The Hammer and the Nails: What POSIWID Misses

The Limitations of Systems Analysis

Introduction: The Appeal of Clear Answers

There's something deeply satisfying about the POSIWID framework—"the purpose of a system is what it does." It cuts through mission statements, good intentions, and complexity to offer stark clarity: forget what they say they're doing; look at what actually happens.

Applied to long COVID, it reveals patterns: systems that mobilized $18 billion for vaccines in ten months but allocated $1.15 billion for treatment research over six years. Systems that identified biomarkers approaching 80% accuracy but haven't deployed clinical tests. Systems where diagnosis clusters among those with healthcare access despite higher symptom burdens elsewhere. The framework says: these outcomes reveal purpose, regardless of what anyone claims.

But POSIWID is a tool, and like any tool, it has limitations. Use only a hammer and everything looks like a nail. Apply only systems analysis and every outcome looks like purposeful design. This essay examines what POSIWID illuminates and what it obscures—not to abandon the framework, but to understand its proper scope.

What POSIWID Does Well

Before cataloging limitations, it's worth acknowledging where the framework has real analytical power.

It reveals divergence between stated and actual priorities. When resource allocation contradicts stated commitments—vaccines receiving $18 billion and emergency authorities while long COVID research gets $1.15 billion and standard timelines—the gap itself is data. POSIWID forces us to take that seriously.

It identifies patterns across variation. When Sweden's openness strategy and New Zealand's elimination strategy converge on similar 2025 endpoints (endemic management, withdrawn prevention infrastructure), the convergence suggests something beyond individual policy choices—structural constraints or gravitational pulls toward certain equilibria.

It cuts through performative complexity. Organizations are skilled at generating activity that looks like progress—studies, committees, initiatives. POSIWID asks the uncomfortable question: does all this activity produce outcomes, or does it produce the appearance of response?

It highlights systematic disparities. When outcomes consistently favor certain groups (formal diagnosis clustering in well-resourced populations), individual explanations miss the pattern. POSIWID reveals systemic dynamics.

These are real contributions. But the framework operates at a level of abstraction that sands down grain. Four kinds of grain matter: agency, causality, constraint, and context.

Where POSIWID Cuts Away: Four Meta-Limitations

Agency Gets Flattened

POSIWID analyzes systems as coherent actors with unified purposes. Outcomes don't record the researchers building assays, the clinicians running long COVID clinics on shoestring budgets, the disability advocates fighting for formal recognition, the patients turning symptom diaries into datasets. Aggregates don't intend; people do. Yet people labor inside scaffolds not of their making.

Saying "the system maintains diagnostic ambiguity" invisibilizes those struggling against it. A researcher who devoted years to identifying IL-6 and PTX-3 as biomarkers might read "the system sustains research without translation" as dismissive of their scientific contribution. A clinician managing 200 patients without approved treatments might hear "the system provides only supportive care" as criticism of their work rather than recognition of their impossible position.

This isn't just about feelings—it's an analytical gap. By focusing on aggregate outcomes, POSIWID can miss the internal conflicts, resource battles, and advocacy efforts that reveal systems aren't monolithic. Understanding why translation doesn't happen requires talking to people inside the translation pipeline. POSIWID's outcomes-only focus discourages this.

The framework shows the scaffold's silhouette, not the fight within it.

Causality Is Messy; POSIWID Is Tidy

POSIWID treats outcomes as if they reveal purpose, but many system outcomes are emergent properties of interacting constraints, not anyone's goal.

"The system rations recognition by socioeconomic status" sounds intentional. Reality: insurance coverage gates specialist access; access gates who gets referred to post-COVID clinics; referral gates diagnosis; diagnosis produces stratified recognition. No one designed this as a rationing mechanism. It emerged from insurance-based healthcare interacting with specialist-dependent diagnosis.

It is tempting to call an emergent regularity a "purpose." That looks like rationing by design. It is rationing by structure—the difference matters for where leverage actually is. Hearts pump blood, but their "purpose" is retrospective interpretation, not a design goal. Systems don't necessarily "intend" their outcomes—they produce them through complex interactions of incentives, constraints, path dependencies, and feedback loops.

Calling these emergent patterns "purposes" can make them seem more coordinated and controllable than they are. This paradoxically might make change seem easier ("just change the purpose!") when structural dynamics are harder to shift than intentions.

The economist's concept of "revealed preference" is more precise here: choices under constraint reveal what's possible and what's prioritized when tradeoffs must be made. But even this is cleaner than reality. Systems don't "prefer" in any psychological sense—they allow certain outcomes through incentive structures and infrastructural configurations. Precision about this matters.

Constraint and Path Dependency Masquerade as Choice

Systems often produce outcomes because they're locked into historical paths, not because those outcomes serve current purposes.

Why no long COVID diagnostic test in routine clinical use? POSIWID says: strategic ambiguity serves administrative convenience. But consider the infrastructure: diagnostic test development pathways designed for acute conditions with clear biomarkers. Regulatory frameworks assuming single-marker clarity (like glucose for diabetes). Reimbursement codes that don't accommodate novel post-infectious syndromes spanning multiple systems. Electronic health record systems lacking fields for symptom tracking over months. Lab infrastructure optimized for high-volume standardized tests, not research-grade multi-marker panels.

The system isn't choosing ambiguity in any meaningful sense—it's moving through it as fast as its plumbing allows. Laboratories, reimbursement codes, EHR schemas, trial endpoints—these are infrastructures tuned for acute, single-marker disease. Translating multi-marker, multi-system, post-infectious syndromes through those pipes is slow even with maximal will.

POSIWID can make everything look like choice when much is institutional inertia and path dependency. Systems often can't easily change even when actors within them desperately want to. Treating this as "revealed purpose" misdiagnoses the problem: the issue isn't that the system wants the wrong thing, but that it's structurally constrained from doing the right thing.

That said: time itself can be an instrument of rationing. Even when lag is genuine, delay benefits some parties (avoiding obligations, deferring costs) and harms others (prolonged suffering, accumulating disability). Acknowledging pipeline realities doesn't preclude asking who benefits from slow pipelines.

Context Decides What's Possible

POSIWID treats "the system" as if it operates in a vacuum, but system behavior is shaped by cultural values, political constraints, and public demand.

Finland's smooth normalization reflects high institutional trust (people generally believe government health guidance), universal healthcare (no one loses coverage), small population (5.5 million allows coherent national policy), and social cohesion. The U.S.'s contested normalization reflects low institutional trust especially polarized by politics, fragmented healthcare, vast scale (330 million people across diverse contexts), and cultural individualism.

Same "purpose" revealed by POSIWID (endemic management with withdrawn infrastructure) can minimize the radically different contexts producing that outcome through different mechanisms. High-trust, universal-coverage, small-population systems and polarized, fragmented, continental-scale systems can arrive at similarly normalized endpoints by utterly different routes. Calling both "purpose" risks mistaking convergence for coordination.

More fundamentally: "what systems do" is partially what publics allow or demand. By 2025, democratic publics broadly signal exhaustion with restrictions and desire for normalcy. Systems respond to this. Maybe the revealed purpose isn't "systems don't care about chronic disability" but "democracies struggle to maintain collective sacrifice for diffuse, long-term threats when immediate crisis passes."

This doesn't mean such sustained efforts are impossible. Democracies have maintained long-horizon public health commitments: lead abatement infrastructure over decades, smoking cessation programs, auto safety regulations that continuously ratchet standards upward. These aren't perfect analogues to pandemic response, but their existence keeps "impossible under democracy" from sounding like a law of nature. The question becomes: under what conditions can democracies sustain collective effort, and why don't those conditions obtain for long COVID?

On Epistemic Humility: Pace, Intent, and Honest Uncertainty

Three limitations cluster around a common theme: POSIWID struggles to distinguish between different kinds of slowness and ambiguity.

Temporal myopia: POSIWID evaluates systems at single time points, but change takes time. Biomarkers identified in 2023 aren't clinical tools in 2025—but a realistic pipeline runs: discovery (2023) → validation studies (2024-2025) → clinical trial endpoints (2026-2027) → regulatory approval (2028) → reimbursement codes (2029) → provider adoption (2030+). We might be in the normal middle, not evidence of avoidance.

Scientific difficulty: Is long COVID one condition or several? Should treatments target viral persistence, immune dysregulation, autonomic dysfunction, or all three? These are genuinely unsolved questions. Post-viral syndromes have been studied for decades in ME/CFS and post-Lyme without definitive mechanisms or treatments. The science really is hard.

Legitimate uncertainty: If saying "we don't know yet" gets interpreted through POSIWID as "strategic non-knowing," scientists and policymakers might feel pressured to claim certainty prematurely. This would be worse than ambiguity—false clarity leading to ineffective interventions.

The synthesis: some slowness is stalling, some is science; some ambiguity is strategy, some is the honest state of knowledge. POSIWID can't always distinguish. This doesn't absolve systems from accountability—it prevents analytic hubris. The work is determining which is which, not assuming all delay is deliberate.

On Selection Bias and Constructive Engagement

Two related limitations reveal how POSIWID gets deployed:

It's typically applied to suspected dysfunction. We apply POSIWID to long COVID response but not to childhood vaccination programs (where stated purpose and actual outcomes align reasonably well), or rapid genomic sequencing that tracks variants in real-time, or the dramatic decline in smoking rates following decades of sustained public health effort. This selective application makes the framework seem universally damning when it could also confirm alignment.

It privileges critique over construction. POSIWID is a diagnostic tool, not a design tool. "The system individualizes collective harm"—okay, but what's the alternative? Collective prevention requires sustained public buy-in for ongoing precautions, economic trade-offs, political will to maintain disruption, enforcement mechanisms that don't trigger backlash. In democracies, policies require consent. Pure POSIWID analysis can become nihilistic if it only reveals dysfunction without engaging the difficulty of alternatives.

Every option has tradeoffs. Critique without engaging those tradeoffs can read as detached from constraints real decision-makers face. Saying "the system should do X" requires showing how X is achievable given resource limits, political constraints, institutional structures, and human behavior.

When POSIWID Works Best

Despite these limitations, the framework has value in specific contexts:

• When stated purposes and outcomes diverge dramatically (vaccines in 10 months vs. treatments in 6+ years)
• When resources are allocated inconsistently with stated commitments ($18B vs. $1.15B quantifies the gap)
• When patterns persist across different contexts (Sweden/New Zealand convergence)
• When outcomes systematically benefit some groups over others (diagnosis by healthcare access, not symptom burden)
• When systems avoid measurement that would create accountability (not tracking long-term outcomes of endemic management)

These patterns warrant POSIWID analysis because alternatives don't explain them well. The tool doesn't change, but showing that it can confirm alignment as well as divergence inoculates against charges of universal pessimism.

What to Pair With POSIWID

To compensate for limitations, systems analysis needs supplementation:

Institutional ethnography: Talk to people inside systems. What constraints do they face? What did they try that failed? What resources do they lack?

Historical analysis: Trace path dependencies. Why were things set up this way? What would changing course require?

Comparative policy analysis: How do different systems handle similar problems? This provides the counterfactuals POSIWID lacks.

Stakeholder perspectives: What do patients experience? What do providers observe? What do researchers need?

Explicit engagement with tradeoffs: If the system shouldn't do X, what should it do instead? What would that cost? What would be gained and lost?

Conclusion: Holding Both

POSIWID earns its keep by refusing to be hypnotized by mission statements. It forces attention to budgets and behaviors, to divergences between rhetoric and resource allocation, to the durable advantages that survive every declaration of solidarity.

But a system is not a mind; it cannot "want," it can only allow. What it allows is set by incentives, infrastructure, and publics. If we stop at "purpose," we risk moralizing constraint and mistaking pace for intent. If we ignore "what it does," we collude with performance.

The work is to hold both: trace the budgets and the behaviors, and then ask which parts are the will, which parts are the wiring, and which parts are the humans. POSIWID identifies patterns worth investigating—but understanding why systems produce those patterns, whether they could do otherwise, and what they should do instead requires going beyond observation of outcomes to engagement with causes, constraints, and possibilities.

Use it to spot the divergences that warrant explanation. Then investigate with tools that capture individual agency, historical constraints, cultural context, scientific uncertainty, and genuine tradeoffs. The purpose of a system may be what it does—but that's the beginning of analysis, not its conclusion.