Free and Easy Tax Filing

The Future of Tax Filing

05. 12. 2026 • Gabriel Zucker and Chris Given, Future of Tax Filing Fellows

Lessons from Direct File and a roadmap for building a simpler, more accessible public tax filing system.

Download the Full Report

Table of Contents
Chapter 1: The story of Direct File and tax access reform in the Biden Administration
Chapter 2: Direct File and data import / pre-population
Chapter 3: Direct File and tax scope
Chapter 4: Direct File and state taxes
Chapter 5: Direct File and identity verification
Chapter 6: Direct File usage and cost levels
Chapter 7: Direct File promotion and outreach
Chapter 8: Integration between Direct File and other IRS systems
Chapter 9: Direct File, fact graph, and a new way of building civic technology products
Chapter 10: IRS communications
Chapter 11: Real-time taxes
Chapter 12: Dependents
Chapter 13: Additional tax filing assistance
Chapter 14: Non-filers
Chapter 15: IRS and private providers

This webpage includes an overview of the full report, brief summaries of all 15 chapters, and links to download each chapter individually. You can also download the complete report below.

Introduction

Introduction: How to use this report

This report represents a look back and a look forward from civic technologists who were involved in the tax reform movement of the Biden Administration, and in the creation of Direct File in particular. It documents what happened and what was learned in the run-up to and implementation of Direct File from 2021 to 2025, distills key policy conclusions, and lays out a program of future reform. It is an attempt at nearly-exhaustive documentation. As such, different components of the report will be helpful to different policymakers, technologists, and administrators attempting to advance the project of tax access reform in the future. The content is intended to be comprehensible for anyone with some grounding in government administration in general and the basics of tax administration in particular.

The report is organized in 15 chapters, each of which can be read alone or in conjunction with the rest of the report. Where appropriate, readers are directed to other sections or chapters to help find more detailed information on related topics. The report’s 15 chapters are presented in three parts: 1) the Story of Direct File; 2) Building Direct File: Policy and Strategy; and 3) Beyond Direct File: a Vision for the Future of Tax System Access.

On this web page, you’ll find introductory information about the entire report, and brief summaries of each of the 15 chapters. Below each chapter summary you will find a link to a PDF of the full chapter. Alternatively, you can download the full report, including introductory information and chapter summaries, here.

Part 1: The Story of Direct File (Chapter 1)

Chapter 1 is a first draft of history, telling the story, from the perspectives of the authors, of how Direct File went from relatively-little-discussed advocate pipe dream to reality during the Biden Administration. We include portions of this story that were not publicly shared at the time and highlight a handful of themes that run through the narrative. We believe having this thumbnail of what happened may help future reformers move faster and more effectively in reviving Direct File, or pursuing other similar initiatives.

Part 2: Building Direct File: Policy and Strategy (Chapters 2-9)

Launching an entirely new tax filing product is a hard and multi-faceted project, and over our years of working on it, we built a great deal of institutional knowledge around the various policy and strategy issues at play. Chapters 2-9 provide an overview of some of those issues — how decisions were made to date, where things may have been on track to go in 2026 or 2027 had Direct File continued, and how future decision makers should consider approaching these issues moving forward.

What’s in these chapters: a framework and reference for future implementers 

These chapters reflect our best effort to be useful rather than merely opinionated: prescriptive where experience gave us clarity, and descriptive where it did not. Our goal is to ensure that any future teams reanimating something like Direct File do not have to start from square one. Where our experience left us with clear opinions about the right path forward, we lay out that path. Where the right next steps are not clear, or where several paths are viable, we lay out a framework of how decision makers may think about their next moves. Where policy and strategy choices became entangled in a web of implicit barriers and inherited policies, we describe that web as much as possible, including parts that generally weren’t previously publicly discussed. We also try to highlight issues that we feel were not well understood in the public discourse, and issues that, as implementers, we had an especially hard time getting our arms around, in the hopes that future teams can have a head start.

These chapters are intended to be useful overviews for any implementation teams picking up a Direct File project in the future, particularly policymakers and technologists at the Treasury, IRS, and state revenue departments. They should also be useful to those who intend to influence and guide the work of the tax reform space more broadly — funders, thought leaders, filing advocates, and organizers. 

These chapters are not overviews of how the code works or how the product team was organized; for that, readers should consult the Direct File open source code, and the appendices of the Filing Season 2025 Direct File report. Likewise, these chapters are restricted to the central policy and strategy questions of the core of Direct File. Direct File tangentially touches a wide range of IRS functionality, and indeed could in principle expand directly into these other areas in the future. Those other topics are discussed in Chapters 10-15.

The case for Direct File

This report is not primarily a defense of Direct File or a description of its past or potential future impact — all of which has been documented at length elsewhere. That said, we of course believe deeply in Direct File and our analysis throughout this report is rooted in the specific significant advantages it offers over the status quo tax filing system. At the most basic level, we believe that if the government requires people to file taxes, it has an obligation to provide a free, accessible, public way to do so — and that outsourcing that function entirely to commercial industry that charges for it is indefensible. We envision Direct File as a free, easy tax filing option for a majority of taxpayers — and especially for those who cannot afford commercial offerings, or who face particular obstacles accessing the syste. We envision it, furthermore, as the centerpiece of a growing constellation of reforms intended to ease access to the tax system, save taxpayers money, and close the gap in tax benefits access. Compared to existing tax prep offerings today, Direct File has four foundational key advantages, the importance of which come into play throughout the various issues discussed below:

1. Free. Direct File is unconditionally and always free. There are many other tax filing options that are advertised as free, sometimes free, partially free (for example, free federal returns paired with paid state returns), or free with caveats (for example, you do not pay a fee to use the service, but your data may be sold for a profit). Direct File is free without caveats.
2. Easy and automated. Direct File was and can be built around the assumption of using IRS data to streamline the filing process. In time, this means the Direct File experience will eradicate tax filing as we know it for many taxpayers, turning a stressful chore into a quick, painless process.
3. Public and official. Direct File is a publicly-owned product and an official offering of the government. This means taxpayers do not have to engage with third-party providers they may not trust. It means that taxpayers can have faith that the tax interpretations implemented by Direct File are fully endorsed by the government. It means that Direct File can pursue the goals of empowering taxpayers and making their returns transparent and comprehensible, rather than painting the tax system as an impossibly complex knot of arcana that only experts could understand, in the interest of upselling bewildered taxpayers. And it means Direct File can be straightforwardly promoted as the official tax filing option, breaking through the noise and choice overload of dozens of slightly different private tax filing options — making a particularly big difference in getting through to hard-to-reach populations, including new and intermittent filers.
4. Integrated with other IRS services across the tax system. As an IRS product, Direct File need not be an island. Rather, it can be positioned as a connector and a front door within a suite of services that collectively enable all taxpayers to access tax benefits and fulfill their filing obligations — including connections to higher-touch filing services (e.g., the Volunteer Income Tax Assistance program, or VITA), to year-round engagement with the tax system, and more.

Chapters 2-9 at a glance

2. Data import / pre-population: using IRS data to streamline the filing process through Direct File. A conceptual framework for the goals and limits of pre-population, and a detailed discussion of the various data sources that have been and can be used, including policy changes needed to make them viable for Direct File use.
3. Tax scope: which tax situations Direct File does and does not cover. How to think about and approach the sequencing of new tax scope items, how to balance the competing pressures of agile development with advance planning, and how far Direct File should eventually go.
4. State taxes: how Direct File supports taxpayers’ state (and local) filing requirements. Background on the status quo interaction of federal and state filing systems, an exploration of the different high-level approaches Direct File could have taken to the state filing question, and a detailed discussion of the implementation of Direct File’s chosen solution, including limitations of the approach and future alternatives.
5. Identity verification: what identity verification requirements Direct File did and should impose on taxpayers. A discussion of the cross-government dynamics, including NIST standards and bureaucratic ambiguity, that led Direct File to be put behind rigorous identity verification in the first place, and possible alternative approaches for government in general and Direct File in particular.
6. Usage and cost levels: an exploration of the dynamics of Direct File’s costs and usage. A detailed analysis of usage in 2024 and 2025, including demographic and state-based patterns, and benchmark projections of what usage may have looked like in future years. A discussion of different usage goals a team may set.
7. Promotion and outreach: a brief discussion of how to get Direct File into the hands of taxpayers, what we know about what has worked so far, and which avenues future organizers should explore.
8. Integration between Direct File and other IRS systems: a brief discussion of how Direct File could become the seed of wider online transformation at the IRS.
9. Fact graph, and a new way of building civic technology products: a proposal for a new way of building civic technology products, by building on the key backend components that drove Direct File — including a robust fact graph logic engine and a backend GUI that allows cross-functional teams to envision, understand, and collaborate on the product. Fully implemented, this way of working would open up new possibilities for state tax support in Direct File, and other opportunities across the government technology space.

Part 3: Beyond Direct File: A Vision for the Future of Tax System Access (Chapters 10-15)

Because filing a tax return is the most important interaction most taxpayers have with the tax system, Direct File is naturally the central component of any program to improve tax access. But it is not the only component. These chapters explore a variety of other issues that are critical to a functioning and accessible tax system, particularly for low-income populations and new or intermittent tax filers — all beyond the bounds of the core project of Direct File.

The nature of these chapters varies significantly. Some envision statutory changes while some describe administrative projects; some contain specific policy proposals while others simply provide a framework for action.

As these projects are not based on Direct File itself, some of them could in principle be advanced in the short term, even with Direct File currently canceled and new IRS initiatives essentially on hold at the federal level.

The original user research that informs these chapters

These chapters stem not just from our experience working on Direct File and our own desk research, but also from a round of 15 user research sessions we conducted in the summer of 2025, which explored primarily the topics in Chapters 11 and 12 — and, tangentially, the topics in Chapters 10 and 14. These were one-hour structured sessions with one participant at a time, probing attitudes and beliefs about different aspects of the tax system. User research in this context should not be confused with usability testing: we were not putting software products in front of research participants and asking them to use them. Rather, we were learning how taxpayers think about and interact with the tax system. Plenty of reforms that may seem natural and important to tax policy experts may not actually align with the desires and preferences of most taxpayers. If we want to build an intuitive tax system that meets people where they are, we indeed need to learn where they are. A single round of user research sessions is not gospel; we are not proposing to overhaul the entire American tax system based on the opinions of 15 people. But, even a small number of conversations can generate a relatively strong signal about the intuitions of the average taxpayer.

We recruited the 15 taxpayers through an online platform for research participants. We spoke with a janitor and business owners, delivery drivers and teachers; we spoke to a mix of married, divorced, or never married people, and a mix of household makeups ranging from individuals living alone to families who shared intergenerational homes or had mixed citizenship status. Nine participants had children under the age of 18 currently in their care (two of whom also had children over 18); one had previously had children under the age of 18 in their care; and five had never had children in their care. We observed adults who claimed their children every year on their tax returns, those who claimed children in alternating years with another parent, and those who had nonbiological children in their lives whom they treated as their own. Household incomes ranged from less than $30,000 up to $100,000, with an average household income of roughly $58,000. We prioritized (for purposes of the material discussed in Chapters 11 and 12) taxpayers with gig economy income, and taxpayers with children and non-traditional family situations. As a basic demographic breakdown:

Gender	Female: 10 Male: 5
Race	White: 8 Black: 3 Hispanic/Latino: 2 Middle Eastern and North African: 1 Asian: 1
Household income	<$30K: 2 $30-60K: 6 $60-100K: 7
Marital status	Divorced/separated: 6 Never married: 5 Married: 4
Filing status	Single: 6 Head of Household: 4 Married Filing Jointly: 4 Married Filing Separately: 1
Earned income types	1099 and W-2: 9 1099 only: 4 W-2 only: 2
Kids <18 in your care	Yes: 9 No: 6

Where possible, we include quotes from these research participants to illustrate key points. To protect their anonymity, we refer to them as, for example, “Participant #7.”

Systemic reform and coherence

As we explore topics beyond Direct File, we encounter tension between interventions that might best advance a narrow or short-term goal on one hand and those that would promote the overall tractability and organization of the tax system on the other. Throughout these chapters, we take as axiomatic the importance of the latter goal: we believe that, in the long run, taxpayers will be best served by an overall tax system that is well-defined and coherent. In particular:

• We believe the tight integration of federal and state returns is a major boon to tax system access and reducing administrative burden, and that any reforms that would interrupt that integration would likely do more harm than good in the long run.
• We believe that reforms should aim to make standard tax filing work better across the board, rather than creating complex, bespoke workarounds for specific populations. A system built on carving out alternate processes risks becoming fragmented and inequitable — as well as harder to administer and to navigate. This theme is explored in Chapter 2: Direct File and data import and again in Chapter 14: Non-filers.

Chapters 10-15 at a glance

Chapters 10 through 12 cover the basic functioning of the tax system:

10. IRS communications. We explore how the IRS does and should communicate with taxpayers, especially regarding electronic communications. While this topic is not the most significant on its own, we believe it is an important prerequisite for many of the other potential reforms we describe in later chapters.
11. Real-time taxes. We explore topics related to taxpayers’ year-round interactions with the tax agency, including estimated payments, withholding, and potential advance payments of credits — ensuring taxpayers are paying roughly the right amount throughout the year, and do not receive too-large lump-sum refunds or owe money back to the IRS at tax time. These topics are especially relevant for gig economy taxpayers, and the overall topic represents a major gap in the current system, which largely conceives of tax filing as a once-a-year exercise. We believe there is a strong case for a new product that would help taxpayers manage their year-round tax obligations, supported by new reporting and data pipelines. Taken together, these measures would constitute a glide path to a pay-as-you-earn system.
12. Dependents. We discuss which taxpayers are allowed to claim which children, and how the tax system handles cases of conflicting claims. The chapter argues that the current rules are not well aligned with taxpayer intuitions — creating classes of children who cannot be claimed, creating widespread perceptions of noncompliance and coverage gaps that are largely illusory, and creating significant administrative burden and overhead. We propose a new framework for dependent rules that would resolve most of these issues.

Chapters 13 and 14 discuss reaching taxpayers who fall through the cracks:

13. Additional tax filing assistance. Direct File is a DIY option for taxpayers who can file returns without much hands-on support. This chapter explores what IRS policy should be for taxpayers who do need more assistance, and how the VITA program in particular can evolve to better meet these needs.
14. Non-filers. Even in a world of Direct File and a highly-accessible VITA, there will be some taxpayers who still fail to file returns. This chapter explores how to serve those persistent non-filers and proposes a framework to ensure they receive their refunds without compromising the overall structural integrity of the tax system. We propose in particular the establishment of a “non-filer filing season” in May-November and the creation of a Non-Filer Team at the IRS that would target non-filers in this period; and we lay out some of the programmatic interventions the team should experiment with. We also discuss the role of “simplified filing” in the tax system, and the state-level implications of our recommendations.

One final chapter concerns the management of the tax ecosystem:

15. IRS and private providers. Private providers play a large role in the tax ecosystem, and even in the most aggressive visions of IRS modernization explored here, they will continue to do so in the future. This chapter explores how the IRS ought to conceptualize its relationship with these providers, and proposes a realignment to better promote competition and consumer protection.

Of course, this list of topics is not comprehensive. It does not include items like IRS organization and staffing, especially in rebuilding from its current weakened state; it contains only limited engagement with the substantial issue of regulating non-credentialed return preparers and other third-party actors; it does not include structural technology reforms, like replacing aging IRS systems; there are various niche taxpayer service items that we do not particularly address, including amended returns (a persistent technological and administrative challenge), the IP PIN program (which can create access barriers for low-income taxpayers), math error notices, and many more. Still, we think an IRS reform program built around the items we discuss here would yield significant progress — promoting taxpayer access, advancing trust in the integrity of the U.S. tax system, and building confidence in the IRS as an honest broker of the nation’s shared resources.

Part 1: The Story of Direct File

Part 1: The Story of Direct File

Chapter 1: The story of Direct File and tax access reform in the Biden Administration

Gabriel Zucker

• The story of how Direct File came to be is of relevance for reformers seeking to create similar change in the future — perhaps with more staying power. Though we do not present the story with any particular thesis, readers may keep in mind a handful of themes: time to launch, implementation strategy, opposition from private industry, legislative versus administrative action, and agency independence.
• By the outset of the Biden Administration, filing reform was nearing the political agenda due to a variety of independent influences: the faltering of the Free File Alliance (1.1.1), the interest of the civic tech movement (1.1.2) and child allowance advocates (1.1.3), and experience of pandemic-era assistance programs administered through the tax code (1.1.4).
• When the expanded CTC was enacted in the American Rescue Plan in March 2021, the government needed an easy-to-use simplified filing tool for traditional non-filers, but opted not to build one, working with external non-profit Code for America to create GetCTC instead. The experience whetted the appetite in government to take control of its own destiny, and galvanized support from aligned advocates. GetCTC could later have been adopted into government and developed into Direct File, but ultimately was not, as the expanded CTC fell out of the policy spotlight. (1.2)
• Instead, the decision to build Direct File was the result of about a year and a half of internal administration debate and study. Intuit’s departure from the Free File Alliance in July 2021 was a critical part of this process (1.3.1). The debate featured two separate policy processes, in late 2021 (1.3.2), and early 2022 (1.3.3), and was pushed to a final resolution by the passage of the Inflation Reduction Act in August 2022, which directed the IRS to study a free, public filing option (1.3.4).
• The lack of a public commitment to pilot Direct File until after completion of the report to Congress meant opposition did not coalesce early in 2023 (1.4.1), though support from states, external stakeholders, and VITA programs was also hard to organize (1.4.3, 1.4.4, 1.4.5). Building the product was an all-out sprint to the finish line (1.4.2) and ultimately created a very tight schedule during the pilot roll-out in 2024, which featured some clashes between agile methodology and traditional government ways of working — but the product was clearly a success (1.5).
• It was announced in May 2024 that the Direct File pilot would be converted to a permanent program, generating more support and enthusiasm from a range of angles, though also further galvanizing right-wing opposition (1.6). When Democrats lost control of all branches of government in November 2024, it was only a matter of time before Direct File would be killed, which it eventually was — effectively in March 2025, and officially in fall 2025 (1.7).
• Many other taxpayer experience initiatives in the IRS Strategic Operating Plan (SOP) to implement the Inflation Reduction Act largely had no more staying power than Direct File, and in some cases took even less root in the agency. (1.8)

Part 2: Building Direct File

Part 2: Building Direct File: Policy and Strategy

Chapters
Chapter 2: Direct File and data import / pre-population
Chapter 3: Direct File and tax scope
Chapter 4: Direct File and state taxes
Chapter 5: Direct File and identity verification
Chapter 6: Direct File usage and cost levels
Chapter 7: Direct File promotion and outreach
Chapter 8: Integration between Direct File and other IRS systems
Chapter 9: Direct File, fact graph, and a new way of building civic technology products

Chapter 2: Direct File and data import / pre-population

Gabriel Zucker, Chris Given

• Expanding pre-population — the use of IRS data to streamline the tax filing process — has the potential to make tax filing quick, easy, and painless; it would end tax filing as we know it. (2.1.1)
• Within the confines of the existing U.S. tax system, though, we cannot eradicate tax returns altogether. Attempting to do so would, in fact, likely hurt the taxpayers we are trying to help. The most automated tax system we can achieve is a Direct File-like product with maximum pre-population, and reformers should keep energy on that track, rather than attempting to build a parallel and more-automated track. (2.1.2)
• Maximizing pre-population in Direct File will require technical improvements in the timing and processing of third-party information returns. (2.2.1)
• It will also require using data from a number of other data sources (2.2.2-2.2.5), though perhaps not from state tax departments (2.2.6).
• Next steps and recommendations:
  • Direct File should continue to expand pre-population to include more data sources and data items, remembering that delivering pre-population is an iterative process. (2.4.1)
  • Direct File may have to reimagine the connection of pre-population to the taxpayer experience, to achieve the full promise of the functionality. (2.4.2)
  • Direct File should work with state partners to explore the pre-population of state tax department data, though it is possible the effort required will not be worth it. (2.4.3)
  • A variety of improvements will be needed to ensure the sufficiency of third-party information return data, including retrieving and storing W-2 Boxes 12-20 data, and changing reporting deadlines for various Forms 1099 (which will likely require statutory changes to 26 USC 6071). (2.4.4)

Chapter 3: Direct File and tax scope

Chris Given, Gabriel Zucker

• The tax code exhibits complexity along multiple dimensions, which are important inputs to any discussion of Direct File’s scope. (3.1)
• Direct File never determined its “end point” — how much tax scope would the product support at full maturity? The team discussed a handful of options, ranging effectively from 50 to 95 percent coverage. The lack of an agreed-upon endpoint made nearer-term scope decisions more difficult to manage. (3.2)
• Direct File’s decision to create a best-in-class taxpayer experience meant that adding each new component of tax scope would be time-consuming, and that reaching “full” coverage would be a slow, iterative process. (3.3)
• Generating consensus on scope decisions was sometimes challenging, often highlighting the conflict between agile technology methodology and government practice (3.4.1). At a granular level, though, Direct File used a data-driven four-part decision framework for new scope that worked relatively well (3.4.3).
• Perhaps counterintuitively, Direct File coverage is sometimes determined much more by small-ticket scope items (a series of small scope restrictions covering various edge cases) than by big-ticket items, like a new form or a new source of income (3.4.5). The impact of all these small restrictions makes it incredibly hard to design effective eligibility screeners (3.4.6).
• Next steps and recommendations:
  • Direct File should decide on an “end point” — the scope it will have at full maturity. This will allow the project to tell a clear story about its development and provide meaningful public multiyear clarity to all stakeholders, allowing all manner of decisions to be made within a more coherent framework. We do not explicitly endorse an end point, though we are inclined toward more modest options, perhaps in the vicinity of 70% of taxpayers. In the near term, Direct File must reach a point of “baseline maturity,” at which it covers about half of taxpayers. (3.5.1)
  • While policymakers may make large-scale decisions about Direct File’s end point and pace of overall development, smaller-scale scope details should be determined by the implementation team. (3.5.2)
  • Direct File should publish decisions about the coming year’s scope earlier, even at the risk of pursuing a more conservative roadmap. Unexpected excess capacity can be moved to other parts of the product. (3.5.3)
  • Direct File’s existing process for smaller-scale scope decisions was largely working well and should be continued (3.5.4). Because of the importance and difficulty of eligibility screeners, Direct File should prioritize using pre-population to automate significant portions of the eligibility determination (3.5.5). For communications reasons and due to modest level of effort, Direct File should also prioritize addressing items of scope that generate de facto income caps (3.5.6).

Chapter 4: Direct File and state taxes

Gabriel Zucker, Chris Given

• State (and local) taxes are an integral part of the U.S. income tax system, and the interconnection between them and federal taxes is complex. While the systems are separate, vast parts of their administration overlap, and there is considerable technological consistency across states. (4.1)
• Before launching the product, Direct File considered five options for state support in Direct File. The option ultimately selected — state filing in a separate product, supported by a data sharing API — was option #2 of 5, and was the lightest-weight option other than the option of providing no state support at all. More intensive options were essentially infeasible under Direct File’s early timelines. (4.2)
• Integrated state filing solutions worked well for most taxpayers in 2024-2025 (4.3.3), but the solution was not good enough for a meaningful minority of taxpayers, and there was also no clear path to scale to multi-state returns (4.3.4).
• Thanks in part to careful design and an intentional engagement strategy, Direct File was making fast progress with states, and we believe it was on the path to coverage in about 45 states by 2027 (4.3.5). That said, the economics of separate state filing solutions in the long run was not a solved problem (4.3.7), and the pursuit of different options in different states exacerbated some of the financial issues by preventing states from taking full advantage of economies of scale (4.3.8).
• Next steps and recommendations:
  • Direct File should not compromise on state filing. Direct File was right to restrict access to those states that offered specified state filing solutions. Free, high-quality state filing should be guaranteed for any Direct File user — even if it means accepting a slower pace of national scaling. (4.4.1)
  • Given the benefit of time to revisit the issue, Direct File should explore creating a way for state tax departments to configure their own state tax functionality directly in the Direct File product. We believe the reimagined tech stack discussed in Chapter 9 could make this approach possible. It would solve otherwise-thorny issues regarding federal-state dropoff, multi-state returns, and the economics of state filing options. States pursuing this option would not need to maintain their own standalone filing software. (4.4.2)
  • For states that continue to provide standalone state filing options via the integrated API model from 2024-2025: Direct File should explore supporting the creation and maintenance of a public multi-state state filing product, to ensure options and a healthy state filing ecosystem (4.4.3). Direct File should also make improvements to the integration model (4.4.4), and set user experience standards for the standalone state products, so as to ensure a consistently excellent user experience (4.4.5).
  • Direct File was correct to reject the possibility of having federal Direct File staff directly build state filing functionality. (4.4.6)
  • Congress should consider providing financial resources for states joining Direct File. (4.4.7)
  • Despite all of this attention to state filing, the federal return is the front door of the American tax system, and policymakers at the federal and state level need to be careful to ensure federal options remain at the center of policy reform and taxpayer communications. (4.4.8)

Chapter 5: Direct File and identity verification

Gabriel Zucker, Chris Given

• Direct File was classified as an IAL2 system, which meant that taxpayers had to undergo rigorous identity verification procedures to use it. Identity verification was a barrier in 2024 and 2025.
• Direct File’s classification was, on its face, an accurate implementation of NIST 800-63-3, the government-wide guidance on identity verification at the time. The fact that it was the correct implementation suggests issues with the NIST 800-63-3 guidance in practice (5.1). The Direct File classification also introduced de facto inconsistencies between IRS’s treatment of public and private tax filing software (5.2).
• Because of its IAL2 classification, Direct File had to use the private service ID.me to identity-proof taxpayers, as ID.me was the only service then able to implement IAL2 at scale. Details of ID.me’s implementation introduced further barriers on top of the inherent barriers of IAL2, which might have been somewhat ameliorated by the use of Login.gov, if that service had been ready at the time. (5.3)
• Next steps for government digital identity in general:
  • NIST 800-63-4, the newest revision to the government-wide guidance, was released in 2025. It makes improvements to NIST 800-63-3 on some fronts, though does not fully resolve the issues. (5.4.1)
  • Culturally, the relationship between NIST and other federal agencies lends itself to buck-passing of risk assessments, which tends to lead information systems to be overly restricted. This culture must change. (5.4.2)
  • Login.gov is a vital service, which needs to provide robust offline fallbacks for users who fail online identity verification, and must continually improve as it makes verification as easy and equitable as possible. (5.4.3)
• Next steps for Direct File in particular:
  • Direct File should consider reclassifying portions of functionality at IAL1, in light of NIST 800-63-4 (5.5.1), and creating alternative non-logged-in service patterns. That said, the growth of pre-population functionality makes this choice less clear cut than it otherwise might be (5.5.2).
  • Direct File should adopt Login.gov as an option for identity verification. (5.5.4)
• This entire section is a point-in-time snapshot of the identity verification space; generative artificial intelligence is very likely to vastly disrupt the entire ecosystem. This may well make many of our specific comments here irrelevant. (5.6)

Chapter 6: Direct File usage and cost levels

Gabriel Zucker, Chris Given

• Direct File had 1% take-up among eligible taxpayers in both 2024 and 2025. 2025 take-up was significantly hampered by the Administration’s sabotage of outreach efforts, driving up the service’s cost per return. Actual costs in 2025, though, validated prior estimates. (6.1)
• Based on the experience to date, in an alternate world where Direct File had been supported by the administration and had continued to develop, we project there would have been nearly 10 million returns a year by filing season 2028, at a cost per return of $12. (6.2)
• Direct File had not answered the question of its target usage: what would be enough usage to justify its ongoing existence? We believe Direct File should probably have targeted a modest cost per return figure, perhaps with target usage weighted toward certain priority populations. If Direct File’s goal were net social savings (cost per return below average cost to file), it would have already achieved this goal in 2025. (6.3)
• Direct File users in its early years tended to be young, single, childless wage earners. Because this was the user base, family credit claims tended to be lower and therefore refunds smaller than the population at large. (6.4)
• Per capita Direct File usage varied only somewhat modestly by state in 2025. States that had participated in year one had higher usage in year two, as did states where larger fractions of the population self-prepare their returns. (6.5)

Chapter 7: Direct File promotion and outreach

Gabriel Zucker, Chris Given

• There is not ironclad evidence regarding which promotion and outreach methods were most effective for Direct File, though two years of experimentation allow for some tentative conclusions.
• With any outreach and promotion, a single, clear message and a direct call to action are paramount.
• Once Direct File reaches “baseline maturity,” the level at which at least half of taxpayers are eligible, it could be promoted as a default option, including on IRS.gov. This formal messaging is likely to be transformative.
• The IRS should augment this website positioning with additional direct communications to taxpayers.
• The federal megaphone, in general, is very strong, and efforts by other actors, though they can be meaningful at the margins, tend to be poor substitutes.

Chapter 8: Integration between Direct File and other IRS systems

Gabriel Zucker, Chris Given

• As is relatively normal for transformation digital products, Direct File was essentially built on an island from other IRS systems. In the medium term, it would have been appropriate to integrate Direct File more closely with other IRS online systems and services.
• Closer integration might allow for: (1) promoting Direct File more systematically across IRS.gov, (2) applying Direct File practices to the deployment of more regular year-round communications to taxpayers, and (3) moving taxpayers seamlessly between filing and services like installment agreements, withholding, estimated taxes, payments, and amended returns.

Chapter 9: Direct File, fact graph, and a new way of building civic technology products

Jennifer Thomas, Chris Given, Gabriel Zucker

• Translating the tax code into a product that asked taxpayers tractable questions and produced accurate returns was a challenging process, which could have been made easier with more and better back-end tools and innovations. Direct File employed two innovations in this regard — (1) the fact graph, a declarative business rules engine that modeled the tax code and allowed the product to make inferences about incomplete information, and (2) a nascent “taxpert interface” that allowed non-engineers to visualize and interact with the product. We envision expanding on both of these components to develop a new way of building civic technology, which we dub FactML in honor of its principal component. (9.1)
• Such back-end innovation would: (1) allow Direct File to be built and scale faster and more reliably, (2) allow states to configure state filing functionality directly in the Direct File product, and (3) open the door for better, more efficient products throughout the civic technology space, including cross-domain linkages. (9.2)
• The impact of these changes can best be envisioned by picturing the process of states building new tax functionality in Direct File, in a FactML world. (9.4.1)
• FactML fundamentally consists of four components: (1) Fact Markup Language (FactML), a generalized XML-based specification for defining legal code as software code, (2) Factual, a portable business rules engine that parses the relationships in FactML, (3) Formative, an opinionated interface library for a user-facing application that collects and asserts information, based on the previous layers, and (4) Formative Studio, a GUI for viewing, annotating, and testing the interfaces and logic contained in the previous layers. Formative Studio is the linchpin that enables effective cross-functional collaboration and makes the entirety of the product tractable in a single interface. Direct File employed functional versions of (1), (2), and (3) (though they each require some documentation and additional abstraction), but only the most nascent version of (4). (9.4.2)
• FactML would solve collaboration and documentation challenges inherent in building such a complex product, it would use this new collaboration potential to unlock state support directly in the product, and it would dramatically speed testing. (9.4.3)
• Taken together, these reforms are a way of working more than they are a single deliverable product. They could be advanced in a variety of contexts, but getting to a FactML world is not a simple linear process. (9.5)

Part 3: Beyond Direct File: A Vision for the Future of Tax System Access

Part 3: Beyond Direct File: A Vision for the Future of Tax System Access

Chapters
Chapter 10: IRS communications
Chapter 11: Real-time taxes
Chapter 12: Dependents
Chapter 13: Additional tax filing assistance
Chapter 14: Non-filers
Chapter 15: IRS and private providers

Chapter 10: IRS communications

Gabriel Zucker, Chris Given

• The IRS has insufficient mechanisms to communicate easily with taxpayers in the 21st century, which has implications for many agency priorities. (10.1.1)
• Taxpayers would appreciate easier-to-access communications from the agency, something they have to come to expect from institutions in their lives — including via email. At present, though, they expect the IRS will only communicate with them for punitive purposes. (10.1.2)
• Recommendations:
  • The IRS must adopt an improved, user-friendly secure messaging platform, for those sensitive messages that require encrypted communication. (10.2.1)
  • The IRS should open the door to communicating with taxpayers via email (10.2.2). The agency should assess each communication on a case-by-case basis to determine how much can be disclosed via email, taking seriously the risk of taxpayers not receiving communications sent through other mediums, and abandoning the too-broad blanket rule of no federal tax information in unencrypted communications (10.2.3).
  • Building trust in IRS communications will take time, and the agency should pursue a program of intentionally building that trust, starting with simpler and lower-stakes messages (10.2.4). Communications should use plain language and a friendly tone, and should be user tested to ensure they are well crafted (10.2.5).
  • The IRS may also consider mass marketing campaigns in support of these direct communications reforms. (10.2.6)

Chapter 11: Real-time taxes (estimated payments, withholding, advance payments, and the gig economy)

Chris Given, Gabriel Zucker

• Even in our fast-paced world, the tax system operates on a fundamentally yearly cadence, with taxpayers paying in throughout the year, and potentially due for a rude surprise if their payments turn out at filing time to have been wrong. Making the tax system more real-time would be a tall order, and we do not necessarily argue it deserves to be prioritized given the immense effort involved. But, this chapter contends it would be a positive project, and lays out what it would entail.
• At present the systems to manage and track payments throughout the tax year are antiquated and leave the IRS without clear data on a taxpayer’s status before filing time. Some taxpayers — especially gig economy workers — also must navigate complex processes to stay in compliance. Advance payments of tax credits, if implemented, add another layer to the real-time tax system, requiring some taxpayers to adjust their withholding or estimated payments to stay in compliance. (11.1)
• Many taxpayers are very uncomfortable when it comes to withholding and estimated payments, especially in the gig economy (11.2.1). Most taxpayers also have a strong interest in ensuring they are paying in the right amount. Owing the IRS at filing time is seen as a very bad outcome to be avoided at all costs; paying in far too much and receiving a large refund is preferable but still far from ideal (11.2.2).
• Taxpayers would find it very useful if the IRS helped them track and fulfill their tax obligations throughout the year, especially for gig economy income (11.2.3). But they would not want such a service to automatically make changes on their behalf, and they would be concerned about overreach (11.2.4).
• Next steps and recommendations:
  • The IRS should introduce withholding for gig economy workers. Step one is probably a pilot with a few large gig economy employers. If the pilot goes well, scaling it further may require statute. (11.3.1)
  • The IRS should create a new product that helps taxpayers track and fulfill their tax obligations throughout the tax year (11.3.2). Using status quo data, there is a limited amount this product could do (11.3.3); to make the product more useful, the IRS needs more real-time data during the year, a reporting reform that may be achievable in the context of H.R. 1 implementation (11.3.4). With this additional data in hand, the real-time product would become immeasurably more powerful and useful (11.3.5).
  • Advance payments of tax credits represent a complex real-time taxes challenge that we believe policymakers should approach thoughtfully. (11.3.6)
  • Taken together, these piecemeal reforms would move the United States closer to a pay-as-you-earn (PAYE) system. Some reformers may advocate a more thorough and holistic PAYE reform than the a la carte measures laid out here; but we believe such smaller iterative measures are the most viable route to larger reforms. (11.3.7)

Chapter 12: Dependents

Gabriel Zucker, Chris Given

• Current law and de facto practice about claiming dependents on tax returns are complex and inconsistent. (12.1)
  • Despite harmonization efforts, tax benefits have different legal definitions of a child (12.1.1). In practice, in most cases of conflicting dependent claims, the de facto resolution process is “first come, first served” (12.1.3).
  • The IRS audits taxpayers and calculates aggregate statistics of dependent claiming on the basis of methods that do not match the de facto filing administration practices. (12.1.4, 12.1.5)
  • Contrary to these rules, taxpayers tend to claim children on the basis of informal agreements within their families or court-ordered formal agreements like custody agreements — arrangements most taxpayers do not know are technically unlawful. (12.1.6)
• The current system creates a number of serious problems. (12.2)
  • Due to the shape of the law and the vagaries of its implementation, some children cannot be claimed by any filer for one or more tax benefits, in apparent contradiction of the policy intent. (12.2.1)
  • Many children are claimed by, technically speaking, the wrong family member; we call these intra-family misclaims. These intra-family misclaims are generally benign mistakes in policy terms, but they create the appearance of tens of billions of dollars of improper payments in the EITC — which have driven policymakers to implement access-restricting measures in the name of reducing fraud. They also may vastly inflate statistics measuring the EITC participation gap, causing policymakers and third-party stakeholders to go looking in the wrong place for Americans missing out on their tax benefits. (12.2.2)
  • The de facto adjudication rules can make it difficult for the family to claim a child, in case of conflict (12.2.3). Meanwhile, the sheer complexity of the system also creates administrative burden (12.2.4) and feeds the damaging widespread impression that the tax system cannot be navigated by everyday people (12.2.5).
• Taxpayers’ intuitions about the child-claiming rules — explored in a series of user research sessions — differ significantly from the current rules, and point toward a new paradigm. (12.3)
  • In a series of scenarios where we asked taxpayers which of multiple adults should claim a child, only 21% of votes actually matched current law. (12.3.2.1)
  • There was widespread disagreement about who should claim, and many taxpayers could not articulate a consistent principle that would identify the right claimant in every case. (12.3.2.2)
  • Taxpayers generally believe that, within reason, families should be able to decide among themselves who claims a given child (12.3.3.1). Some of these intra-family choices are consistent with behavior some policymakers would consider “gaming” — but taxpayers did not tend to see such behavior as inappropriate or immoral (12.3.3.2). Taxpayers tend to believe legal agreements like divorce decrees or custody agreements should be the first line of defense in determining who claims a child, and tend to see these kinds of agreements as a special case of the family working it out among themselves (12.3.3.3).
  • In the case that there is a conflict between family members, though, taxpayers expect there to be clear and objective rules to resolve the conflict (12.3.3.4).
  • When thinking through who truly ought to claim a given child, most taxpayers had an intuition of applying some type of relationship test and some type of support test to determine claims. There was disagreement, though, about how those tests would be defined. (12.3.4.1, 12.3.4.2)
  • Taxpayers did not voice any support for the idea that children might be classified differently for different benefits, or any understanding that such splitting is part of current law. (12.3.4.3)
• Recommendations:
  • All tax programs should use a common dependency definition, under which one child is only ever claimed by one tax unit. One tax unit should claim one child, with that child consistently eligible or ineligible for all relevant programs to the greatest extent possible. (12.4.1)
  • Rather than prescriptively identify the one tax unit who is definitively able to claim a given child, the law should define a permissive set of claimants, any of whom may validly claim the child. Within this set of claimants, families are encouraged to decide for themselves who claims the child. As long as only one of the valid claimants claims the child, the government does not need to second-guess which one it is. We provide a few possibilities of how this permissive class may be defined. (12.4.2)
  • In cases that multiple valid claimants do claim the child, there should be a detailed set of tiebreaker rules that the government will use to adjudicate between the rival claims. These rules only come into play if there are multiple claims actually made. The tiebreakers should be based on the taxpayer intuitions laid out in earlier sections. (12.4.3)
  • This new paradigm raises implementation challenges, since not all claims come in at the same time. Equitable administration may require some child claims to be held provisionally until the government is confident there is not a rival claim. (12.4.4)
  • The IRS needs a quick and human-centered resolution process to adjudicate rival claims, one which does not require snail mail for administration. (12.4.5)
  • The new regime does increase the capacity for unsympathetic “gaming” — that is, deliberate manipulation for tax advantage — in child claims. Policymakers may reduce the incentive to game by removing quirks in existing tax law, such as the three-child limit for EITC. They may also consider granting IRS statutory authority to prevent specific cases of unsympathetic gaming — specifically those that entail high-income parents shifting child claims to lower-income family members. (12.4.7)
  • While many of these recommendations may seem challenging to implement, we encourage policymakers to consider the myriad untenable contradictions in the status quo, including widespread legal instruments that violate the terms of the tax law; widespread audits, enforcement actions, and alleged overpayments based on benign intra-family misclaims; and the simple fact that, more than 20 years after the establishment of the current policy regime, we are as far as ever from everyday taxpayers actually knowing, understanding, and following the letter of the law.

Chapter 13: Additional tax filing assistance

Gabriel Zucker, Chris Given

• Policymakers have not articulated coherent goals when it comes to high-touch, hands-on tax filing assistance. Do all taxpayers have the right to free assistance more involved than DIY filing? Only some taxpayers? If so, which? We contend, at a minimum, that as long as publicly-funded high-touch hands-on assistance remains relatively limited, it should be targeted to those taxpayers with the highest needs. (13.1)
• The IRS and/or other tax access stakeholders should explore the creation of additional levels of public tax preparation assistance between the extremes of DIY tax filing with chat support (Direct File), on one hand, and full-service, someone preparing your return for you (VITA), on the other. It is possible there is a use case for additional models between these extremes. (13.2)
• The IRS should manage a centralized front door expediting access to all publicly-supported tax filing services, helping taxpayers understand their options and find the one that is right for them (13.3). Moreover, Direct File should be expanded to support additional levels of service, such that most taxpayers can get started with their return on Direct File and be routed dynamically to the appropriate level of assistance they need, should live chat turn out to be insufficient. This means, for example, a half-finished Direct File return could effectively become a digital VITA intake form for taxpayers who find themselves unable to complete the return alone (13.4).
• VITA program rules should be amended so that programs are judged not on the total number of returns they prepare, but rather on how well they do at engaging specific hard-to-serve populations with specific hard-to-prepare returns — and at helping taxpayers, perhaps over the course of several years, achieve self-sufficiency in filing. These changes will help ensure that limited VITA capacity is focused on those things that only VITA can do. (13.5)

Chapter 14: Non-filers: better data, a post-April focus, a dedicated team, automated second-best returns, and simplified filing

Gabriel Zucker, Chris Given

• In this chapter, we define true non-filers as those taxpayers who either must or should (to get a refund) file, who remain non-filers even when the distortions induced by dependent rules have been eliminated from the analysis (see Chapter 12), and who remain non-filers even once Direct File has been scaled up as a free and easy option.
• Though the IRS may embrace some extraordinary special measures to engage non-filers, the paramount long-term goal is always that taxpayers ultimately file a standard full return. Thanks to the fundamental parameters of the U.S. tax system, only a full return can guarantee that taxpayers are getting the complete benefits they deserve from the tax code. (14.1)
• Because of the subtleties and unknowns in engaging non-filers, our best recommendations in this section concern not specific programmatic interventions, but rather organizational ideas: meta reforms that create a fertile environment for closing the non-filer gap. (14.2)
  • The IRS needs better internal and external data on the non-filer population. Internally, the agency needs a by-name list of (likely) true non-filers, to target interventions and measure success. Externally, it needs to publish credible aggregate-level estimates of the size and characteristics of the non-filer population, to track success and guide third-party efforts. (14.2.1)
  • The IRS should establish an ecosystem-wide practice of engaging non-filers after April 15, thereby splitting the filing season into a traditional season (January-April) and a non-filer period targeted at those remaining (May-November). This approach would allow messaging and interventions to be better targeted at the right taxpayers; it would take advantage of the additional confidence the IRS can have in its own data later in the year; and it would better spread capacity across the year. (14.2.2)
  • The IRS should create a Non-Filer Team whose dual mandate is to get non-filers as much of their refund for recent tax years as possible, while also setting them back on the path to full filing in future years. The Team’s interventions should be run as randomized controlled trials, so they can quickly learn what works and iterate. For a variety of reasons, such a team inherently has to exist at the IRS, rather than at a third-party organization. (14.2.3)
• The Non-Filer Team’s mandate would be to experiment and figure out what works. We discuss a few of the ideas they may explore implementing. (14.3)
  • User research suggested that many taxpayers do not know they can file late, and a mere reminder of this fact could be helpful in moving the needle. Taxpayers also appeared universally enthusiastic about the idea of the IRS sending reminders to file, the type of thing they generally expect institutions in their life to do (14.3.1). There are a handful of design issues to resolve regarding any such notices (14.3.2).
  • The IRS might also consider sending automated “second-best” returns to true non-filers during the second half of the year. Despite the general principle that taxpayers should file full returns, the agency might conclude that something is better than nothing for taxpayers who otherwise might not file at all (14.3.3). Taxpayers, though, were split about this idea in user research (14.3.3.2), and it is not entirely clear if there is a promising implementation here that does not collapse back into a Direct File referral (14.3.3.3).
  • The Non-Filer Team may also explore collaboration with VITA to target true non-filers. (14.3.4)
• Simplified filing is a term of art defining a niche alternate filing process whereby filers with functional-zero income (low enough that they are not required to report it, and that they are not eligible for any credits which depend on the precise amount of their income) might file without reporting any income whatsoever. Unlike in 2020-2022, the current federal tax code has no provisions that would occasion simplified filing, but even if it did, we believe it would not be worth pursuing a standalone simplified filing solution; better to streamline the standard Direct File pre-populated filing pathway as much as possible than to create competing pathways to file. In the case where states have use for simplified filing, we still believe it would be more productive to funnel those taxpayers through non-simplified Direct File. However, in a world without Direct File, relevant states may pursue their own limited standalone simplified filing products. (14.4)
• States may consider creating their own Non-Filer Teams along the lines discussed in 14.2.3. These Teams could target the additional population of taxpayers who file federal, but not state, returns. (14.5)

Chapter 15: IRS and private providers

Gabriel Zucker, Chris Given

• The IRS has historically played a limited oversight role regarding the private tax software providers and paid preparers who comprise the broader tax filing ecosystem — in large part a function of the IRS’s limited capacity and leverage. In a world with a resurgent Direct File, we believe it would be time for this posture to self-consciously change, to ensure better outcomes for taxpayers.
• While taxpayers should of course be allowed to use other services, the IRS should — unlike in 2024 and 2025 — unapologetically promote Direct File as a default, and even preferred, option for tax filing. (15.1)
• The IRS should more rigorously enforce accuracy standards for private tax preparation software — a task that would be made immeasurably easier by the existence of Direct File tax logic. This comprises both hard accuracy (given a set of inputs, does the software determine the correct eligibilities and amounts) and soft accuracy (given a certain taxpayer situation, how likely is it that they are successfully guided to claim all the provisions available to them). A dedicated team would probably be necessary to enforce these standards. (15.2)
• The IRS should also enforce rigorous privacy and security standards for software providers, holding them to the same standards it would impose on its own products — at the price of noncompliant providers losing access to IRS systems (15.3). This is especially critical as the IRS gives private providers access to ever more taxpayer data (15.3.1).
• The IRS should also regulate paid preparers, which will require statutory change. (15.4)