The Digital Pillar Rests on Evidence It Doesn't Have

The National Strategy to Advance Social Connection, released alongside the Surgeon General’s advisory in May 2023, organizes its response into six pillars. Five of them concern institutions that have existed for decades: schools, health systems, workplaces, neighborhoods, government programs. The fourth pillar is different. It calls for reforming digital environments — data transparency, safety standards, and what the strategy calls pro-connection technology design. The Department of Health and Human Services’ hub for the strategy lists this pillar alongside the others as though the evidence supporting it were of comparable weight.

It is not. This is worth stating plainly, because the policy conversation around technology and loneliness has a habit of borrowing the authority of the mortality literature — the well-established, decades-old body of work on social isolation and cardiovascular and all-cause mortality — and applying it to a much newer and much less settled question: what specific role do digital platforms play in loneliness, and what would change if they were redesigned.

What the advisory actually says about technology

The Surgeon General’s advisory itself is careful on this point, more careful than much of the commentary it has generated. It does not claim that social media causes loneliness. It notes that technology has changed how people connect and disconnect, and that both effects run in both directions — some digital contact substitutes for in-person contact, some facilitates it. The advisory’s own framing acknowledges that the directionality and magnitude of these effects are not settled.

That caveat tends to disappear by the time the advisory’s headline figures — roughly half of U.S. adults experiencing loneliness, a mortality risk compared to smoking up to fifteen cigarettes a day — get paired with the fourth pillar’s call to reform digital environments. The pairing implies a causal chain: technology use produces loneliness, loneliness produces the mortality risk, therefore changing technology reduces the risk. Each link in that chain is plausible. None of it has been demonstrated as a chain.

Where the actual evidence sits

Consider what the survey evidence can support. Gallup and Meta’s global survey of social connection, fielded from June 2022 to February 2023 across 142 countries, found 72% of respondents felt socially connected — a figure that sits uneasily against Gallup’s separate finding, from the same period, that 24% of people worldwide report feeling lonely. These are not simply inverse: connectedness and loneliness track different things, and a population can report high connectedness on average while a substantial minority reports loneliness. Neither figure isolates technology’s contribution, because neither survey was designed to. They ask about connection and loneliness as outcomes, not about the pathway that produced them.

The Harvard Graduate School of Education’s Making Caring Common survey, from February 2021, found 61% of young adults aged 18 to 25 reporting serious loneliness, with 43% saying their loneliness had increased since the pandemic began. This is frequently cited as evidence that heavy platform use among the young is driving the loneliness gap between generations. But the survey measured loneliness and measured age; it did not measure platform use as an explanatory variable, and its design cannot separate technology’s role from the pandemic disruption, economic precarity, or the loss of in-person institutions that the survey’s own authors point to as contributing factors.

The American Enterprise Institute’s 2021 survey on American friendship offers a similar structure. It documents that 12% of Americans report no close friends at all, up from 3% in 1990, and that the share of men with six or more close friends fell from 55% to 27% over the same period. Both are before-and-after comparisons across a thirty-one-year span during which nearly everything about daily life changed — work patterns, marriage rates, geographic mobility, and, yes, the arrival of smartphones and social platforms partway through. A before-and-after comparison across three decades cannot assign the decline to any one of those changes. It can only establish that the decline occurred.

The specific gap: no intervention evidence

None of this means technology is irrelevant to loneliness. It means the literature has not yet produced what a reform-oriented policy pillar actually requires: evidence that changing a specific feature of a specific platform changes loneliness outcomes for a defined population, measured before and after, ideally against a comparison group that did not experience the change.

The BMC Public Health review of the state of loneliness and social isolation research, published in June 2023, is explicit that inconsistent measurement across studies is one of the field’s central barriers, and that this makes comparison across studies — let alone across platforms or design changes — difficult to do with any confidence. A Frontiers in Public Health editorial collecting pandemic-era work on isolation and mental health makes a related point: most of what exists documents association between digital behavior and psychological outcomes, not the effect of a deliberate change in digital environment.

This is the same gap that shows up, in a different form, in the American Heart Association’s 2022 scientific statement on social isolation and cardiovascular risk — that statement is explicit that the absence of intervention evidence is the field’s central limitation, even for isolation interventions delivered face to face, which have existed far longer than any proposed technology reform. If intervention evidence is scarce for programs with decades of practice behind them, it is close to nonexistent for the specific, novel proposition that redesigning a notification system, an algorithm, or a data-sharing practice measurably reduces loneliness at population scale.

Why the distinction matters for what comes next

A policy pillar built on association rather than demonstrated effect is not necessarily wrong to pursue. Data transparency and safety standards for platforms may be justified on other grounds entirely — consumer protection, child safety, market competition — independent of any claim about loneliness. The risk is in how the claim gets used. If “reform digital environments” is justified primarily by appeal to the mortality statistics and the youth loneliness figures, and those figures were generated by instruments that never measured technology’s causal contribution, then the pillar is borrowing certainty it has not earned. That borrowed certainty makes it harder, not easier, to design the kind of study that would actually settle the question — because the policy conversation proceeds as though the question is already settled.

The Foundation for Social Connection’s resource library, which tracks the federal response to the advisory, is useful here mainly as a signpost: it shows how much of the current policy architecture around technology and connection is built on the advisory’s framing rather than on a body of platform-specific intervention research that does not yet exist.

What would close the gap is not more prevalence surveys. The field has those in abundance, and they increasingly disagree with one another on particulars while agreeing on the general shape of the problem. What is missing is a study that randomizes or otherwise varies a specific platform feature — a notification cadence, a default privacy setting, an algorithmic ranking choice — across comparable groups, and follows loneliness and social network measures over time using a consistent instrument. Until that kind of study exists, the fourth pillar of the National Strategy is a reasonable hypothesis dressed in the language of an established finding. Policy can proceed on a hypothesis. It should not describe it as more than that.