Modern WisdomInside DOGE, The IRS & How to Scam the US Government - Sam Corcos
CHAPTERS
Sam Corcos joins Treasury: DOGE origins and the IRS modernization “plane to land”
Sam explains how he was recruited into government via the DOGE orbit to tackle IRS modernization—an IT effort decades in the making and billions over budget. He introduces his current remit inside Treasury with a primary focus on the IRS and describes his expectations vs. what he found.
- •IRS modernization described as one of government’s worst-managed IT programs
- •Motivation: public service, national debt concerns, spouse/board encouragement
- •Brought in through DOGE in March; embedded as a Treasury employee focused on IRS
- •Initial expectations: structural change finally “on the table” under this administration
What a government CIO actually is (and why many weren’t technical)
Sam unpacks the CIO role in government, tracing it from an information “librarian” function to modern IT leadership—without updated technical qualification standards. He contrasts public-sector CIO expectations with private-sector CTO norms and outlines his own technical background.
- •CIO role evolved from record-keeping, but qualification standards didn’t evolve
- •Many past CIOs were non-technical due to lack of formal requirements
- •Difference between private-sector CTO vs. government CIO framing
- •Sam’s engineering background and coding experience as credibility for the role
Office politics in politics: the IRS IT leadership shake-up (50 on admin leave)
Sam describes a dramatic early intervention: recommending the IRS IT leadership be placed on administrative leave during peak tax season. He explains the rationale—technical incompetence upstream causes billion-dollar downstream mistakes—and what changed once technical decision-makers were installed.
- •Leadership team lacked basic technical literacy for multi-billion-dollar decisions
- •Administrative leave used as a reset: remove from roles without firing
- •Short-term panic followed by faster delivery and improved decision velocity
- •Empathy for displaced leaders: tenure-based promotions put people in wrong seats
Why executive orders don’t “self-actualize”: inertia, committees, and fax machines
The conversation shifts to the reality of implementing directives in government: policy signals don’t automatically translate into execution. Sam illustrates bureaucratic inertia with the IRS’s massive dependence on faxing, and the difficulty of locating who has authority to change entrenched practices.
- •Government incentives reward risk avoidance: committee decisions and optics
- •Example: IRS receives ~60M faxes/year and maintains ~50k active fax lines
- •Myth of fax security persists; hard to find the actual policy “lever” to change it
- •Key insight: policy existing on paper is only a small part of implementation
Civil service protections and performance management: why bad outcomes persist
Sam explains why it’s so hard to hire, evaluate, and remove poor performers in government, grounding it in the history of the spoils system and civil service reforms. He describes how performance reviews, unions, and tenure-based systems create perverse incentives—like promoting weak performers to move them along.
- •Civil service created for continuity; firing is procedurally difficult
- •Performance reviews skew high; negative reviews create major administrative burden
- •Common workaround: promote or transfer poor performers instead of removing them
- •Government engineering org ratios inverted: far more managers/administrators than engineers
Contractor bloat and vendor incentives: why government becomes a ‘soft touch’
Sam details how vendor relationships become distorted when decision-makers don’t feel personal cost for overspending. He gives concrete examples of pilot pricing that balloons, arbitrary scope changes, and contract renewals that persist even when tools aren’t used—while engineers can’t be paid market rates.
- •Incentive misalignment: vendors push lock-in; agencies often don’t negotiate hard
- •Example: $2M/year pilot proposal later renegotiated to $100M/year
- •CIO role: massive spend responsibility but low pay and heavy restrictions
- •Contract reviews revealed unused cyber vendors still being paid tens of millions annually
Procurement is broken: FAR, contested bids, and ‘value-added resellers’
Sam walks through why buying even simple software is slow and wasteful under federal procurement rules. He explains competitive bidding delays, bid contests that halt progress, and the widespread practice of using value-added resellers as a formalistic cover for “competition,” adding cost with little value.
- •RFPs disappear into procurement ‘black box’; engineers lose influence post-handoff
- •Bid contests can pause projects repeatedly—costing millions per day in delay
- •Paper digitization initiative stalls despite clear ROI due to procurement churn
- •Value-added resellers skim margins to satisfy competitive-process formalities
Waste, fraud, and abuse—plus how to scam procurement (8(a) fast lane)
The discussion turns to the blurred lines between waste, fraud, and abuse, and how weak oversight creates opportunities for graft. Sam describes license contracts that auto-renew despite never being used, and the 8(a)/Small Disadvantaged Business pathway that can be exploited to bypass competition with contracts conveniently capped under thresholds.
- •Unused software licenses renewing for years: unclear accountability and oversight
- •Investigating reveals revolving-door relationships can make ‘waste’ look like ‘fraud’
- •8(a)/SDB contracting can bypass competitive bidding; contracts often capped near $25M
- •Middlemen may take 10–50% while doing little; estimates of tens of billions in graft
DOGE beyond the meme: mission, high-agency operators, and why ‘savings’ are disputed
Sam reframes DOGE as a loose network of mission-oriented people focused on debt and dysfunction rather than a single monolithic org. He then dives into the tricky accounting of savings (fixed-price vs. BPA/IDIQ), poor government data quality, and why headline numbers are hard to validate mid-year.
- •DOGE described as high-agency, execution-focused people enforcing priorities
- •Finding the few reliable deliverers inside agencies is critical to getting anything done
- •Savings math is complicated: cancellations, auto-renewals, BPAs/IDIQs aren’t straightforward
- •IRS example: weeks to determine true budget; claims of savings depend on definitions
Why cutting spend matters—and why modernization often makes systems worse
Sam argues spending cuts are necessary not only for debt control but also because fewer vendors and fewer overlapping systems improve operational coherence. He critiques “modernization” as a flawed one-time funding mindset and explains how the IRS accumulated dozens of disconnected systems and over 100 competing ‘sources of truth.’
- •Spending less can improve performance by reducing vendor-driven fragmentation
- •IRS has dozens of non-communicating compliance systems; ‘single source of truth’ violated
- •Modernization myth: build once, declare done, then replace—creating system sprawl
- •Mainframes can be optimal for batch processing; the problem is integration and latency
Security, APIs, and the media narrative: the ‘hackathon’ that wasn’t
The conversation addresses fears that DOGE is reckless with taxpayer data. Sam describes how security is handled in practice, why ‘mega-API’ stories can be misleading, and recounts a Wired article that mischaracterized a strategic planning session as a DOGE hackathon—fueling congressional inquiries and internal confusion.
- •IRS/Treasury security posture generally strong; other agencies vary
- •Unified API effort framed as normal system interoperability, not ‘nefarious mega-API’
- •Strategic planning session misreported as ‘hackathon’; narrative snowballed into hearings
- •Media ‘trading up the chain’: insinuations become accepted “facts” over time
Hiring and talent: engineering org recomposition, HR bottlenecks, and why youth matters
Sam explains why the government struggles to recruit and retain technical talent, including salary caps, HR-driven screening, and weak technical interviewing. He outlines a push to ensure “engineering is for engineers,” rebuild hiring loops so engineers can assess candidates, and discusses how reduction-in-force rules favor tenure over performance—often pushing out younger workers.
- •IRS historically skipped technical interviews; resume keywords treated as proof of skill
- •HR controls qualification screening; engineers often see only final PDF packets of candidates
- •“Engineering is for Engineers” memo: move non-engineers out of engineering org
- •RIF mechanics are tenure/veteran-preference driven, disproportionately impacting younger staff
DOGE governance and workforce reductions: DRP, RIF, and ‘unchecked’ accusations
Sam explains how DOGE-affiliated efforts are constrained by career staff processes, and why large-scale firings are rare outside formal programs. He describes the Deferred Resignation Program uptake at IRS and the limits and brutalities of reduction-in-force processes, addressing claims that DOGE operates without oversight.
- •In Treasury/IRS, most actions still require working through career staff channels
- •Deferred Resignation Program: ~25,000 IRS staff opted in (per Sam’s estimate)
- •RIF is cumbersome and often not performance-based, creating painful tradeoffs
- •‘Unchecked’ critique vs. reality: authority exists, but levers are hard to find and use
What the IRS really does—and how tax policy becomes code
Sam demystifies the IRS’s function as primarily a software-and-ledger organization rather than a tax-collection shop, with BFS handling much of the payments flow. He then details how new tax laws are interpreted by policy/legal teams and ultimately implemented through mainframe logic, forms, and edge-case handling at massive scale.
- •IRS as ‘QuickBooks’ ledger; BFS as ‘Stripe’ payments analogy
- •Core job: reconcile reported income with third-party data and trigger enforcement workflows
- •Policy implementation pipeline: statute → interpretation → systems logic → forms and interfaces
- •Edge cases drive complexity; every change is a “problem at scale” affecting diverse taxpayers
How long it takes to fix government IT: data integrity first, leadership/culture second
Sam offers a pragmatic timeline for IRS improvement: solving data integrity within roughly three years with incremental deliverables, assuming procurement doesn’t stall. He broadens the lens to other agencies with startling examples (e.g., HHS payroll workflows) and argues sustainable fixes depend more on technical leadership, organizational design, and continuous software lifecycle practices than big-bang rewrites.
- •Priority #1: data integrity and shared truth across systems; target ~3 years for major resolution
- •Incremental improvement beats big-bang rewrites; software requires continuous care (“puppy” analogy)
- •Cross-agency example: HHS payroll involves spreadsheets, FAA mainframes, DOD processing, manual entry
- •Systemic fix: technical CIO standards, real engineering leadership, and better lifecycle discipline
