The TL:DR version
The breakthroughs we inherit (the internet, genomics, AI) took decades to imagine and build, funded by people who knew they might never see the results. Today, short-termism dominates: quarterly earnings, election cycles, viral news, shrinking attention spans. To recover our capacity for long-horizon thinking and doing, we need to reinvest in four pillars: Patience, Imagination, Institutions, and Memory. But these aren't just about sustaining effort over time. Building toward the genuinely unknown expands what becomes imaginable in the first place; technology at its most transformative is a scaffold for seeing what couldn't be seen before. This essay makes the case for why and how.
Photo by Andy Song on Unsplash
The long form version
Introduction
Invention and innovation are not the same thing but we seem to have compressed them into one and the same.
That compression produces novelties, not breakthroughs, it selects for ideas that can demonstrate traction within funding cycles of eighteen months, not the ones that might reshape the world in eighteen years. The bias toward immediacy has become so normalised that we barely notice it, yet it represents a profound narrowing of what we allow ourselves to attempt.
Consider the technologies that transformed civilisation: the telephone, the internet, genomic sequencing, artificial intelligence. None emerged from a single sprint of invention. Each required decades of experimentation and embedding (across regulation, standards, manufacturing capability, and social practice) before becoming infrastructure rather than curiosity. The transistor was invented at Bell Labs in 1947; the smartphone arrived sixty years later. ARPANET sent its first message in 1969; the web became culturally dominant in the 2000s. The Human Genome Project took thirteen years of coordinated international effort to deliver a single reference sequence.
These timelines reveal that the breakthroughs we inherit were funded and protected by people who knew they might never see the results. They built architecture: not just artefacts. They built the scaffolding of institutions, standards, and cultural practices that allowed difficult ideas to survive long enough to become indispensable, and while the pace of innovation and adoptions moves faster as we evolve and build on what we know, the need for these foundations is still important.
I call this an architecture of wonder because wonder has two meanings, and both matter here. Wonder is awe: the astonishment at what exists, the recognition that the universe is stranger and more magnificent than we imagined. But wonder is also the act of imagining what could be and experimenting to make ‘it’ : I wonder what it would take to cure this disease, reach that star, build a society where human flourishing is the default. The first meaning grounds us in humility; the second propels us toward ambition AND action. A culture that loses either becomes arrogant or stagnant.
This essay argues that sustaining long-horizon innovation requires more than a bit of time, a set of quarterly OKR’s or capital. It requires deliberate structural investment in four foundational pillars: Patience, Imagination, Institutions, and Memory. These are not soft virtues, they are engineering requirements for any society that wants to build the future rather than merely consume it.
Why Long Horizons Matter
The case for long-horizon thinking is structural.
First, it enables innovation unconstrained by immediate utility. AT&T’s decision in the early twentieth century to fund pure research at Bell Labs, giving scientists latitude to explore fundamental science without obvious products in sight, produced the transistor, the laser, information theory, Unix, the C programming language, and the discovery of cosmic microwave background radiation. Nine Nobel Prizes and five Turing Awards emerged from that single institution. These breakthroughs did not come from setting out to increase next year’s profits. They came from decades-long inquiry into the nature of matter and information, protected from the pressure to demonstrate quarterly returns.
Second, long horizons expand the canvas of problems worth tackling. Short-termism fixates on incremental improvements; long-term thinking and doing encourages engagement with generational challenges (curing diseases, transitioning to sustainable energy, extending human presence beyond Earth) that cannot be solved overnight but are immensely worthy of pursuit. When CERN commits to building a particle collider that takes twenty-five years from conception to first collision, it is asking questions whose answers matter for centuries: What is the origin of mass? What else exists beyond the Standard Model? Plans are already underway for a Future Circular Collider that may not operate until 2070.
Third, long-term thinking and doing cultivates resilience and responsibility. It forces confrontation with trajectory rather than snapshot. Climate change, biodiversity loss, infrastructure decay: these are problems that compound invisibly across decades until they become crises, and then we try and mitigate, whilst forgetting to innovate for adaptation. A long view makes preventive action possible. It makes intergenerational responsibility concrete. If we consider people a century from now as stakeholders in today’s decisions, our ethical calculus shifts. The future is ancestral and a partnership between the present, the past and what’s to come. The individuals who planted sequoia saplings 150 years ago, or who set aside national parks, were making choices whose true benefits they themselves would never fully enjoy. They acted so that others could inherit a world of natural wonder.
This practice has been called “cathedral thinking”: launching projects on timelines longer than a human life, sustained by each generation laying part of the foundation for the next. The medieval architects of European cathedrals often began construction knowing completion would come long after they were gone, sometimes a century or more later. Yet they persisted, motivated by a vision dedicated to future worshippers. We see secular versions of cathedral thinking in modern science: the quest for controlled nuclear fusion has stretched over sixty years and may require decades more, but its promise (virtually inexhaustible, clean energy) is so revolutionary that scientists pass the torch from mentor to student, across generations, rather than give up.
Finally, long horizons inspire. They connect us to something larger than ourselves. When we cast our perspective to a distant goal, whether building a 10,000-year clock in a mountain or sending explorers to Mars, we rekindle a sense of purpose that transcends the everyday. As Dario Amodei has argued, we need genuinely inspiring visions of the future, not just risk mitigation: “some positive-sum outcome where everyone is better off, something to rally people to rise above their squabbles and confront the challenges ahead.” Fear of catastrophe can spur avoidance but it is hope that drives construction.
Our time is one of converging scientific revolutions: AI, quantum technology, space exploration, neuroscience, bioengineering. These fields (and new fields that will emerge at the intersections of these) present immense promise and peril reaching far beyond our own generation. They place upon us a responsibility to ensure that our accelerating knowledge is guided by wisdom, foresight, and respect for those who will inherit the world we shape today. The convergence of consciousness research with neuroscience, of quantum physics with psychology, of data science with art: these boundary-crossing syntheses demand values as much as velocity.
History Tells Us What the Future Can’t
History reveals cycles, reversals, and contingencies that should humble any straight-line projection of the future.
For most of recorded history, people did not assume that the future would automatically improve on the past.
The Greeks wrote of a Golden Age already lost.
Aristotle observed that empires grow, flourish, decay, and die in cycles, much as organisms do.
The fall of the Western Roman Empire led to centuries of fragmented capability in Europe.
The recipe for Roman concrete, superior in many ways to modern formulations, vanished.
Knowledge was lost and techniques were forgotten - this was all genuine regression!
Even our current faith in progress has been jolted repeatedly: the twentieth century delivered electricity, aviation, antibiotics, and computers, but also world wars, genocides, and nuclear weapons. Technological advancement does not automatically produce moral or social advancement if long-term values are not actively crafted or maintained.
History also teaches that context determines whether a discovery transforms the world. Hero of Alexandria built a working steam engine in the first century CE, a small spinning sphere driven by steam but it did not launch an industrial revolution. Perhaps cheap slave labour reduced the incentive; perhaps the supporting scientific theory was absent. It took 1,700 years, and different socioeconomic conditions in Britain, for steam power to create a transformation. Similarly, the Chinese invented gunpowder in the ninth century and used it mainly for fireworks.
The future is not predetermined by invention alone; it is shaped by how society chooses to deploy and develop what it has.
One particularly relevant lesson concerns forgetting. Vint Cerf, one of the architects of the internet, has warned of a “digital dark age”: the possibility that despite creating more data than ever, our era could become poorly remembered if that data is not preserved in accessible ways. Digital formats become obsolete; storage media decay; software required to read old files disappears. Much of antiquity is known to us only through copies of copies of manuscripts. What was saved was often a matter of chance. The loss of the Library of Alexandria is a collective tragedy. We should not assume that our digital abundance protects us from similar fragility.
A crucial historical lesson concerns reversibility. Societies can and have undone progress through negligence or misplaced priorities. The Ming Dynasty in fifteenth-century China had the world’s largest naval fleet and undertook grand voyages of exploration under Admiral Zheng He, reaching Africa decades before Columbus reached the Americas. Yet political shifts led to banning further voyages and even destroying ship blueprints. Its naval technology stagnated as Europe’s leapt ahead. This was a voluntary relinquishing of a long-term vision.
The past teaches us to expect non-linearity: golden ages and dark ages, leaps and setbacks, context and contingency. It cautions against arrogance and forgetting, while also giving hope that enlightened action can steer outcomes. History tells us what the future alone cannot: that we have been shortsighted before, and paid the price, but also that we are capable of greatness when united by long-term vision.
History also provides uplifting evidence that foresight and perseverance can alter seemingly predestined outcomes. When scientists discovered that chlorofluorocarbons were depleting Earth’s protective ozone layer, it looked like an environmental catastrophe in slow motion. Yet the world’s nations agreed in 1987 to the Montreal Protocol, phasing out CFCs. The ozone layer is now on track to recover, a rare case of global long-term cooperation succeeding. The eradication of smallpox, an international campaign lasting from the 1960s to 1980, rid the world of a disease that had plagued humanity for centuries. Such victories show that long-term projects can succeed when guided by science, shared purpose, and institutions willing to think beyond immediate interests.
The Present’s Challenge
We are surrounded by achievements that would have stunned our ancestors. We can sequence genomes in days, communicate across the globe instantly, and deploy artificial intelligence to solve problems. Yet even as our capabilities have expanded, our temporal horizons in many arenas have contracted. The faster our tools allow us to act, the less time we allot for reflection and long-range planning. This misalignment, powerful means married to myopic ends, is one of the greatest threats to sustaining an architecture of wonder.
The erosion of sustained attention is measurable. Research shows that average attention spans on screens have shortened significantly over the past two decades. In 2004, people might spend several minutes on a task before switching; now many struggle to go seconds without checking for updates. Deep concentration is a prerequisite for complex problem-solving and creativity. If an individual cannot stay with a challenging problem longer than a TikTok video, how will they engage with a goal that might take years to accomplish?
This attenuation is not merely personal; it is institutional. News cycles operate on a 24/7 breaking-news model, leaving little room for historical perspective. Social media amplifies immediacy: what’s trending now eclipses what was important yesterday. The collective memory of online discourse is remarkably short. Meaningful analyses get buried under viral trivia. Slow-burning crises like climate change or infrastructure decay require consistent attention over years, yet they compete against the latest outrage in a race they struggle to win.
Thirty years ago, I studied psychology and computer science together, convinced that human and machine intelligence would converge. The world said stick to maths; I ignored it. I’ve spent my career since building long bets: cloud, data, space tech in Silicon Valley research labs; then translating science into product across sectors, always working ten to fifty years ahead. I know what it takes to build things that cannot yet be seen or adopted. I also know that the mechanisms for doing this are under threat.
In business, the pressure toward quarterly capitalism has intensified. Corporations judged by quarterly earnings face relentless pressure to deliver immediate results, often at the expense of R&D investment or workforce development whose payoffs lie years ahead. In the mid-twentieth century, companies like AT&T could fund Bell Labs for fundamental research without clear short-term profit. Today, even tech giants with enormous resources hesitate to pursue projects without marketable products in a few years. The venture capital model, while spurring rapid innovation, often expects explosive growth or quick exit rather than steady building over decades.
In government, election cycles drive short-termism. Leaders operating on two, four, or five-year terms have incentives aligned with visibly delivering something within that term. Long-term policies (overhauling education, combating climate change, building resilient infrastructure) may not show benefits until many years later, perhaps under a successor’s watch. Politicians may shy away from such investments or from necessary but initially unpopular measures whose benefits lie beyond their tenure.
What if there was a model that rewarded delivering part of a long-term outcome and then passing the baton well? I’ve started, scaled, and transformed organisations, but I’ve also known when it’s time to move on and let someone else carry the next wave. That only works, though, when successors build on foundations rather than tear them down. Too often, new leaders assume the previous wave failed and start from scratch, breaking what was working rather than extending it. The result is cycles of destruction dressed up as reinvention: waves of pain rather than conscious progress. A short-term mind cannot think in long-term change, so it spirals, breaking and rebuilding the same ground over and over.
There are counter-currents. Climate activists pleading for a habitable planet when they are old represent deeply long-term thinking. Some investors promote patient capital and longer investment horizons. DARPA explicitly invests in high-risk, high-reward projects that may take a decade to mature; the internet and GPS emerged from exactly this model. But these remain exceptions against a dominant trend of acceleration.
Long-horizon research has slowed just as adoption of established innovations accelerates. We are consuming the pipeline of foundational ideas faster than we are replenishing it. The technologies transforming industries today (AI, cloud computing, genomics) emerged from research conducted decades ago. If we do not invest in the next wave of foundational science now, we will find ourselves, a generation hence, with nothing transformative left to adopt.
The challenge can be summarised as a deficit of long-range attention and continuity at nearly every level. People are not incapable of thinking long-term (many yearn for greater purpose and stability) but the structures and rhythms of contemporary life do not reward it. Reversing this will require intentional effort: redesigning institutions, cultivating habits, and shaping narratives that re-value the long view. But it will also require something more fundamental: different leaders.
The Leaders We Aren’t Building
Consider the arithmetic of human development. A leader who will guide critical decisions in 2075 is being shaped right now, in schools, in early careers, in the assumptions they absorb about what ambition looks like and what success means. If we want leaders capable of fifty-year thinking by mid-century, we needed to start instilling these skills a while ago. The lead time for developing human judgement is measured in decades, not quarters, yet we have no approach to identifying or cultivating people who can think and act on civilisational timescales. The education system doesn’t build these skills, and other mechanisms are corroding them.
Our current leadership pipelines optimise for the wrong timelines. Business schools train for quarterly performance, political systems reward those who can win the next cycle, not those who can see past it, scientific training, which should cultivate patience, increasingly pressures young researchers toward rapid publication rather than deep problems.
From secondary school onwards, we sort students into disciplines and reward depth over breadth. A chemist who understands economics, a historian who can code, a biologist versed in ethics: these combinations are rare not because they lack value but because curricula and career structures discourage them – because it’s easier for organisations to put people into boxes that can be administrated. Yet the problems that matter most (climate change, AI governance, pandemic preparedness, space civilisation) sit at disciplinary intersections. We are mass-producing leaders calibrated for sprints while the challenges ahead demand ultramarathons.
What would it mean to deliberately develop long-horizon leaders? Several capacities seem essential.
Temporal range as a trainable skill. The ability to hold multiple time horizons simultaneously (what must happen this year, this decade, this century) is a cognitive capacity that can be developed through practice. Scenario planning, historical analysis, and exposure to genuinely long-term projects can stretch default planning windows. Leaders should practise moving fluidly between operational urgency and long-term consequence, learning to see how decisions made today ripple forward through decades, or even centuries.
Comfort with outcomes that a successor might get the credit for. Most leadership development assumes you will see the results of your decisions, but long-horizon leadership requires a different psychology: the willingness to start what you cannot finish and to hand over well, rather than clinging to control. The ego structures that drive conventional ambition and rewards may interfere with this. We need leaders who can find meaning in contribution as well as completion, who understand their role may be to advance a relay race.
Knowing what you don’t know, especially about the future. Long-horizon leaders need strong convictions to sustain decades of effort, but they also need the humility to recognise that the world will surprise them. This isn’t sitting on the fence or lacking the confidence to make a decision, it’s the ability to change your mind without losing your way, adjusting the path as needed.
Cross-domain fluency. A leader who understands only technology, or only policy, or only finance, will be outmatched by challenges that integrate all three. Long-horizon leadership requires what the Renaissance called sprezzatura: the appearance of effortless competence across domains, built on genuine (even if imperfect) understanding of multiple fields and the ability to synthesise insights that generalists see but specialists miss.
Being a teacher and graciously passing on the baton. Most leaders neglect this (because they are not incentivised to do so), but if you’re working on something that will take longer than your own career, preparing your successors is essential. This means writing down not just what you decided but why, building organisations that can learn and change, and actively developing the people who will continue after you. The medieval cathedral builders trained the next generation of masons, handing on skills and vision along with materials.
All of this suggests a need for new kinds of developmental environments: fellowships that embed emerging leaders in genuinely long-term projects (climate adaptation infrastructure, space settlement planning, nuclear waste stewardship), apprenticeships with the rare institutions that already think in centuries, or deliberately constructed cohorts that maintain connection across decades of separate work while pursuing a shared long-term vision.
Some models already exist. The Long Now Foundation thinks in ten-thousand-year spans, multi-generational family businesses carry knowledge about succession, indigenous governance structures formally give responsibility to future generations. We need to stop smart leaders from being sorted into conventional career paths, compressed to match the systems and leaders around them.
The Four Pillars of Wonder
Rekindling long-horizon thinking requires reinvesting in certain values and practices. In envisioning an architecture of wonder that can endure, I’ve identified four foundational pillars: Patience, Imagination,Institutions, and Memory. Each represents a quality or system that, when strengthened, gives structure to long-term endeavour.
Patience: Cultivating the Time for Greatness
Patience is the willingness to invest time in what matters, without guarantee of quick reward. It underscores everything from scientific research to personal mastery of a craft. In a fast-paced era, patience can feel anachronistic, but it is in truth a radical act of faith in the future.
The James Webb Space Telescope was conceived in the 1990s. After multiple redesigns, delays, and budget overruns, it finally launched over thirty years later in 2021. During that time, teams of engineers and scientists maintained faith through political scrutiny and technical setbacks. Now operational, Webb is sending back images of the earliest galaxies. The quest for nuclear fusion energy has stretched over sixty years, leading to cynicism in some quarters, yet steady accumulation of knowledge culminated in a net energy gain experiment in 2022. If society’s patience wears thin and funding dries up just short of the goal, we squander decades of investment.
That doesn’t mean throwing money at things without expecting operational excellence. Invention creates the new; innovation scales it. Both demand rigour, but they need different timelines and metrics. Patience means understanding how to fund, measure, and monitor the long game of invention, not abandoning accountability.
Culturally, we can cultivate patience by celebrating slow and masterful work. In Japan, the concept of monozukuri, the art of making things, encapsulates deep dedication to craft. A swordsmith folding steel for the thousandth time, a potter who has thrown clay every day for fifty years to achieve subtle grace in form: these are examples of patience yielding quality that no shortcut can attain. Japan designates certain artisans as “Living National Treasures” for preserving traditional crafts, signalling to younger generations that devoting a life to something worthwhile is respected.
At a societal level, embedding patience may require structural changes: funding models that provide secure, long-term support for critical research, buffered from short-term political winds; longer job tenures or apprenticeships that allow people to grow with a project; educational systems that teach patience explicitly through languages, musical instruments, or assignments that cannot be completed in one sitting. Time is an ally in creation, not just an enemy to beat.
Imagination: Envisioning and Building Possible Futures
If patience provides the temporal canvas, imagination provides the content. It is the capacity to envision and build worlds and possibilities that do not yet exist, the pillar that allows us to dream beyond present circumstances. Without imagination, patience has nothing to work toward.
The generation of engineers and astronauts who took humanity to the Moon often cite childhood reading of Jules Verne or watching early space adventure films. Those imaginative works seeded concrete reality later. Arthur C. Clarke imagined geostationary communications satellites before they were technically feasible; decades later, they became the foundation of global communication. Star Trek’s depictions of handheld communicators and tablet computers in the 1960s inspired technologists who eventually created cell phones and iPads. Imagination in art and culture can precede and catalyse real scientific progress.
Building Beyond Sight
The examples above describe imagination as precursor: the dream that precedes the artefact. But there’s a deeper mechanism at work. The act of building toward the genuinely unknown doesn’t just realise existing dreams, it generates new ones. It expands the space of what can be imagined.
When we built MemoryNet in the early 2000s, we weren’t implementing a vision of Instagram. We were exploring what happened when physical objects acquired digital presence, the early experiments in what would realise the Internet of Things. The prototype revealed possibilities: social sharing, location-aware content, ambient computing. These couldn’t have been fully specified in advance. The technology became a tool for thinking, not just doing. We saw further because we had built something to stand on.
This is the paradox of long-horizon work: you cannot fully plan what you cannot yet imagine, but you can create conditions that reliably produce encounters with the genuinely new. The four pillars aren’t just about patience in waiting. They’re about building the capacity to recognise and act on things that don’t fit existing categories.
Technology, at its most transformative, serves as imagination scaffold. Not faster or bigger or more, but genuinely orthogonal. The medieval cathedral builders knew roughly what a cathedral was. What we’re pointing at is different: creating systems whose implications exceed the imagination of their creators. The transistor’s inventors understood switching but couldn’t have specified smartphones. The ARPANET engineers understood packet switching but couldn’t have specified TikTok.
This changes what the pillars actually do: patience gives the time for implications to surface (you can’t see what you’ve built until the unexpected emerges), memory provides the accumulated vantage points that make new perception possible, and institutions let each generation start from where the last one arrived rather than rebuilding from scratch.
At a societal level, active imagination (not just dreaming) fuels paradigm change. Every significant reform began as a radical idea that many found utopian or dangerous. Over time, through advocacy and incremental progress, those ideas became reality. This should encourage us in confronting present challenges: imagining a carbon-neutral economy or an AI-augmented society that is equitable and humane might seem daunting, but framing these as aspirational narratives is the first step toward engineering them.
How do we strengthen this pillar? By protecting spaces for imagination: interdisciplinary learning in schools, design fiction, arts integrated with STEM to form whole-brain thinkers and doers who learn hands on. By making room for visionaries and cautionary voices alike. By changing the approach from “that’s unrealistic” to “what would need to change to make it possible eventually?” Imagination’s ultimate gift is meaning: it allows us to craft stories and objects about why we do what we do. Frame curing cancer as “slaying the dragon” and suddenly people find motivation that carries them through setbacks.
A word of caution, though: many people call themselves visionaries, polymaths, or systems thinkers. The labels are easy; the work is not. Most organisations are crafting the past in the present, disguised as the future. Real imagination (thinking and doing) requires a different kind of mind: one that makes odd connections, holds contradictions, lives in uncertainty and builds what others can’t yet see. That capacity is rarer than the job titles suggest, and it needs room to operate.
Institutions: Building Lasting Frameworks
If patience and imagination are qualities that individuals and cultures must cultivate, institutions are the concrete structures that embody and support those qualities at scale. By institutions, we mean the organisations, systems of governance, and cultural practices that endure beyond any one person. Institutions are the vessels that carry forward long-term endeavours, allowing them to outlive their founders.
HP Labs’ Cooltown project in the late 1990s envisioned a world where physical objects, places, and people would have web presences, with mobile devices seamlessly bridging physical and digital realms. When I joined HP Labs, I entered through 1501 Page Mill Road in Palo Alto, and the history of expansive ambition hit me immediately. This was a place charged with rigorous experimentation, curious courage, and a fearless commitment to uncertain yet scientifically meaningful ideas. They called it The HP Way: a philosophy built on deep trust and empowerment. On my first day, my boss handed me a $5 million budget, a set of car keys to drive over to our partners at the Exploratorium, and asked me to make something happen. So I did. We created a hundred-year worldview and mobilised under-explored areas of research. The team didn’t just imagine this future; we built working prototypes, foundational technologies, and product demonstrations (including MemoryNet, which anticipated Instagram by six years) that foreshadowed QR codes, location-based services, the Internet of Things, and smartphones as bridges between worlds. We partnered with NASA, Stanford, Berkeley, Georgia Tech, MIT, IDEO, Apple Labs and others on unconventional projects that generated possibility. The global infrastructure for cloud computing wasn’t understood: bandwidth was limited, mobile devices primitive. The project wound down, but its experiments seeded thinking that others would later realise. Cooltown demonstrates that institutional patience sometimes means building toward a vision before the world is ready, and that apparent ‘failures’ often lay groundwork for others to build upon. The technologies I worked on during my time at HP Labs now underpin critical sectors of the economy.
Xerox PARC in the 1970s invented the graphical user interface, Ethernet, laser printing, and object-oriented programming. Xerox itself famously struggled to commercialise most of these innovations, but they became the foundation of the personal computing revolution. Apple, Microsoft, and others built empires on PARC’s inventions. This illustrates a crucial point about institutional research: the organisation that invests may not be the one that reaps the commercial reward, but society benefits regardless. Long-horizon institutions require funding that understands the returns will not fit traditional models.
Bell Labs thrived under unique conditions: the AT&T monopoly provided steady funding and a mandate to improve telecommunications over the long haul, effectively shielding researchers from quarter-by-quarter profit concerns. Its culture mixed scientists and engineers in open campuses to encourage serendipitous collaboration. Management set broad goals but gave researchers freedom in how to reach them. When institutions commit to fundamental research and give talented people time and space, the payoff can be world-changing.
DARPA demonstrates how government can foster long-horizon research. Unlike typical funding agencies, DARPA is small, nimble, and emphasises high-risk projects. Program managers rotate every few years, each empowered to fund daring ideas, and failure is accepted as the cost of originality. Beyond the internet and GPS, DARPA-backed projects contributed to stealth aircraft, drones, and early personal computing research. The model has inspired proposals for similar bodies in non-military domains: ARPA-E for energy, ARPA-H for health.
CERN stands as an institution of international collaboration aimed at fundamental discovery. Since 1954, it has brought together dozens of countries to build ever more sophisticated particle accelerators. These experiments take decades: planning for the LHC began in the 1980s; construction continued through the 1990s and 2000s; runs started in 2008 with upgrades planned into the 2030s. CERN’s governance, a council of member states, reduces the risk that one country’s political shifts will derail the whole project. The World Wide Web was invented at CERN in 1989 to share data. The existence of CERN proves that even in a divided world, we can create durable institutions united by curiosity and long-term purpose. Having worked with the CERN teams, it’s clear that the long game is central to what they commit to working on and that the science will continue to advance over decades and centuries.
Germany’s Mittelstand companies, often family-owned for generations, demonstrate how institutional culture can embed long-termism. These mid-sized manufacturers prioritise steady excellence over rapid growth, invest heavily in apprenticeships and workforce development, and think in decades rather than quarters. Many have dominated global niches for a century or more. Their governance structures, often insulating management from short-term shareholder pressure, enable the patience that precision engineering demands. The lesson is that institutional longevity requires not just mission but structure: the right ownership, governance, and incentives to resist short-term pressures.
Cultural institutions and practices matter too. Japan’s concept of monozukuri (mentioned above) is institutionalised in education and industry as an ethos, reinforced by government recognition and companies that embody it. As a result, Japan has an unusual number of extremely old companies. Some breweries and inns have operated for centuries, weathering wars and upheavals by adhering to values of quality that inspire loyal support. The oldest, Kongō Gumi, was a temple-building company operating continuously from 578 AD until it merged in 2006, an astonishing 1,428-year run. Such longevity is only possible with a strong sense of identity and memory preserved through institutional practice.
Building and supporting institutions that last is our best strategy to embed long-termism into society’s fabric. Institutions act as the skeleton of collective action, keeping it standing when individual agents come and go. They are the cathedrals of our secular age.
Memory: Honouring and Harnessing the Past
The final pillar is memory: the capacity to remember, record, and learn from what has come before. Memory, in a collective sense, is our shared brain: archives, traditions, historical consciousness. Without memory, each generation starts from scratch, repeating mistakes and losing hard-won insights. Ancestry, ritual and culture are memory.
Consider how ancestry shapes long-term thinking. Cultures that venerate ancestors tend to think more naturally about descendants. The Māori concept of whakapapa connects individuals to genealogies stretching back to creation and forward to unborn generations, making intergenerational responsibility visceral rather than abstract. These practices keep the past alive in ways that mere archives cannot, binding communities across time through ritual and obligation.
Memory serves as the tether between past and future, ensuring that progress is cumulative rather than cyclical in the negative sense. A society with good memory can avoid reinventing the wheel or stumbling back into solved problems. Consider public health: because historical records preserved information on prior pandemics, scientists and leaders in 2020 could consult how the 1918 influenza pandemic was handled, gleaning what measures helped and what pitfalls existed. That collective memory, transmitted via history books and public commemorations, suddenly became urgently relevant a century later. Those who heeded it fared better than those who ignored it.
Singapore has built a ‘digital twin’ of the entire city-state: a living data model that captures everything from traffic patterns to building energy use to population flows. This infrastructure isn’t just for today’s planners; it’s designed to accumulate data across decades, enabling future generations to understand how the city evolved and to model interventions before implementing them. The investment only makes sense on a long horizon. It represents a new form of institutional memory: not archives of documents but living models that grow smarter over time.
Consider the rechalking of the Uffington White Horse in England, a 3,000-year-old figure carved into a chalk hillside. It survives only because local people have continually maintained it, weeded and refilled the chalk, on at least an annual basis for centuries. This ritual of maintenance, passed down through generations, has preserved something of beauty from the Bronze Age into the internet age. The act of remembering is itself a form of devotion to the future.
Modern memory infrastructure includes the Internet Archive’s Wayback Machine capturing web pages, national archives of electronic records, and the Svalbard Global Seed Vault, a secure facility in the Arctic holding seeds of millions of plant varieties, built to last a thousand years. Teams dealing with nuclear waste storage have pondered how to mark sites with warnings intelligible to future humans who might not speak any current language. Proposals range from obvious symbols to a hypothetical “atomic priesthood” that keeps the danger alive through ritual and lore.
Memory also lives in language through proverbs and rituals. “A society grows great when old men plant trees whose shade they know they shall never sit in”: a Greek proverb praising long-term altruism. Annual days of remembrance renew commitment to “never forget.” These are miniature time-capsules carrying forward principles that proved valuable.
Strengthening memory entails preserving knowledge carefully, teaching history insightfully, honouring lessons learned, and keeping our link between past and future intact. When memory is strong, progress becomes accumulative rather than cyclical. We can stand on the shoulders of giants rather than perpetually chasing our tails.
The Responsibility of Our Time
We are now a species wired together in real time, armed with technologies capable of planetary impact, and standing at the brink of transformations that could redefine life, intelligence, and matter itself. The convergence of fields (AI reaching for general cognitive abilities, biotechnology editing the code of life, quantum physics unlocking new computation, neuroscience mapping the mind) presents a once-in-millennium opportunity to elevate humanity. It also presents grave risks of misstep.
The stakes of our decisions in the coming decades are colossal. We could cure diseases and alleviate poverty, or entrench inequalities and create new dangers. We could foster a sustainable planet, or destabilise Earth’s climate irreversibly. We could coexist with intelligent machines to augment human flourishing, or lose control of creations that surpass us. We hold the future of civilisation in our hands as never before.
What responsibilities follow? First, stewardship: to pass on a world at least as capable and healthy as we found it. This spans environmental stewardship (not consuming resources in ways that leave future generations deprived) and cultural stewardship (leaving not only our knowledge but also a functioning society). We are the first generation to fully understand the mechanics of global warming and possibly the last to prevent its worst effects.
Second, restraint and foresight with emerging technologies. We are unleashing powerful forces: AI that could automate decision-making at massive scales, gene editing that can alter the human germline. We owe it to the future to ensure these are developed safely and inclusively. The CRISPR babies incident in 2018 was broadly condemned. The scientific community enforced ethical memory, drawing on lessons from premature forays in the past. Now global dialogues on gene editing governance aim to set norms protecting future humans from unintended harm.
Third, preservation and transmission of hard-won knowledge. It would be tragic if, through negligence or disaster, the archives of human creativity and understanding were lost. Maintaining strong educational systems, countering misinformation, and supporting free inquiry are responsibilities we bear to ensure future generations inherit revolutionary values rather than fragmented echo chambers.
We are also the first generation to produce data at civilisational scale. More information is created each day than existed in all human history before the digital age. This presents both opportunity and responsibility. New technologies, particularly AI, can find patterns across this ocean of data that no human could perceive: connections between diseases, climate signals, economic trends, and scientific findings scattered across millions of papers. Properly curated and preserved, this data could serve as a resource for centuries, enabling future generations to understand our era and build upon our knowledge. But data decays. Formats become obsolete. Platforms disappear. Without deliberate investment in data stewardship, our age of information abundance could become an age of information loss.
Finally, there is a psychological dimension: to inspire hope rather than despair and that being a good ancestor is profound and attainable.
What I’ve Learned About Leading Long-Horizon Work
Over two decades of building at the edge of what’s possible, from research labs to frontier science to AI transformation, I’ve learned a few things about what it takes to invest in and lead long-horizon work…
Invention emerges from structured collisions. Transformative ideas emerge at the convergence of disciplines, perspectives, and systems. We need to design and operate for these, not wait for them to happen by accident.
People, not systems, build breakthroughs. Trust and responsibility turn ideas into outcomes. The first question to ask talented people is not “what are your deliverables?” but “what problems, ideas, or opportunities make your heart race?”
Strategic signals shape momentum. Patents and publications are not the goal, but they matter. They shape timing, attract talent, and send clear signals that an idea is worth taking seriously, inviting others to build on it.
Rethink the infrastructure of invention. We are past symbolic moonshots. Focus on orbital redesign: architectures that adapt, evolve, and make room for systemic reinvention.
Aim far and move fast to learn. Some ideas need decades to mature. Bridging temporal horizons requires pairing long-term ambition with near-term experimentation. Design layered risk architectures that tolerate failure and enable cumulative learning across timeframes, teams, and domains.
Turning these into reality requires deliberate design
Say yes to outliers. Make it the operational default. Build a culture of autonomy, trust, and excellence by curating high-agency teams who combine deep technical insight with systems-level thinking. Establish peer-led programme boards that enable research to work differently.
Facilitate layered bets. Create a multi-horizon portfolio: generational commitments, platform-level investments, and near-term provocations. Build talent scaffolding that enables commercialisation along the way, even when the whole answer is decades away. The component parts of innovation often have value before the full vision is realised. These layers generate pull for scientists, founders, and investors, building momentum through a rhythm of experimentation, learning, and acceleration. Reinforce accountability through regular forums that demo progress at each horizon.
Embed dual rhythms of invention and execution. Groundbreaking outcomes require both. Create a culture defined by operational excellence, low-friction execution, and trust with accountability. High-agency teams need transparent milestones and rituals that reward learning, not just output.
Build reciprocal partnerships. Position long-horizon institutions as catalytic nodes for invention and collaboration. Structure partnerships with feedback loops: horizon advisory groups and sparring partners who challenge assumptions and extend reach.
Redesign funding for the 21st century. Funding architectures should be adaptive, learning-centred, and velocity-aware. Prototype microgrants for rapid iteration alongside horizon-specific funds for deep science. Trial metrics like ‘idea velocity’ to track momentum across levels of uncertainty. Reward risk-informed learning, not just risk avoidance.
Closing
The Architecture of Wonder is a nod to human potential. It reminds us that we are the species that remembers, the species that dreams, and the species that builds. We have inherited the accumulated hopes and knowledge of those before us. What we build next will be the legacy we leave to those after us. Each generation lays some bricks, erects some scaffolds, perhaps raises a spire or opens a gateway, but the cathedral of civilisation and new forms of humanity across different worlds is an intergenerational masterpiece that we need to be thoughtful about. Who goes to Mars first will dictate what future will be created there.
Today, our generation works with materials our ancestors could scarcely imagine: artificial minds that learn and create, genetic scripts that compose life, instruments that peer to the edge of the universe. But our brains have been rewired for immediacy, not for the long game these tools and opportunities demand. Progress is not automatic; it requires people who can see past the next quarter, the next election, the next funding cycle, and trace the downstream consequences of what we build today.
The longevity agenda offers one lever. If we expect to live longer, we have more reason to care about the world we’re creating. That perspective shift needs to reshape how we govern, how we invest, how we educate. But longevity research itself is a cautionary tale: it remains predominantly funded around men’s health, while the women’s health agenda, though gaining traction, has yet to be applied to longevity. If we create a world where one gender lives longer than another, we will have used long-horizon thinking to entrench imbalance rather than transcend it. The tools are not enough; how it impacts the long term matters as well.
We have to embrace long-horizon thinking and doing. When we plan a city, we plan for a century hence. When we design a technology, we ask not just “what can it do?” but “what world will this create five decades from now?” We nurture institutions that learn and last. We tell stories that celebrate the arc of history and the promise of the future.
The four pillars (Patience, Imagination, Institutions, Memory) support each other. Memory informs active imagination with rich source material. Imagination inspires institutions with vision. Institutions enable patience with structure and stability. Patience allows knowledge to accumulate into memory. Together, they form a stable foundation upon which long-term aspirations can be built and maintained.
From Pillars to Practice
The four pillars have operational implications. Domains where long-horizon thinking is not optional have evolved practices that embody these pillars: defence research, intelligence analysis, space exploration, and the national laboratories that steward nuclear and quantum capabilities. These are not perfect institutions. But they have solved problems of sustained innovation and in doing so have enabled new startup ecosystems that are now able to translate these technologies into the mainstream.
Five practices recur across these domains:
Treat Memory as Infrastructure. The essay describes how Roman concrete vanished, how digital formats decay, how the Uffington White Horse survives only through continuous rechalking. But memory loss is not just about artefacts. It is about interpretive capacity. When India conducted nuclear tests in 1998, the US intelligence community was caught off guard, not because the satellites failed, but because the analysts who understood India’s nuclear signalling had retired without transferring their knowledge. The documents remained; the ability to read them did not. Build mentorship, documentation, and redundancy before the crisis reveals the gap.
Close the Loop Between Horizons. Most organisations treat innovation as a pipeline: blue-sky research eventually matures into products. This 3 horizon mental model encourages neglect. Long-term research becomes a luxury, first to be cut when budgets tighten. Bell Labs worked differently: fundamental science and applied engineering existed in the same corridors, informing each other. The better model is a cycle: near-term operations generate value and learning; that value funds the next generation of seeds; research reshapes what operations become possible.
Build Scaffolding Before Adding Tools. Patience is the willingness to invest time in what matters. But patience is not only about waiting. It is about building the human systems that allow capability to be used wisely. When the US Air Force fielded persistent drone surveillance, the technology worked brilliantly, but operators experienced burnout and moral injury because the human systems (crew ratios, shift patterns, psychological support) had not been designed with the same rigour as the sensors. Technology scales; human cognition does not.
Be an Architect, Not a Tourist. The essay distinguishes between those who dream and those who build. Innovation culture celebrates the breakthrough moment: the demo, the prototype, the proof of concept. But breakthroughs are a small fraction of the work. Stealth aircraft took fifteen years from concept to combat deployment. GPS took four decades of satellites, receivers, doctrine, and integration. The cathedrals invoked throughout this essay were not built by people who sketched a spire and moved on. They were built by generations who laid foundations, raised walls, and solved problems their predecessors could not have foreseen.
Demand Stewardship, Not Spectacle. Powerful technologies create risks alongside benefits. The temptation is to move fast and address governance later. But by then, path dependencies have hardened and public trust may already be damaged. The CRISPR incident, mentioned above, showed what happens when capability races ahead of constraint. Toronto’s Quayside smart city project collapsed not because the technology failed, but because questions of data governance were only considered after the fact.
The future is not read as science fiction; it is built. Future generations will live within the architecture we build today, so we must design it with deep time in mind.
References
Amodei, D. (2023). Machines of Loving Grace: How AI Could Transform the World for the Better. [Online]. Available at: https://www.darioamodei.com/essay/machines-of-loving-grace (Accessed 31 August 2025).
Brand, S. (1999). The Clock of the Long Now: Time and Responsibility. New York: Basic Books.
Carr, N. (2010). The Shallows: What the Internet Is Doing to Our Brains. New York: W. W. Norton.
Gertner, J. (2012). The Idea Factory: Bell Labs and the Great Age of American Innovation. New York: Penguin Press.
Krznaric, R. (2020). The Good Ancestor: How to Think Long Term in a Short-Term World. London: Headline Publishing Group.
Maffeo, L. (2015). ‘Google’s Vint Cerf on how to prevent a digital dark age’. The Guardian, 29 May 2015.
Ord, T. (2020). The Precipice: Existential Risk and the Future of Humanity. London: Bloomsbury Publishing.
Roser, C. (2017). ‘Monozukuri – Japanese Work Ethics’. AllAboutLean.com, 12 Dec 2017. Available at: https://www.allaboutlean.com/monozukuri/ (Accessed 31 August 2025).
The Economist (2021). “The agency that shaped the modern world.” The Economist, 439(9248), 5 June 2021, pp. 67–68.
This essay was written by me and refined through sparring with various LLMs