Berkshire Hathaway: The 27% Baseline Problem
Twenty-Seven Percent Is a Ghost That Haunts the Structure
For twenty-two years, from 1960 through 1982, the partnership that became Berkshire Hathaway compounded capital at twenty-seven percent annually. That number, twenty-seven percent, functions now less as a historical fact than as a structural baseline against which every subsequent decision must be measured. It represents not merely a return profile but a set of market conditions: information asymmetries available to a small pool of capital, mispricings in mid-cap industrials and regional insurers that could absorb meaningful investment, and the absence of institutional players sophisticated enough to arbitrage away obvious discounts. When the architect of that record acknowledges, explicitly, that such performance is “not going to be the case this time,” he is not issuing a disclaimer. He is describing a physical constraint. The machine has grown too large for the terrain that nourished it.
This is the defining structural tension. Berkshire Hathaway generates economic returns by deploying massive capital, checks exceeding one billion dollars, into durable consumer franchises and financial assets possessing structural moats: brand equity dense enough to survive managerial incompetence, regulatory positioning that locks in float, operational excellence in low-tech, high-margin niches. The compounding engine relies on insurance operations that collect premiums years before claims come due, creating a self-renewing pool of permanent, interest-free capital. Yet as the asset base swells past one hundred billion dollars, the universe of investable opportunities capable of moving the needle shrinks to a small set of highly visible mega-cap equities. In this terrain, information asymmetry vanishes. The edge that produced twenty-seven percent, finding the undervalued, the misunderstood, the temporarily distressed, cannot survive the visibility required to deploy ten-figure sums. The business has become, by mathematical necessity, a hunter that must pursue its quarry in open daylight, while competitors with equivalent capital and faster algorithms circle the same targets.
Permanent Capital Without a Maturity Date
The economic engine rests on a liability structure that no competitor can replicate without decades of organic construction. Insurance float represents premiums collected upfront and held against future claims. In the interval, sometimes years, sometimes decades, Berkshire invests these funds. Unlike debt, float requires no interest payments. Unlike equity, it dilutes no ownership. It is, in effect, a revolving line of credit from policyholders that renews itself as long as underwriting discipline maintains favorable loss ratios. This is not merely a funding advantage; it is a different category of capital. While competitors must pay for leverage or answer to shareholders quarterly, Berkshire commands a resource that compounds without friction.
The durability of this advantage rests on regulatory barriers and operational complexity. Building a property-casualty insurer capable of generating tens of billions in float requires navigating state-by-state regulatory mazes, sustaining underwriting discipline through soft markets when competitors slash prices to capture premium volume, and maintaining claims-paying credibility through catastrophes that rupture lesser balance sheets. These are not strategies; they are moats dug over half a century. Yet the float mechanism contains an implicit assumption: that the capital can be deployed at returns exceeding the cost of carrying it. When the asset base was small, this was trivial. A single overlooked regional bank or niche manufacturer could absorb the year’s cash flow and generate the twenty-seven percent. Now, with annual cash generation measured in tens of billions, the float requires a pipeline of billion-dollar opportunities simply to maintain velocity. The liability structure remains perfect. The asset side of the equation faces arithmetic compression.
The Network Effect of a Single Node
In 1997, at Allen & Company’s Sun Valley Conference, Bill Gates shared a stage with Roberto Goizueta, then chief executive of Coca-Cola, and Warren Buffett. The panel was moderated by Don Keough, former president of Coca-Cola and Buffett’s neighbor from Farnam Street in Omaha during the 1950s. Gates, at the height of Microsoft’s dominance, referenced Buffett’s quip that he could improve Coca-Cola’s earnings simply by employing a ham sandwich as management: testament to the brand’s operational durability. The moment captured a generational shift in business power, from consumer brands to technology, but it also revealed something structurally specific about Berkshire’s competitive position: the deals do not arrive through investment banking pitches or auction processes. They emerge from a social graph built over fifty years, from White House dinners and Hood Canal weekends, from relationships that predate the fortunes they now govern.
Consider the Coca-Cola investment. In 1987, following the New Coke debacle, Berkshire deployed $1.2 billion, six percent of the company, into a distressed equity. The entry price reflected temporary operational failure and brand mismanagement. By the mid-1990s, the position had appreciated tenfold. This was not the result of financial engineering or operational intervention; it was the consequence of relationship capital. Don Keough, by then chief operating officer of Coca-Cola, had reconnected with his former neighbor at a White House dinner. The proximity provided early, informal intelligence about the company’s trajectory that was unavailable to the market. More critically, it provided access to the boardroom during the crisis. Similarly, during the Salomon Brothers Treasury bond scandal in 1991, Buffett’s personal intervention, his credibility deployed to reverse a federal decision that would have destroyed the firm, generated returns measured not merely in dollars saved but in reputation earned. Berkshire became the white knight of last resort, the capital source that could move faster than committees because it answered to a single node.
This is the underappreciated structural advantage: operational leverage in capital allocation. Berkshire can deploy a billion dollars in a single decision without partnership committees, due diligence delays, or regulatory filings that signal intent to the market. This speed allows it to capture distressed assets, New Coke, Salomon, before repricing occurs. Competitors face governance friction; Berkshire faces only the limitations of its founder’s bandwidth. Yet this advantage is non-transferable. It took fifty years to construct the network of relationships that generate proprietary deal flow. The next generation of management cannot inherit the friendships formed on Farnam Street in the 1950s, nor the credibility earned by intervening in the Salomon crisis. The advantage is biological, not institutional.
Hunting in the Open
The relationship capital matters because the alternative, sourcing opportunities through public markets, has become structurally disadvantageous. When managing ten billion dollars, one can find twenty exceptional opportunities annually and generate outsized returns. When managing one hundred billion dollars, one must find two hundred such opportunities, or concentrate heavily in a handful of massive positions. The latter path leads inevitably to large-cap equities where every detail is analyzed by thousands of institutional investors, where pricing efficiency approaches perfection, and where the information asymmetries that produced the twenty-seven percent returns cannot exist.
The Coca-Cola trajectory illustrates this entropy precisely. The tenfold appreciation between 1987 and the mid-1990s reflected both operational recovery and multiple expansion in a less efficient market. Over the subsequent twenty-five years, the position appreciated only 3.5 times. The moat remained intact, Coca-Cola still dominates global beverage distribution, but the returns compressed because the entry price reflected full visibility. The market had learned to price the durability correctly. When Berkshire writes a billion-dollar check into Coca-Cola today, it is not exploiting an information edge; it is making a macro bet on global consumer habits and currency movements, indistinguishable from the thesis of a hundred other large-cap value investors. The twenty-seven percent ghost whispers that this is not the same activity that built the fortune.
Pattern Recognition Against Paradigm Shifts
The reliance on pattern recognition, identifying durable moats through historical analogy, creates specific fragilities when technological discontinuities alter the underlying terrain. In 1991, Buffett expressed concern that cable television would erode newspaper readership, threatening the economics of the Washington Post and other media holdings. The observation was correct as far as it went. Cable did fragment attention. But the existential threat was not cable; it was the Internet, which would destroy the classified advertising monopoly that subsidized print journalism. The pattern recognition identified the proximate competitor while missing the paradigm shift.
Similarly, in 1992, Bill Gates, by then a close friend and bridge partner, predicted that television networks would survive the Internet because they owned content libraries that could be repurposed across platforms, while Kodak was “toast” due to digital photography. The forecast about Kodak proved prescient, but the confidence in network durability underestimated how completely digital distribution would atomize content ownership. Berkshire’s largest consumer holdings, Coca-Cola, See’s Candies, rest on assumptions about brand preference stability and syrup-based distribution models that assume technological continuity. If direct-to-consumer delivery, personalized nutrition, or synthetic alternatives disrupt the physical distribution of sugar water, the tenfold returns of the past become mathematically impossible to repeat. The moat assumes the landscape remains stable; the landscape is shifting beneath it.
The Personal Position Problem
In 1991, as Microsoft’s dominance became apparent, Buffett purchased one hundred shares of the company personally, not through Berkshire, to “keep up with Bill Gates.” The purchase was trivial in dollar terms, a rounding error on a personal balance sheet already exceeding five billion dollars. Yet the decision reveals a structural conflict that operates as an economic variable. By holding Microsoft personally rather than institutionally, Buffett created an information firewall. Deep operational diligence, access to strategic plans, competitive intelligence, management’s unfiltered assessment of threats, would have constituted material non-public information. As a personal shareholder, Buffett could not receive such briefings without facing insider trading constraints. As Berkshire’s capital allocator, he could not deploy the firm’s capital into the sector’s dominant player because the personal position blocked the information channel required for fiduciary duty.
This is not merely a governance footnote. It represents a opportunity cost measured in tens of billions of dollars. While Berkshire accumulated zero exposure to the technology sector’s dominant platform, Gates explicitly recommended Microsoft and Intel as the only holdings necessary for a diversified portfolio. The pattern recognition that excelled at evaluating See’s Candies and Coca-Cola, businesses where the product and distribution remain static for decades, could not adapt to software economics, where network effects and zero-marginal-cost scaling create entirely different competitive dynamics. The personal shareholding served as a substitute for institutional conviction, allowing curiosity without commitment, but it also prevented the accumulation of expertise that might have altered Berkshire’s trajectory away from the stagnating returns of consumer staples.
When the Node Goes Dark
All of these tensions, scale compression, technological blind spots, the capital deployment ceiling, converge on a singular fragility: the architecture is founder-dependent. The speed of decision-making, the proprietary deal flow, the ability to intervene in crises like Salomon or negotiate board positions during New Coke, hese derive not from Berkshire’s institutional structure but from a specific individual’s fifty-year accumulation of social capital and pattern libraries.
The twenty-seven percent returns of the early partnership era relied on information asymmetry. The current model relies on relationship asymmetry, ccess to deals and interventions available only to a specific node in the network. When that node disappears, whether through mortality or retirement, the mechanism does not transfer. Successors cannot inherit the friendships formed with Don Keough in the 1950s, the trust earned during the Salomon crisis, or the informal intelligence networks that identify distressed mega-caps before they become auction processes. The next generation of leadership will inherit the float, the permanent capital structure remains, but they will hunt without the relationship infrastructure that has sourced the highest-return investments.
At Sun Valley in 1997, as Gates joked about the ham sandwich, the photograph captured three men representing different economic eras: the industrial brand builder, the platform technologist, and the capital allocator who bridged both. The image also captured a closing window. Berkshire’s model, converting insurance float into permanent capital and deploying it through a singular decision-making node, generated returns that defined an epoch. But the twenty-seven percent ghost reminds us that the conditions producing those returns no longer exist. The capital base has grown too large for the inefficiencies that nourished it, the moats face technological erosion, and the social graph that sources the best deals cannot survive its architect. The machine will continue to compound, float ensures that, but the rate of compounding faces structural gravity. The hunter must now pursue larger, more visible quarry, without the guide who knew every hidden path.
The Salomon Brothers intervention in 1991 provides the clearest available case study of how the economic engine converts reputation into returns, and why that conversion cannot be institutionalized. When the Treasury Department discovered that Salomon’s traders had submitted unauthorized bids in government auctions, the firm faced not merely a liquidity crisis but an existential regulatory death sentence. The federal government moved to suspend Salomon’s primary dealer status, a designation that functioned as the firm’s oxygen supply. Without it, the investment bank would have collapsed within days, taking substantial Berkshire capital with it.
The standard institutional response to such a crisis involves committees, legal reviews, and carefully negotiated settlements that extend over weeks. Salomon had no weeks. The mechanism that reversed the federal decision was not Berkshire’s balance sheet, though that helped, but a specific individual’s accumulated credibility capital. The Treasury Secretary took a phone call. The suspension was modified. The firm survived, and Berkshire’s position ultimately generated substantial returns. But the economic variable at work was not capital allocation skill in the abstract; it was the specific gravity of a reputation built over four decades of under-promising and over-delivering. A successor, however capable, cannot inherit the phone calls. They inherit the phone, but not the relationships in the contact list, nor the trust that causes a Cabinet secretary to pause during a market meltdown and reconsider.
This creates a structural asymmetry that worsens with scale. As the capital base grows, the frequency of crisis-driven opportunities required to maintain historical returns increases proportionally. Yet the founder’s bandwidth remains fixed. In the 1960s, a single distressed textile manufacturer could absorb the year’s investable funds. By the 1990s, the machine required multiple Salomon-level interventions or billion-dollar equity stakes simply to maintain velocity. The personal network that sources these opportunities, Don Keough reconnecting at a White House dinner, the bridge games with Gates that reveal management temperament, operates on human time, not capital time. The relationships renew through physical proximity and decades of reciprocal favors, not through institutional processes. When the founder’s calendar becomes the constraint on deal flow, the business has transformed from a capital allocation vehicle into a personality-driven merchant bank, whether it admits the transition or not.
The Coca-Cola investment mechanics reveal the same dependency on non-replicable social infrastructure. In 1987, following the New Coke disaster, the market priced Coca-Cola as a damaged brand facing permanent erosion of market share. Berkshire’s $1.2 billion deployment, six percent of the company, represented a contrarian bet on durability that required more than financial modeling. It required board-level certainty that Goizueta’s management team would not repeat the error, that the distribution infrastructure remained intact, and that the panic was temporary. Don Keough, who had played with the neighborhood children on Farnam Street in Omaha during the 1950s, provided the informal intelligence channel that validated these assumptions. This was not due diligence in the institutional sense; it was embedded knowledge of character formed over thirty years of social continuity.
The returns reflected this access. By the mid-1990s, the position had appreciated tenfold. Yet the subsequent twenty-five years generated only 3.5 times return: compression that precisely tracks the transition from relationship-driven entry to market-efficient holding. Once the position size grew too large to be managed through personal networks, once Coca-Cola became a liquid mega-cap that traded on global macro factors rather than temporary distress, the edge disappeared. The float required deployment, so the capital remained, but the twenty-seven percent ghost whispered that this was now a passive bet on consumer stagnation rather than an exploitation of temporary dislocation.
The constraint operates arithmetically. With $100 billion in equity capital, a position size sufficient to move the needle, say, five percent of the portfolio, requires a $5 billion deployment. How many publicly traded companies can absorb $5 billion without moving the price against the buyer? How many of those trade at prices offering meaningful margin of safety to a buyer who cannot sell without collapsing the bid? The answer approaches zero. The result is a forced migration toward index-like concentration in the largest available equities, where the only available “edge” is the patience to hold through volatility that institutions with redemption pressures cannot tolerate. This is a real advantage, Berkshire’s permanence does allow it to endure drawdowns that would force leveraged competitors to liquidate, but it generates mid-single-digit returns, not twenty-seven percent.
The technological blind spot compounds this arithmetic trap. In 1991, Buffett worried publicly that cable television would fragment newspaper readership, threatening the Washington Post and other publishing holdings. The observation demonstrated functional pattern recognition: media attention is zero-sum, and new distribution channels steal time from old ones. But the pattern matched cable against newspapers, missing the Internet’s existential threat to classified advertising: he economic subsidy that had sustained print journalism for a century. By 1992, Bill Gates had articulated a more complete vision, predicting that Kodak would become “toast” while believing networks would survive through content repurposing. Both predictions were partially correct, but both missed the deeper restructuring that platforms like Microsoft itself would enable.
Berkshire’s failure to participate in the technology platform shift was not merely a missed trade. It was a structural consequence of the capital allocation firewall created by personal shareholdings. When Buffett purchased one hundred shares of Microsoft personally in 1991, to “keep up with Bill Gates”, he created an information barrier that prevented institutional investment. Deep operational diligence would have required access to Microsoft’s strategic roadmap, competitive intelligence, and management’s assessment of emerging threats. As a personal shareholder, Buffett could not receive such material non-public information without insider trading constraints. As Berkshire’s allocator, he could not deploy the firm’s capital without that diligence, lest he violate fiduciary duty. The personal position, trivial in dollar terms, functioned as an insulation mechanism that kept the dominant platform of the era outside the portfolio while the partnership accumulated more Coca-Cola and See’s Candies.
See’s Candies represents the anti-Microsoft in this taxonomy: business where the product, distribution, and competitive dynamics remained static for decades, allowing pattern recognition to function perfectly. Acquired in 1972, See’s required no technological adaptation, faced no platform risk, and generated cash flows that could be predicted with actuarial certainty. The twenty-seven percent returns of the early partnership era depended on finding dozens of See’s Candies, mall, predictable, mispriced by markets focused on glamour growth. By the 1990s, with $10,000-per-share stock prices and a net worth exceeding $5 billion for the founder, the See’s model could no longer absorb the capital generated by the float machine. The hunt turned necessarily toward larger prey, where the predictability that characterizes boxed chocolates and sugar water becomes dangerous when applied to software, networks, or platform economics.
The Sun Valley Conference in 1997 captured this inflection with documentary precision. On stage sat Roberto Goizueta, architect of the New Coke recovery and the global expansion that had generated Berkshire’s tenfold return; Bill Gates, whose Microsoft had by then surpassed Berkshire in the velocity of wealth creation; and Buffett, the allocator who had selected Goizueta but passed on Gates. Moderating was Don Keough, the Farnam Street neighbor who had facilitated the bridge between Omaha and Atlanta. The “ham sandwich” reference, Buffett’s observation that Coca-Cola’s moat was so wide that a ham sandwich could run the company, drew laughter, but it also marked the boundary of the old model’s applicability. The comment assumed operational stability, predictable consumer behavior, and distribution moats that technology could not breach. Gates sat beside him, representing exactly the force that would eventually breach them.
The generational shift visible at Sun Valley was not merely a changing of the guard in the Forbes rankings. It represented the emergence of a new economic grammar, network effects, zero marginal cost, platform dominance, that rendered the vocabulary of moats and ham sandwiches increasingly archaic. The twenty-seven percent returns had come from identifying businesses protected from change. The next era’s returns would come from businesses that accelerated change. Berkshire’s structure, permanent capital, conservative underwriting, relationship-driven deal flow, optimized for the former and could not adapt to the latter without violating its own economic logic.
The succession problem follows necessarily. The next generation of leadership will inherit the float mechanism, the insurance operations, and the portfolio of consumer brands. They will not inherit the White House dinner invitations, the Salomon crisis phone calls, or the Farnam Street neighborhood history that turned a chance encounter into a billion-dollar entry point. The deal flow that generated the highest returns, istressed opportunities identified before auction processes, board seats secured during management crises, preferential terms negotiated in private placements, epends on a social graph that took fifty years to construct and that expires with its central node.
Some analyses suggest that culture can be codified, that the “Berkshire way” can be transmitted through written principles and decentralized management. This misses the economic function of the founder’s specific relationships. Culture is not a substitute for deal flow; it is the residue of deal flow once the transactions close. The underwriting discipline that protects the float requires not just principles but the specific judgment of individuals who have seen multiple credit cycles and who cannot be replicated by hiring actuaries with impressive credentials. The speed of decision-making that allows billion-dollar checks to clear in hours rather than months requires not just streamlined governance but the confidence of boards who trust the specific signature on the check.
The fragility becomes visible when examining the implicit assumption that large, visible opportunities with structural moats will continue to emerge at a rate sufficient to deploy the growing capital base. This assumption already shows cracks. The Coca-Cola returns compressed from 10X to 3.5X as the position grew and the entry price reflected full market visibility. The newspaper holdings, once paragons of local monopoly economics, faced secular decline that pattern recognition identified too late. The technology platform that Gates represented at Sun Valley, Microsoft, Intel, the emerging internet, generated returns that dwarfed Berkshire’s consumer staples, yet remained inaccessible due to the information firewall created by personal shareholdings and mental-model incompatibility.
As the asset base approaches $100 billion and beyond, the machine faces a deployment trap. The float continues to grow, requiring investment. The available universe of billion-dollar opportunities capable of generating historical returns shrinks. The founder’s network, while still functional, cannot scale to source twenty such opportunities annually. The result is a gradual, almost imperceptible migration toward index-like returns, masked by the accounting performance of earlier vintage investments and the mechanical growth of the insurance operations. The twenty-seven percent ghost haunts not because the current returns are poor, mid-teens compounding remains impressive against market averages, but because the structure that generated them has been replaced by a different organism entirely: not a value-extraction partnership exploiting market inefficiencies, but a permanent capital vehicle making macro bets on global consumer stagnation, praying that the moats survive technological shift long enough for the float to find a home.
The hunter has grown too large for the forest, yet cannot survive outside it. The guide knows every path because he walked them for fifty years, but the quarry have moved to terrain his maps do not chart. When the guide steps back, the hunting party will possess the rifles, the ammunition, and the licenses, ut they will stand at the edge of an empty clearing, wondering why the game no longer appears.
The specific mechanics of this trap become visible when examining the distinction between See’s Candies, acquired in 1972, and Microsoft, purchased personally in 1991 but never institutionally. See’s represented the perfect fit for the emerging Berkshire model: a regional confectioner with entrenched West Coast distribution, zero technological obsolescence risk, and pricing power that allowed for annual increases without volume loss. The capital required was modest, $25 million, allowing the position to generate the twenty-seven percent returns that the partnership structure demanded. The business required no adaptation; chocolate consumption patterns in San Francisco in 1972 differed imperceptibly from those in Omaha in 1952 or, crucially, those in Los Angeles in 2022. The moat was geographic and psychological, immune to semiconductor cycles or software versioning. Management mattered only at the margin; the “ham sandwich” theory held because the product and distribution infrastructure had remained static for fifty years, requiring only that executives not actively destroy the brand through incompetence or adulteration.
Microsoft presented the inverse profile. By 1991, when Gates suggested that Buffett purchase shares to “keep up” with the technology revolution, the business required not just capital but interpretive frameworks that Berkshire’s architecture could not accommodate. Software economics violate the See’s Candies template: returns come not from physical scarcity and brand habit but from network effects, switching costs, and rapid obsolescence cycles that demand continuous managerial adaptation. The moat in technology is dynamic, requiring constant excavation; it erodes the moment management stops digging or misses a platform shift. Gates understood this distinction intimately; his 1992 prediction that Kodak would become “toast” while networks survived revealed an operator’s grasp of digital entropy that the capital allocator, fixated on syrup distribution and newspaper local monopolies, could not internalize. The “ham sandwich” manager at Microsoft would have lasted approximately eighteen months before the platform became obsolete, replaced by competitors who understood that in software, stasis is death.
The personal purchase of one hundred Microsoft shares functioned as a structural relief valve, allowing curiosity without commitment, but it simultaneously constructed an information barrier that prevented institutional deployment. Had Berkshire sought to accumulate a meaningful position, say, five percent of Microsoft, comparable to the Coca-Cola stake, the firm would have required board representation and operational diligence. Such access would have entitled Buffett to material non-public information regarding Windows development timelines, OEM negotiations, and the nascent Internet strategy. As a personal shareholder holding a symbolic position, Buffett could receive such intelligence without triggering insider trading constraints; as the fiduciary for Berkshire Hathaway, he could not act on it without violating Chinese walls or creating disclose-or-abstain obligations. The personal position thus served as an insulator, preventing the accumulation of knowledge that would have created legal barriers to trading.
The contrast with the Coca-Cola investment mechanics reveals the asymmetry. In 1987, Keough’s position as COO allowed informal conversations about New Coke’s failure and Goizueta’s recovery strategy that would have constituted material non-public information had Berkshire been actively accumulating shares. Because the relationship predated the investment and existed in a sector, consumer packaged goods, where operational details change slowly, such intelligence could be absorbed and acted upon without triggering the same insider trading scrutiny that would attend real-time software development updates. Technology moves faster than disclosure windows; chocolate recipes do not.
This dynamic illustrates how the relationship capital that generated the tenfold Coca-Cola return became, in the technology sector, a liability masquerading as a friendship. While Berkshire continued accumulating Coca-Cola shares through the 1990s, averaging up in a position that would appreciate only 3.5 times over the subsequent twenty-five years, the software platform that Gates had built generated returns of one hundred times or more over the same period. The float, that magnificent permanent capital, was deployed into the decaying moat of carbonated sugar water while the living moat of operating systems remained unmapped. The specific opportunity cost is stark: a $1 billion Microsoft position in 1991, held through the decade, would have generated returns sufficient to purchase the entirety of the Coca-Cola holding several times over, with capital remaining to fund the Geico acquisition and still compound at rates that would make the twenty-seven percent ghost weep.
The cognitive cage tightened as the scale grew. The pattern recognition that identified See’s Candies as a fortress, correct in 1972, when the moat was physical and local, became a filter that excluded platforms in 1992. The successor managers will inherit not just the rifles and the ammunition but these specific cognitive maps, etched over decades of successful consumer-brand investing. Those maps mark the territory where the game once gathered: branded physical goods, regulated utilities, insurance float arbitrage. They do not chart the algorithmic landscapes where value now accumulates, nor do they account for moats that require management to sprint merely to maintain position. When the guide steps back, the hunting party will possess the permanent capital and the governance speed, ut they will deploy it using mental models optimized for a terrain of physical scarcity in an era of digital abundance, hunting ham sandwiches while the ecosystem requires hunters who understand that the sandwich itself is being reconstituted at the molecular level by software.