Key Takeaways
Make technologists partners in business outcomes, not an internal help desk.
The old divide persists. Schwartz traces the separation of business and IT to the early computer era, when technical specialists seemed like outsiders speaking an alien language. Companies responded by treating IT as a contractor: business units wrote requirements, IT estimated and delivered, and both sides tried to manage risk at arm’s length. The arrangement made accountability look clear, but it often rewarded delivery of specifications rather than useful results.
A young programmer called Gerald seemed distracted by obscure technical work while payroll systems failed. The author later realized Gerald’s work was intended to remove a bottleneck that threatened payroll. The apparent conflict reflected a deeper problem: business leaders could not see the value of technical judgment, and IT could avoid responsibility for business outcomes.
The argument is less about abolishing professional boundaries than aligning incentives across them. Research on organizational silos repeatedly finds that local targets can undermine system-wide performance: a department can meet its metric while the company loses time, customers, or adaptability. Shared outcomes create a reason for technical and business experts to exchange knowledge early. Still, integration needs clear decision rights. If everyone is responsible in the abstract, no one may be accountable in practice. The useful shift is from internal customer service to joint ownership, with technical expertise respected rather than subordinated.
In uncertain systems, set direction and let informed teams act locally.
Plans lose freshness quickly. At Borodino, Napoleon issued battlefield orders from a distant position. By the time messages reached him or returned to commanders, conditions had changed, and some orders were based on events that were no longer true. Schwartz uses the episode to illuminate a modern organization: information moves slowly through layers, while customers, competitors, and technology keep changing.
A complex adaptive system is a group of people or units pursuing their own aims while interacting in ways that produce unpredictable results. Leaders cannot know every detail or command every response. Their better role is to clarify the desired outcome, establish shared principles, and give capable teams authority to respond to what they see directly.
This resembles the distinction between centralized intent and decentralized execution in military doctrine and emergency response. Local actors often have better, more current information than distant decision makers. The idea is especially powerful when feedback is fast and consequences are observable. It is not a license for every team to optimize its own corner: teams need common objectives, information sharing, and boundaries. Without those, decentralization can reproduce the same silos in smaller units. The leadership challenge is to create coherence without pretending that the center can know everything.
Ship tiny changes frequently to improve speed, quality, and recovery.
Small releases reduce exposure. Agile development breaks work into usable pieces, while Lean practices reduce waiting, handoffs, and unfinished work. DevOps combines these ideas with cross-functional teams and automation, so development, testing, security, and operations can deliver changes together. Instead of betting on one large launch, teams release small increments, learn from use, and correct problems quickly.
A DORA study cited in the book compared high-performing organizations with low performers. The top group deployed 46 times more frequently, moved code from completion to deployment 2,555 times faster, recovered from failures 2,604 times faster, and had one-seventh the change failure rate. These are reported comparisons, not a promise that every organization will reproduce the same numbers.
The counterintuitive point is that frequent change can strengthen stability when each change is small, tested, and reversible. This fits reliability engineering, where systems are made safer by rehearsing routine recovery rather than relying on rare, dramatic interventions. The metrics also expose a useful distinction between activity and flow: more work started is not progress if it waits in queues. DORA's findings are compelling, but observational comparisons cannot isolate every cause. Leaders should track operational outcomes and customer impact, not treat deployment frequency as a performance target that can be gamed.
Treat every requirement as a testable guess about what users need.
Specifications conceal assumptions. A request for a particular screen or data field is really a belief that the feature will help someone achieve a business result. Schwartz recommends making that belief explicit, then testing it cheaply before investing in a full solution. This is the Lean Startup approach: build a minimal experiment, observe behavior, and use the evidence to revise or discard the idea.
At USCIS, a team designed digital forms meant to speed immigration adjudication. After a long build, processing took 75 minutes instead of the 15 minutes required with paper. Adjudicators had relied on quickly flipping through documents and validating every field. A mock-up observed with real users could have exposed these assumptions before the system was built.
This is a practical application of scientific reasoning: distinguish a goal from a proposed mechanism, then test the mechanism. Observing real behavior is often more informative than asking people what they prefer, since users may not anticipate how a new workflow changes their work. The discipline also requires choosing meaningful measures. A click rate, for example, may show interest without demonstrating improved service. Experiments work best when they connect user behavior to the outcome the organization actually cares about, and when teams can act on inconvenient results.
Define IT value by strategic outcomes, including the agility it preserves.
There is no universal ROI. Return-on-investment calculations depend on forecasts that are often unreliable in changing markets, and they struggle to compare unlike outcomes, such as improved security, increased revenue, or a public agency's mission. Schwartz argues that business value means what the organization values, so senior leaders should set clear objectives and use them to guide technology choices.
He also asks leaders to look beyond current features. Technical quality, organizational capacity, and access to useful data all affect how quickly the company can respond to new needs. Poorly structured systems accumulate technical debt, making future changes slower and more expensive. Agile systems and processes have latent value: they preserve options to adapt, even when the next use is not yet known.
This expands capital budgeting beyond forecasted cash flows toward option value: the value of keeping a future choice available. In finance, an option becomes more useful when uncertainty rises and the cost of waiting for information is high. That logic supports investment in maintainable systems, reusable components, and accessible data. But agility should not become a vague justification for any modernization spend. Leaders still need evidence that flexibility matters to their strategy, and should compare its cost with the specific constraints or opportunities it addresses.
Manage risk by reducing the cost of change, not freezing it.
The status quo carries risk. Leaders often fear an unfamiliar technology or process more than a familiar but failing arrangement. Schwartz connects this tendency to status quo bias, a documented preference for existing conditions. Yet a fixed plan can be dangerous when markets and threats change faster than the plan can be updated.
The alternative is not reckless risk-taking. It is to make smaller, reversible decisions, test assumptions, and increase investment when evidence improves. A small experiment can buy information while limiting the amount at stake. Automated guardrails can check security, compliance, or spending continuously, allowing teams to move quickly within defined limits. In this view, experimentation is a way to reduce risk, not to celebrate failure.
The framing echoes real-options reasoning and the distinction between uncertainty and danger. A cautious decision can still be hazardous if it locks an organization into an obsolete position. Small tests preserve the ability to stop, redirect, or expand based on new evidence. The important qualification is that reversibility varies: a temporary interface test is easy to undo, while a change to safety-critical infrastructure may not be. Risk management should therefore consider both the scale of a bet and the cost of reversing it, rather than applying one appetite for experimentation everywhere.
Fund objectives in stages instead of approving giant projects upfront.
Governance should learn as work proceeds. Traditional investment committees often approve a large bundle of requirements and then monitor whether the team follows the original plan. Schwartz calls this a Star Chamber approach: a small group judges proposals before delivery, when uncertainty is greatest. Staged funding lets leaders make smaller commitments, inspect results, and decide what deserves continued investment.
At USCIS, an E-Verify modernization effort spent four years producing requirements and binders. The team replaced that plan with five business objectives, funded work in short increments, and reviewed measures frequently. Teams included both technologists and operational staff. One objective plateaued, so its remaining funds moved elsewhere. The broader effort ended after two and a half years because the goals had been met, not because the original project scope was exhausted.
This approach turns governance from a gate that approves predictions into a learning loop that reallocates resources. It resembles venture investing, where staged commitments preserve the right to invest more after uncertainty falls. It also makes stopping a successful effort rational: once its marginal value declines, capacity can move to better opportunities. The challenge is choosing objectives that are specific enough to guide teams without prescribing their solutions. Some outcomes, such as security or public trust, resist simple metrics, so oversight must assess credible evidence as well as dashboards.
Cut bureaucratic waste, then automate guardrails that keep work safe.
Rules can preserve knowledge. Schwartz does not treat bureaucracy as inherently harmful. Rules can record lessons, support fairness, and help an organization meet its obligations. The damage comes when procedures are slow, duplicative, or hard to update. At the Department of Homeland Security, an analysis document could take about 18 months to prepare, despite often formalizing a decision already made. The question is whether that effort produces enough value to justify its delay and cost.
Security is also an enterprise responsibility, not a task to hand off to IT. Good hygiene, automated tests, and continuous controls can catch common vulnerabilities early and leave an audit trail. The goal is to make protection part of ordinary work, rather than a final inspection that stops delivery.
This is a useful distinction between a rule's purpose and its paperwork. In safety engineering, a checklist or automated interlock is valuable when it reliably prevents a known failure mode; redundant documentation that proves a box was considered may add little protection. Automation can make controls consistent, but it can also encode outdated assumptions at scale. Rules therefore need review as well as enforcement. Security culture matters because technical defenses cannot compensate for careless handling of information or a workforce that assumes protection belongs to someone else.
Use cloud capacity as a variable resource for faster, cheaper experiments.
Infrastructure becomes adjustable. Cloud computing lets organizations obtain processing, storage, and other services when needed, then reduce or release them when demand falls. This replaces the older pattern of buying enough hardware for expected peak usage, even if much of it sits idle. It also lowers the initial cost of trying an idea and makes it easier to scale a successful one.
The book gives Intuit's tax software as an example: by turning off infrastructure outside tax season, it could save five-sixths of the relevant service's infrastructure costs. Cloud capacity can also be provisioned quickly, so a team can test a product without first building a datacenter. The cloud does not remove market uncertainty; it gives the company more ways to respond to it.
The economic shift is from fixed capacity to a more variable cost structure. That can improve both experimentation and resilience, especially when demand is seasonal or difficult to predict. Yet variable spending can surprise organizations when a service becomes popular, and dependence on a provider introduces new operational and strategic questions. Cost limits, monitoring, and portability choices matter. The strongest case for cloud adoption is not simply that someone else owns the hardware, but that on-demand capacity changes the speed and reversibility of business decisions.
Make the executive suite jointly accountable for cross-functional outcomes.
Digital work crosses executive silos. A customer-facing service may depend on marketing insight, technology, operations, finance, and security. Schwartz argues that the CEO, CFO, CMO, CIO, and board should act more like a team with shared responsibility for enterprise outcomes, rather than separate leaders negotiating across functional boundaries. The CIO contributes technical judgment to strategy; finance steers resources; marketing brings customer signals; operations reveals how work functions in practice.
Transformation should begin with action, not an elaborate plan. Start with a clear objective, give a small cross-functional team the authority and tools to pursue it, and look for a real result within roughly two weeks. Then examine what happened, remove obstacles, and choose the next small step.
Executive collaboration matters because the quality of digital decisions depends on combining knowledge that no single function holds. This resembles interdisciplinary team research: diverse expertise can improve problem solving, but only when members share information rather than defend territory. Joint accountability creates that incentive. A short first result also makes a transformation credible through observable change, not slogans. The caution is that speed cannot substitute for strategic coherence. Teams need a clear purpose, and leaders must still protect long-term capabilities that produce little immediate evidence.
Analysis
Schwartz's central move is to redefine digital transformation as a change in organizational relationships, not a technology installation. Cloud services and modern software practices matter, but their value is lost when IT remains a contractor receiving fixed specifications from a separate business. The book's recurring contrast is between prediction and adaptation: when requirements, schedules, and returns are uncertain, leaders should create conditions for frequent learning instead of demanding precision that does not exist. The strongest synthesis is economic. Small deliveries, staged funding, automated controls, and maintainable systems are not merely engineering preferences. Together they reduce the amount of capital exposed before evidence arrives, shorten the time to value, and preserve choices when assumptions fail. Agility is therefore both an operating capability and a form of risk management. The E-Verify example makes the point concrete: replacing years of documentation with measurable objectives allowed leaders to adjust spending and stop work when further gains were unlikely.
The book's weakness is also the source of its force. It argues broadly against contractor-style governance, but not every system can be divided into tiny releases or judged through near-term measures. Highly regulated, safety-critical, or tightly coupled systems may require extensive upfront analysis, and cloud flexibility can create provider dependence or cost volatility. The answer is not to restore rigid planning everywhere, but to vary controls according to the consequences of failure and the reversibility of decisions.
The most durable lesson is that accountability should attach to outcomes, while the path toward those outcomes remains open to revision. That requires executives to tolerate evidence that disproves their preferred solutions, technical teams to understand business purpose, and governance to move from policing plans to supporting informed adjustment. Transformation succeeds when these behaviors become normal operating practice, not when a company simply adopts the vocabulary of Agile or DevOps.
Review Summary
War and Peace and IT receives mostly positive reviews, with readers praising its insights on digital transformation, agile methodologies, and the role of IT in business. Many find it well-written and informative, appreciating the author's experience and perspective. Some criticize the book for repetition and AWS bias. Readers value its exploration of IT's importance in determining business objectives and fostering collaboration. The book is recommended for executives and managers seeking to understand IT's role in driving business results, though some found it challenging to grasp certain concepts.
People Also Read
Glossary
contractor-control model
IT treated as a vendorA way of organizing work in which business units specify technology needs, submit requirements to IT, and expect IT to deliver against an agreed schedule and scope. Schwartz argues that this model creates an arms-length relationship inside one company, separates technical work from business outcomes, and rewards compliance with the specification over the value actually produced.
Complex adaptive system
Interacting, self-organizing actorsA system made up of people or groups that pursue their own aims and interact in ways that create changing, difficult-to-predict outcomes. In the book, an enterprise is complex adaptive because information is incomplete, actions affect other groups, and small changes can have unexpectedly large effects. Leaders guide such a system through shared purpose and conditions for action, not detailed commands.
DevOps
Integrated delivery and operationsA set of technical and organizational practices that brings software development, testing, infrastructure, operations, and security work into a coordinated flow. Small cross-functional teams use automation and frequent releases to reduce handoffs, detect defects earlier, and respond quickly to users or incidents. Schwartz presents DevOps as a way to improve both delivery speed and operational control.
IT asset
Technology enabling business workSchwartz's managerial view of an organization's overall technology capabilities, including systems and infrastructure, considered as an asset that supports current operations and future change. Its value is not limited to what it does today. The design, flexibility, and reliability of its internals affect the time, cost, and risk of adapting it to new business needs.
Technical debt
Cost of hard-to-change systemsThe accumulated burden created when software's internal design becomes tangled, outdated, poorly tested, or difficult to understand. The system may still perform its current functions, but changes become slower and riskier, and defects are harder to find. Technical debt is like financial debt because it imposes continuing costs and can eventually restrict an organization's ability to adapt.
Functional debt
Gap between systems and needsThe mismatch that develops when a technology system no longer fits the business activities or user needs it is supposed to support. Users may compensate with workarounds or alter their own processes to fit the system. Those adaptations consume time and can prevent the organization from taking advantage of new opportunities.
Impact mapping
Goals linked to behavior and workA collaborative planning technique that begins with a business goal and maps the people whose behavior could affect it, the behavior changes that would support the goal, and the deliverables that might encourage those changes. It helps teams distinguish the outcome they seek from a proposed product or feature, and gives them hypotheses to test.
Metered funding
Investment released in stagesA funding approach in which an organization commits resources incrementally rather than approving the full cost of a large initiative upfront. Each stage should produce useful results or information that can guide the next decision. Leaders can continue, redirect, or stop investment as uncertainty falls, limiting exposure while retaining the option to pursue promising work.
T-shaped people
Broad skills, deep specialtyPeople with working knowledge across several areas and deep expertise in one. The broad skills help teammates communicate, shift work, and understand adjacent concerns; the specialty provides depth when a problem requires expert judgment. Schwartz favors T-shaped teams because they reduce dependence on handoffs between narrow functional silos.
software delivery and operational performance (SDO)
Combined delivery and reliability measureA performance construct used in the DevOps Research and Assessment research discussed in the book. It combines measures of software delivery throughput and operational stability, including deployment frequency, lead time from completed code to deployment, change failure rate, and mean time to repair. Schwartz uses the findings to argue that speed and stability can improve together.
Download PDF
Download EPUB
.epub digital book format is ideal for reading ebooks on phones, tablets, and e-readers.