turn a bus ride or break into time that works for the applicant.
That countercase matters. “Mobile” is not the defect. The defect appears when a service claims mobile access because the page fits while the full task still depends on equipment, time, or handoffs the design never checked.
The federal government already names the hidden work
Office of Management and Budget guidance for federal public-benefit programs tells agencies to examine the beginning-to-end burden of applying for and maintaining eligibility. Its examples include finding and gathering records, traveling, waiting for agency personnel, learning program rules and procedures, and psychological costs.
That list is useful because it moves attention away from the number of questions on one screen. A short form can sit inside a long task. The applicant may first need to understand which program applies, locate a document, ask an employer for a record, wait for a code, travel to a printer, call for help, and later repeat part of the process to keep the benefit.
The burden is not merely how long fingers touch the form. It is the work needed to reach a usable decision.
OMB’s guidance also preserves a necessary counterweight. Eligibility systems may need identity, income, household, residency, disability, or other information to administer law, prevent errors, protect privacy, and direct limited public funds. Not every document request or identity check is disposable friction.
The design question is whether the agency has chosen the least burdensome reliable way to obtain what it genuinely needs. Can an existing government record be reused with appropriate consent and safeguards? Can a person submit a readable phone photo? Does the portal explain why a document was rejected? Is there a workable route for someone whose record does not fit the common case?
A service can be strict and still be humane. It has to explain the rule, preserve progress, and provide a route through exceptions that does not require a private detective.
Hours are already part of the official account
Federal information-collection practice formally measures paperwork burden in hours. OIRA’s live inventory of approved collections aggregates estimated annual burden across individuals, businesses, governments, and other respondents. The total changes as collections are added, revised, and retired, and it is not a household-only measure.
The important point here is not one enormous national total. It is that time is already recognized as a cost of compliance. When a portal moves printing, scanning, correction, and follow-up to the applicant, those minutes do not become free because the agency no longer sees a line in its lobby.
The same caution applies in reverse. A digital process can remove travel, mailing, manual data entry, and uncertainty. The field record has to observe what changed rather than assume that every portal either liberated or punished its users.
Test the task, not the home page
A useful mobile access test follows one realistic path from discovery to decision:
- Begin on a phone without a prepared folder of documents.
- Find the eligibility rule and identify every record the task requires.
- Create or recover the account using the channels the service actually offers.
- Leave the task once, switch applications, lose connectivity briefly, and return.
- Upload a document using the phone’s ordinary camera and file controls.
- Trigger one common correction, such as a rejected image or mismatched name.
- Locate human help, note its hours, and preserve any confirmation or case number.
- Check whether the user can tell what happens next and when.
The test should record steps, waits, failures, recovery, and the equipment used. It should also include assistive technology and people with different reading levels. A fast completion by the person who designed the form is not evidence that the form is easy.
Nor is one failed attempt proof that the entire service excludes a population. The review needs several readers, a stated task, and a record of where difficulty occurred. Did the wording confuse the person, did the connection fail, did the document rule reject a valid file, or did the program require information the reader did not have? Those are different defects with different repairs. “User error” is not a diagnosis. “The portal is broken” is not one either.
The defensible conclusion is narrower than “portals exclude people” and more useful than “the site is responsive.” Sixteen percent of adults in Pew’s 2025 surveys had a smartphone without a home broadband subscription. Federal guidance recognizes that access to benefits can require documents, learning, waiting, travel, and psychological work. Together, those records justify testing the whole service under realistic device and life conditions.
They do not identify the failure rate of any portal. They do not prove a phone is inadequate. They reveal why a completed page layout is not a completed access analysis.
Evidence boundary
This report uses Pew’s 2025 Mobile Fact Sheet, OMB Memorandum M-22-10, and the federal information-collection framework already reviewed for bent.WTF. The 16-percent estimate, survey period and size, smartphone-only definition, 2023 mode break, named burden categories, and official use of burden hours are sourced. The desk metaphor and end-to-end mobile test are bounded editorial interpretations.
The record does not measure portal completion, abandonment, error, processing, or benefit receipt by device type. It does not establish that smartphone-only adults lack computers or cannot finish online tasks. A public claim about a specific service needs a current, observed path with accessibility and reader evidence attached.
Continue through the record
Keep the source record beside the claim.
The Library keeps definitions, exhibits, caveats, and related work close enough to check.
Start with an active file. Open the exhibits when a claim does real work.
Open the Library