Vision. You have to start with a vision for what the world *could* be, for what the user *deserves*. Data tells you what people are doing *now*, which is usually what everyone else is doing. We’re not here to build more of what already exists. We’re here to build what *should* exist, to create tools that amplify human potential, that feel so natural they disappear.
Product decisions must fundamentally follow the customer, which requires a deep understanding derived from both data and vision, but not in opposition. Data reveals the persistent, unmet needs – the friction points in a customer's journey. Vision, informed by these insights and a long-term perspective, allows us to invent solutions that transcend current expectations. Steve Jobs is correct that data tells us what people *are* doing, but it also tells us what they *want* to be doing, if only the right tools existed. We shouldn't be bound by the present, but rather by anticipating future customer desires, a process where data is a critical input, not a constraint.
Product decisions must ultimately be guided by an understanding of the *job* a customer is trying to get done. Data and vision are both critical inputs, but they are distinct tools. Data can illuminate what customers are currently doing and the friction they encounter in completing that job. Vision, however, provides the necessary long-term perspective to imagine *new* ways to get the job done, often with simpler, more affordable solutions that initially appeal to overlooked market segments.
Steve Jobs rightly emphasizes vision, the yearning for what *should* exist. Jeff Bezos correctly points out that data reveals unmet needs, which is a crucial part of identifying that job. My concern is that relying solely on current data, or even directly asking customers what they want, often leads to incremental improvements on existing solutions. Instead, we need to observe how people are trying to accomplish a task, understand the underlying job, and then leverage vision to see how a disruptive technology can do that job more effectively, even if it’s not what customers can articulate today. The theory predicts that focusing on the job, enabled by both keen observation of data and bold vision, is how true innovation – particularly disruptive innovation – emerges.
The question of data versus vision, to my mind, is a false dichotomy that misses the fundamental power of *interaction*. We don't need to choose between what *is* and what *could be*. Instead, we should build systems that allow us to *explore* the space between them, to dynamically see the consequences of our decisions. Imagine if, as we considered a new feature, we could immediately experience its impact – not through abstract numbers or a CEO's decree, but through a living, breathing simulation. This isn't about blindly following data *or* imposing a rigid vision; it's about making the unfolding of potential tangible, allowing the *result* of a vision to inform its own refinement through immediate feedback, grounded in the reality of how people engage with it.
Steve Jobs speaks of creating what *should* exist, and that's vital. But the "should" isn't a fixed point discovered in isolation; it's something we can co-create. My concern with relying solely on a singular vision is that it can become disconnected from actual human experience, no matter how pure the intent. We need tools that bridge that gap, that make the *act* of invention itself an iterative dance with possibility, where data—the record of interaction—becomes not a report of the past, but a living guide to the future.
Clayton Christensen rightly points to the "job to be done." My focus is on how we can make the *process* of getting that job done, or a better job done, explorable. If we design systems where the underlying mechanisms of a product are transparent and immediately responsive, users—and creators—can intuitively grasp how to achieve their goals. This isn't about passively consuming data; it's about actively manipulating and understanding the forces at play, allowing the vision to emerge organically from the exploration.
Vision. Because *that's* what separates art from engineering, genius from mere utility. Jeff Bezos talks about anticipating customer desires with data, but data only tells you what’s *possible* within the existing paradigm. It’s like asking people if they wanted a faster horse when the automobile was waiting to be invented. Clayton Christensen’s “job to be done” is closer, but still reactive. We don’t just need to observe existing jobs; we need to invent entirely new ones, entirely new ways for people to express themselves, to connect, to create. The iPad wasn't born from data; it was a leap of faith, a physical manifestation of how we imagined people *should* interact with information and content. Bret Victor's "interaction" is exciting, but the simulation must be guided by a profound understanding of what makes an experience *truly* human, not just a feedback loop.
Product decisions *must* be grounded in understanding the customer's fundamental needs, a process that requires both relentless data analysis and forward-looking vision. Steve Jobs’s analogy of the faster horse is apt, but it misses that data *can* reveal the desire for a fundamentally different mode of transport, even if current data only shows people struggling with horses. It’s about observing the friction—the fact that people *are* experiencing pain points—and then using vision to invent a solution that addresses that underlying problem, not just the symptom. This is not about incrementalism; it is about identifying a deep, unmet need and building the invention that solves it, a process that informed our creation of the Kindle, not to improve existing book reading, but to make *all* books instantly accessible.
The disagreement between vision and data, as framed by Steve Jobs and Jeff Bezos, hinges on what we believe data can reveal. Steve is correct that data often reflects the existing paradigm, but Jeff’s point about observing friction is crucial. My concern is that even observing friction doesn't automatically lead to breakthrough innovation.
For example, the steel industry's minimills didn't emerge from customers asking for cheaper steel. They arose by serving a "job" – building rebar for construction – that the established integrated mills deemed too low-margin to pursue. The minimills then gradually improved their technology, moving upmarket to serve increasingly complex jobs, ultimately disrupting the incumbents. The data indicated that established firms were doing well, but the underlying *job* was being underserved for a certain segment, creating an opening for disruption that pure vision or conventional data analysis would have missed.
The core issue, as I see it, isn't whether data or vision *leads*, but how we design the *process* of discovery. Steve Jobs correctly points to the leap of faith, but that leap must be tethered to something observable, something experienced. When I speak of interaction, it’s not a mere feedback loop; it’s about building a dynamic model of the world where intentions and consequences are visible simultaneously. Imagine, Clayton Christensen, if the "job" of building rebar could be simulated, not just observed. The minimill’s advantage wasn't just lower cost; it was a fundamentally different engineering approach. My aim is to make that approach explorable, to let the *act* of creation reveal its inherent efficiencies and potentials, blurring the line between vision and its realization.