Sustaining a Shared Reality: How Past Technology Waves Have Impacted Strategy

14 min read Original article ↗

Whitney Zimmerman

Press enter or click to view image in full size

John Ruskin, Doge’s Palace, Venice: 36th Capital (1849–1852), The Ruskin Library, Lancaster University

In 1972, a thirty-year-old McKinsey partner named Lou Gerstner published an article in Business Horizons titled “Can Strategic Planning Pay Off?“ It was a reaction to a great wave of information technology in business — the mainframe data-processing revolution of the 1950s and 1960s — and the effect it had on how companies approached strategy. That wave had taken American business from roughly 240 installed computers in 1955 to about 50,000 worldwide a decade later, anchored by IBM’s System/360, which became the back-office platform of the modern corporation, running payroll, billing, accounting, and inventory for a generation. Two decades later, Gerstner encountered the consequences of a subsequent wave firsthand, when he arrived at IBM as CEO in 1993 and found a company drowning in analysis but, in his words, “paralyzed, unable to act on any predictions.”

Major advances in information technology have more than once made it dramatically easier to produce raw material for strategy. And each time, some companies have misused that potential — more analysis, not better decisions. Generative AI is another, powerful wave of change. Understanding past waves is one of the most useful things strategists can do to make the most of the latest. It is, in Francis Gavin’s terms, thinking historically about the present moment.

Gerstner’s recurring experiences

To write his 1972 article, Gerstner surveyed chief executives about the new discipline of strategic planning that had swept through corporate America, fueled in part by the data-processing capabilities of the previous decade. He recorded their candid reactions. One called it “basically just a plaything of staff.” Another compared it to “a Chinese dinner: I feel full when I get it, but after a little while I wonder whether I’ve eaten at all.” A third was blunter: “A staggering waste of time and money.”

In Gerstner’s view, the problem was the disconnection of analysis from decision-making, and that analytical machinery had given companies a false sense of reality. He drew an explicit analogy to the data-processing investments that had enabled the planning boom in the first place: “Following the widespread introduction of data processing in the 1950s, many companies sooner or later were obliged to recognize that the promise of this great management tool was stubbornly refusing to materialize. Real, tangible return on investment was low or nonexistent.” His prescription: the work must be integrated with the real decisions, every project must pass the “so what?” test, and the CEO must be personally involved. Without those, the technology’s promise would remain unrealized, and companies would strategically struggle.

Even in 1972, Gerstner saw the limits of automation in strategy. Strategic planning, he wrote, “is fundamentally a creative process. It cannot be programmed or systematized. To structure meaningful, practical action programs requires insight, wisdom, and perspective.” The hardest part of strategy — the judgment, the willingness to take a personal stand on a controversial issue — could not be outsourced.

Two decades later, Gerstner experienced the dynamic first-hand. A subsequent IT revolution began in 1979 with two products that shipped that year: VisiCalc, the first electronic spreadsheet, and Oracle, the first commercial SQL relational database. By the mid-1990s, those trends — a spreadsheet on every analyst’s desk and a relational database under every large company’s transactions — together with the PC, enterprise software like SAP R/3, and a growing ecosystem of business intelligence tools, were transforming corporate analytics. IBM was one of the companies that had built the infrastructure that made it possible. IBM had sold the tools and used them itself.

What Gerstner found when he arrived at IBM in 1993, as he describes in his memoir Who Says Elephants Can’t Dance?, was a company that had mastered the production of analysis and lost the capacity to act on it. The apparatus had become elaborate, expensive, and in important places systematically misleading. The clearest example was customer satisfaction data: 339 separate surveys, administered by the sales force, which naturally selected its happiest customers — what Gerstner called “asking the innkeeper how good the inn is.”

His response was to rebuild IBM’s view of the world, with leaders firmly in the loop. He believed “good strategies start with massive amounts of quantitative analysis — hard, difficult analysis that is blended with wisdom, insight, and risk taking.” One of his earliest moves was to recruit senior financial executives charged with instilling “productivity, discipline, and probing analysis” into the company. He launched Operation Bear Hug, requiring each of his top 50 executives to visit 5 major customers over 3 months and report back personally. He rebuilt the customer satisfaction approach as fourteen studies, independently administered, sampling both customers and non-customers, benchmarked against competitors, and integrated into strategic planning on a semiweekly basis. This work was essential to IBM’s strategic transformation. The decisions that followed — keeping IBM together rather than breaking it up, slashing mainframe prices to save the platform, and building services into the company’s largest business — relied on a leadership team that shared a view of the world they helped build.

Gerstner was not the only one to see this recurring dynamic. Henry Mintzberg published The Rise and Fall of Strategic Planning in 1994, a sweeping indictment of the same era of corporate planning practice that Gerstner inherited at IBM. Mintzberg did not center his account on analytical technology — his target was the institution of strategic planning itself — but the dynamic he documented at the heart of the book is recognizable as the same one. He called it the Fallacy of Detachment: the belief that strategy can be formulated by people who are removed from operational reality and informed by hard data alone. Mintzberg traced the same pattern playing out at General Electric in the same era, which had built an elaborate hierarchy of Strategic Business Units and Sectors, enabling the CEO to review the company through six small strategy binders, rather than through detailed engagement. Mintzberg’s verdict: “the more aggregated became the information, the more detached became the CEO.” The corrective at GE, as at IBM, was a new CEO — Jack Welch — who insisted on reconnecting senior executives to reality.

Mintzberg traced this dynamic well before the era of corporate planning. His conclusion is one worth keeping in mind: “Systems do not think, and when they are used for more than the facilitation of human thinking, they can prevent thinking.”

What explains the pattern

Disengagement

When technology takes over the routine parts of a task, humans can disengage not only from that step but from adjacent ones. Lisanne Bainbridge identified this in 1983 in a landmark paper called “Ironies of Automation.” Her central observation is that automating the routine work leaves the human responsible for the hardest parts — the exceptions, the novel situations, the judgment calls — while depriving them of the practice through which the skill to handle those situations is built. The operator needs to be “more rather than less skilled, and less rather than more loaded, than average” precisely when things go wrong. Yet automation systematically erodes both the skill and the readiness.

Parasuraman and Manzey, in a 2010 integrative review, confirmed that this “automation complacency” is attentional, not motivational — it cannot be willed away or overcome with practice. It generalizes from novice to expert operators. The same can happen in strategy work. When a technology too exclusively handles some work — market sizing, scenario modeling, competitive benchmarking — leaders may inadvertently and unintentionally disengage from the framing, debating, and choosing that the analysis is supposed to serve. The quality of the analytical output may rise, but the quality of broader engagement may fall, undermining the end result.

Fragmented mental models

Teams that build their understanding of a problem together — through debate, collaborative analysis, and constructive conflict — develop shared mental models. A substantial body of research (Van den Bossche et al. 2011; DeChurch and Mesmer-Magnus 2010, a meta-analysis of 65 studies) shows that these shared models are what enable teams to coordinate, adapt, and make creative and hard decisions under uncertainty. Critically, the process of co-construction matters as much as or more than the content of the output. Teams that receive a finished analysis, however perfect, do not build the same cognitive alignment and thus gain the same potential as teams that work through the problem themselves.

An experiment by Liang, Moreland, and Argote (1995) tested this. Teams trained together on a task outperformed teams whose members received identical training individually — and the mechanism was transactive memory (knowing who knows what), not cohesion. The implication for strategy is that a leadership team jointly working through a market analysis, even imperfectly, builds a cognitive infrastructure that a team receiving an automated report does not. The output may be superior, but the team’s ability to debate, challenge, and adapt it is weaker.

Gerstner’s IBM is a case study in this erosion. The analytical apparatus had become so elaborate that leadership teams consumed analysis rather than building shared understanding. Operation Bear Hug was an act of forced co-construction: executives had to go see reality for themselves, synthesize it in their own words, and report what they had learned.

Mismatched paces of change

Stewart Brand observed in his 1999 essay on pace layering that healthy systems contain layers operating at different speeds, and that their resilience depends on the relationship between them. Fast layers learn and propose; slow layers remember and dispose. “Fast gets all our attention,” Brand wrote. “Slow has all the power.” The system breaks down when the fast is allowed to drive the slow — when analysis sets the pace of decision, or when the appetite for synthesis outruns the team’s capacity to debate it. Resilience depends on the slower layer constraining the faster, not the reverse.

Strategy is a layered system in this sense. Analysis and data gathering are naturally fast; framing the challenge, building shared reality, debating choices, and developing the personal judgment to commit are slow. The fast layers can produce, but the slow layers govern and commit. Past information technologies have temporarily disturbed this balance by accelerating the analytical layer. Similarly, AI accelerates it, and others like synthesis, by orders of magnitude, while the slow layers remain human-paced. The fast layer races ahead; the slow layer no longer has time to constrain or govern what the fast one produces. What looks like a more capable strategy process risks becoming structurally less capable: the layer that does the strategic work has been outpaced by the layer that does the analytical work.

Recommendations for strategy leaders

None of this is an argument against using AI in strategy. It is an argument for learning from history to best harness its potential for strategic effectiveness.

Stay engaged in the work. The deepest engagement with strategy happens through working sessions, not the consumption of finished decks. CEO Jensen Huang has made this a hallmark of NVIDIA’s leadership culture: senior teams debate strategy together at the whiteboard, in real time, with the substance in front of them. Anthropic CFO Krishna Rao reflected a similar belief recently: “I tell people during the interview process, ‘I’m not really hiring you as a direct report of mine. I’m hiring you as a partner, and I want you to treat it as a partnership, which means there might be things that you and I disagree on. I want to hear that, and I want to whiteboard it. I want to understand.’”

Don’t outpace your team’s capacity to absorb the change. New technology is not the problem. Treating its arrival as if the organization were already ready is. Leaders and processes need time to experiment with new tools, build new habits, and invent new approaches. Automating the production of strategy artifacts before the team has built the discipline to interrogate them produces a worse process, not a better one.

Invest in quality of thought, not volume of work. AI makes it cheap to produce more of everything — more market sizings, more scenarios, more competitor profiles. Strategy is bounded by the quality of the framing, not the volume of the inputs. Olivier Sibony‘s work on strategic problem-solving emphasizes that the hardest and most valuable part of strategy is decomposing the challenge correctly — getting the constituent parts right before combining them. Spend disproportionately on framing, debating choices, and building foundational understanding of market attractiveness, competitive advantage, and how value is created.

Look for the marks of real thought. Strategy that has been grappled with looks different from strategy that has been produced. It contains the irregularities Ruskin saw in Gothic capitals — evidence of someone deciding, interpreting, choosing. The polished, symmetrical, machine-perfect output is precisely what should make a strategy leader skeptical. The work of strategy leaders, increasingly, is to recognize and reward the harder-won thinking that sometimes does not look as polished as what AI might produce.

A note on the artwork

Press enter or click to view image in full size

John Ruskin, Doge’s Palace, Venice: 36th Capital (1849–1852), The Ruskin Library, Lancaster University

The image accompanying this post is a drawing by John Ruskin — the Victorian-era polymath whose work shaped how the nineteenth century understood the relationship between craft, technology, and human dignity. In the mid 1800s, Ruskin studied the Doge’s Palace in Venice stone by stone, drawing all 36 carved capitals — the top of a column — along its lower arcade. Each was carved by a different craftsman. Each depicts a different subject. No two are alike. The 36th, at the corner of the palace where the Great Council of Venice met, places Justice enthroned alongside seven of history’s great lawgivers — Aristotle teaching his pupils, Solon, Moses, Trajan delivering judgment to the Widow, and others. Directly above it, on the corner of the palace itself, stands the great angle sculpture of the Judgment of Solomon. Venice had placed judgment quite literally at the foundation of its seat of government.

The study became The Stones of Venice (1851–53). At the heart of the three volumes is a chapter called “The Nature of Gothic,” which has been called one of the most important texts of the nineteenth century. Writing as factories were changing the nature of work itself, Ruskin argued that the value of the carved capitals lay precisely in their imperfection: each chisel mark recorded a person thinking, interpreting, deciding. His deeper claim was that thought and labor cannot be separated without loss — that the gentleman who only thinks and the operative who only executes are both diminished by the division. “The painter should grind his own colors,” he wrote, “the architect work in the mason’s yard with his men; the master-manufacturer be himself a more skilful operative than any man in his mills.” Only in the union of thinking and doing, Ruskin argued, can either be fully alive.

Ruskin saw the same dynamic that information technology revolutions have reproduced — analytical work delegated from the executives whose judgment it must serve, and which ultimately rely on their involvement to be effective. What Gerstner found at IBM was analytically elaborate but strategically lifeless, and in important places systematically misleading. What he did about it — Operation Bear Hug, the rebuilt customer satisfaction apparatus, the personal engagement of senior leaders with the work — was, in Ruskin’s terms, taking up the grinding of his own colors. Gerstner’s belief, in 1972 and 1993, was that strategy work must be in service of decisions, guided by leaders. Ruskin’s counsel, written a century earlier, said the same of craft: “Never demand an exact finish for its own sake, but only for some practical or noble end.”

To go deeper