One hundred and twenty days. Not eighteen months, not "the better part of a year" โ one hundred and twenty days, from the moment a taxpayer walks in the door to the moment the case is closed. That is the number Jarod Koopman, who runs criminal investigation at the Internal Revenue Service, attached to the agency's planned rewrite of its voluntary disclosure program, speaking not from a podium with a seal on it but from a panel at the UCLA Extension Tax Controversy Conference. The current pipeline, by his own framing, runs a year or longer. The proposed one runs about four months. Call it a seventy percent compression in the single variable that matters most to anyone deciding whether to confess a filing error: how long the uncertainty lasts.
And then there is the second number, which is not a number at all. Penalties under the new program will be, in his words, "significantly lower." Significantly. Lower. No percentage. No bracket. No worked example. In a discipline where the difference between a ten percent haircut and a fifty percent haircut is the entire behavioral question, the agency has published an adjective and withheld the arithmetic.
Those two data points โ a hard deadline and a soft promise โ are the whole story, and almost everything written about this reform so far has misread which one carries the signal. The deadline is administrative hygiene. The adjective is the policy.
Context: what voluntary disclosure actually is, and why the baseline matters
For readers who have never had to think about this, the American voluntary disclosure framework is one of the few places in federal tax administration where the state openly trades severity for speed. A taxpayer who has failed to report income โ offshore accounts, unreported entities, and increasingly digital assets โ can come forward before the agency finds them, file corrected returns, pay the tax and interest, and accept a fixed penalty regime in exchange for a commitment not to refer the matter for criminal prosecution. It is not amnesty. It is a plea structure dressed in administrative language, and it has existed in recognizable form for decades, most visibly through the offshore programs that ran from 2009 until the last one closed in 2018.
The crypto population has been living in the gap ever since. The IRS clarified in 2019 that virtual currency is property for federal tax purposes, then spent the following years building a reporting apparatus โ the 1040 checkbox question, the broker reporting provisions that eventually became Form 1099-DA, the expanded information reporting tucked into successive appropriations. What it did not build was a disclosure pathway calibrated to the actual shape of crypto errors: airdrops that arrived without a cost basis, forks nobody recorded, tokens received in 2021 at a valuation that no longer exists, DeFi positions that were never clearly taxable events at all. The result is a large population of Americans holding small, ugly, hard-to-quantify mistakes, each one too expensive to litigate and too ambiguous to voluntarily resolve at full penalty.
We didn't get here because the rules were unclear. We got here because the rules were clear about the obligation and silent about the remediation.
The last offshore amnesty closed in 2018, and what followed is instructive. The agency did not stop pursuing offshore non-compliance; it stopped asking people to volunteer and started matching them against data received from foreign institutions under FATCA. The voluntary program was not the end of enforcement. It was the transitional mechanism that bought time while the data infrastructure was built. Anyone who believes the current reform is the final word on crypto tax enforcement has not studied how the previous cycle ended.
There is also a baseline problem that the announcement implies but does not state. A year-plus resolution time is not an anomaly. It is what happens when a program is designed for a few hundred high-net-worth offshore disclosures per year and then receives a population measured in the tens of thousands, each with fragmented records and contested valuations. The agency did not build for volume. Volume arrived anyway.
Core: the mechanics, and why the mechanics matter more than the announcement
Take the four components of the reform in sequence, because each one changes a different incentive.
First, the 120-day resolution target. Under the current program, a disclosure can take twelve to eighteen months to reach closure, and during that entire window the taxpayer is in limbo โ unable to file clean returns, unable to plan liquidity, unable to know whether the matter escalates. Anyone who has sat with a client through one of these knows the psychological cost is not linear in time; it compounds. Cutting the window to roughly a third does not merely save calendar days. It converts a decision most people defer indefinitely into a decision they can actually complete.
Second, the penalty reduction. Here the absence of a number is not a communications failure; it is a negotiating position. My read, based on years of watching how enforcement agencies sequence these announcements, is that "significantly lower" means the internal debate over magnitude is unsettled. The compliance wing wants a headline large enough to move behavior โ something in the forty to fifty percent reduction range relative to the standard framework. The criminal enforcement wing, which owns the program and has spent a decade arguing that crypto evasion is willful more often than not, wants a reduction small enough that disclosure never becomes cheaper than not disclosing and hoping. What got published was the compromise: a commitment to reduce, with the magnitude held back so neither side is locked in.
Third, the payment mechanics. Koopman referenced installment agreements and offers in compromise โ the latter being the formal mechanism by which a taxpayer who genuinely cannot pay the assessed liability settles for less. That detail is doing more work than it appears to. An offer in compromise is not something an agency advertises when it believes the population in question is sitting on hidden liquidity. It is what you offer when you have concluded that a meaningful share of the population is underwater. Read that against three years of crypto price action and the implication is uncomfortable but obvious: the IRS expects a substantial cohort of crypto taxpayers to have realized gains in 2021 and hold losses now, and it has decided that a partial recovery beats a theoretical full recovery that never arrives.
Fourth โ and this is the component that has received the least attention and deserves the most โ there will be no separate digital asset track. The IRS is folding crypto disclosures into the general program. A parallel legislative effort, Section 502 of the Digital Asset Taxation Certainty Act, would do the opposite: it would create a dedicated disclosure pathway specifically for taxpayers who made unintentional errors on crypto filings. That bill cleared the House Ways and Means Committee in September and now sits in the queue for a floor vote that may or may not come.
Two paths, two philosophies. The administrative path says crypto is not special; it gets the same door as everyone else. The legislative path says crypto is special; it gets a door shaped to its particular failure modes โ ambiguity of basis, absence of records, the genuine difficulty of distinguishing a mistake from a position. And here is what the "IRS softens on crypto" headlines miss entirely: the administrative path, precisely because it refuses to treat crypto as special, denies crypto taxpayers the one thing a dedicated channel would provide โ a defined safe harbor with defined edges. Being folded into the general program is not leniency. It is the withholding of clarity while extending the courtesy.
Every line of code writes a history of power. The Internal Revenue Code is code, and the choice of which function to call โ the general disclosure routine or a purpose-built one โ determines who gets processed and who gets protected.
One more mechanical detail that reframes the urgency. The agency's civil assessment window for a substantial omission of income runs six years, three for ordinary understatement, and indefinitely where fraud is alleged. A 120-day resolution target is therefore not a race against the statute. It is a race against the agency's own capacity. That distinction matters because it tells you the reform is driven by internal throughput constraints, not by any legal deadline forcing the agency's hand.
The incentive structure, modeled honestly
Strip the announcement to its economics and what you have is a repricing of three variables: time cost, penalty cost, and payment flexibility. All three move in the taxpayer's favor. That is the surface reading, and it is correct as far as it goes.
The second-order effect runs the other way. If penalties are about to fall significantly, the rational move for a taxpayer sitting on an unresolved error today is not to disclose today. It is to wait, because the price of disclosure is a falling asset and the option to disclose remains exercisable. Announce a discount that has not yet taken effect and you do not accelerate disclosures โ you suppress them until the discount lands. This is not speculation; it is the most basic property of a forward-looking incentive. Every compliance professional I have spoken with since these remarks circulated has described the same client conversation: can we wait until December?
Which means the agency has, deliberately or not, created a disclosure vacuum in the interval between announcement and implementation. Taxpayers who might have come forward in October now have a reason to file nothing and see. The tax and interest on those liabilities keep accruing while they wait, so the delay is not free, but the penalty delta is likely to dominate for anyone with material exposure.
There is a countervailing force, and it is the one that ultimately keeps the program honest: the asymmetry between voluntary disclosure and being found. A taxpayer who waits and is subsequently identified faces the standard penalty framework plus the live possibility of criminal referral. A taxpayer who waits and then discloses still gets the discount. The waiting strategy only fails for those already on the agency's radar โ which, given the data the agency is about to receive, is a larger set every quarter.
Contrarian: this is inventory management, not mercy
Here is where I part company with most of the coverage.
The consensus reading is that the IRS is softening. That reading is wrong in a specific and important way. An agency that believed crypto tax evasion were rare, or difficult to prove, or confined to a small population, would not bother rewriting its intake process. It would keep the slow pipeline and let cases die of old age. The decision to compress resolution to 120 days and cut penalties is the decision of an organization with more inventory than throughput. The 120-day target is not a gift to taxpayers. It is a throughput target set by someone who has looked at the backlog and concluded that the constraint is not the number of cases but the number of closable cases per unit time.
Under that reading, "significantly lower" penalties stop looking like mercy and start looking like yield optimization. If your objective is deterrence, you maximize the penalty and accept that most cases never close. If your objective is recovery, you minimize the penalty to the level where disclosure becomes rational and the case actually resolves. The mention of offers in compromise tells you which objective won.
This is the same structural error the tokenization industry has been making for three years, and it is worth naming because the pattern is identical. Everyone assumed traditional institutions wanted to come onto public rails because public rails were philosophically superior. They didn't. They wanted settlement finality and an audit trail, and they were perfectly willing to build permissioned corridors to get them when the public ones came with obligations they could not control. The IRS is making the same calculation in the opposite direction. It does not need a crypto-native disclosure channel to get what it wants. It needs your exchange's 1099-DA, and it needs a disclosure program efficient enough that the people the data flags have somewhere to go other than litigation.
Governance isn't the ceremony of consultation. It is the shape of the pipe through which cases actually flow, and that shape tells you what the operator is optimizing for. This pipe is being widened at the intake and narrowed at the exit. That is not a description of leniency. It is a description of a processing plant.
And the number itself will decide whether the plant runs. A ten percent reduction changes nothing โ the cost of a lawyer still exceeds the penalty saved, and behavior stays frozen. A fifty percent reduction with a four-month clock is a genuine product, one that a competent advisor would recommend. The agency has announced the product without the price, which means it has announced an intention, not a program.
The procedural risk nobody is pricing
The IRS can adjust a voluntary disclosure program through its existing administrative discretion. It does not need Congress to do it. That is precisely why a December timeline is plausible on the administrative side. But administrative discretion is not unbounded. Substantive rulemaking in the United States generally requires notice-and-comment under the Administrative Procedure Act, and nothing in the reporting so far suggests the agency has committed to running that process. If it skips it, the resulting program is exposed to challenge โ not by taxpayers who benefit from it, but by anyone with an interest in arguing the agency exceeded its procedural authority. That is a low-probability, medium-impact risk, and it is the kind of thing that quietly kills programs six months after launch.
The legislative path carries its own timing risk. Section 502 has cleared committee. Committee clearance is roughly the point in the American legislative process where a bill has survived the easy part. Floor scheduling, Senate consideration, and signature remain, and none of them sit on a December timetable. The realistic outcome is that the administrative program launches first, in whatever form survives internal review, and the legislative safe harbor arrives later or not at all โ leaving crypto taxpayers with the general program and no dedicated channel for as long as it takes Congress to act, which is to say indefinitely.
And there is the sequencing question I keep returning to. Broker reporting under the digital asset information regime begins generating data on a rolling basis, and coverage expands with each reporting year. The agency is building the intake valve in the same window the data firehose turns on. That ordering is not accidental. Historically, voluntary disclosure programs have been refreshed immediately before enforcement capacity expands, not after. The door gets a fresh coat of paint right before the letters go out.
What actually changes for a crypto holder, and what does not
If you are a US taxpayer with unresolved crypto errors and no indication the agency has identified you, the reform does not change your legal position today. It changes your expected cost of resolving it, and it does so in a direction that rewards patience and punishes urgency. The rational response is not to sprint toward a program whose rules have not been published. It is to document everything โ acquisition records, cost basis reconstruction, the lot-level history most people never kept โ so that when the rules land, you can move in days rather than months. The 120-day clock only helps people who arrive with their evidence already assembled.
If you are an exchange or custodian, the reform is unambiguously positive in one narrow sense and negative in a broader one. Positive, because a clearer disclosure path reduces the tail risk of your user base being swept into enforcement actions that generate headlines you do not want. Negative, because efficiency gains on the agency's side translate directly into higher expectations on yours โ better reporting, faster 1099-DA generation, cleaner reconciliation. The compliance burden does not fall. It redistributes.
If you are an offshore or lightly regulated venue, the reform is straightforwardly bad news. Every efficiency gain in case processing is a reduction in the expected value of non-compliance, and the population that benefits most from a slow agency is precisely the population arbitraging its slowness.
And if you are reading this as a market signal, you are reading it wrong. This does not move Bitcoin. It will not move Ethereum. It is not a supply event, a demand event, or a liquidity event. The only assets it touches are those whose value derives from regulatory clarity itself โ the compliance software vendors, the tax advisory practices, the audit and analytics firms. That is a real business, and it is about to get busier, but it is not a trade.
Takeaway
The signal in this reform is not that the IRS is going easy on crypto. The signal is that the agency has concluded its crypto caseload is large enough, and its crypto data good enough, that the binding constraint has shifted from finding violations to processing them. Everything else โ the 120 days, the lower penalties, the installment agreements, the refusal to build a dedicated channel โ follows from that single operational judgment.
What to watch, in order of information value. The formal rule text, whenever it appears, because the penalty percentage in that document is the entire behavioral question and the adjective is worth nothing. Whether the agency runs a notice-and-comment process, because that determines whether the program survives its first legal challenge. Whether the December date holds, because a slip past the filing season changes the calculus for everyone planning around it. And whether Section 502 reaches the floor, because a dedicated crypto disclosure channel would be a materially different animal from a general one โ narrower, safer, and far more useful to the people who actually need it.
Truth emerges from transparency, not from silence. An agency that announces a discount without announcing its size has told the market the number is still being fought over inside the building. That is not a policy. It is a position, and positions get repriced.
The question worth sitting with is not whether the IRS is being generous. It is why an enforcement body would choose to optimize for throughput at the exact moment its data collection capacity is expanding. The answer will be visible within a year, in the shape of the first batch of cases processed under the new clock โ and in who, exactly, is standing in that line.