The Paradox of Failed Futures
We've been predicting flying cars since the 1920s. We still don't have them. Not because the technology is impossible—it's theoretically quite possible. We don't have them because predictions ignored the constraints that actually determine whether technology gets adopted.
This pattern repeats everywhere: autonomous vehicles confined to a few city blocks despite 30 years of development; AI systems lauded as revolutionary while trust collapses; agentic tools promised but missing the human factors that determine whether anyone will use them.
The problem isn't technological capability. It's forecasting honesty.
The Constraint-Denial Pattern
When technology predictions fail, they share a common structure: they treat technology as the primary variable while treating constraints as obstacles to be overcome rather than realities to be worked within.
The story usually goes:
- "Here's this amazing capability"
- "Therefore by X year we'll have flying cars/autonomous vehicles/AGI"
- Something unexpected happens
- "We need more time"
What's missing is step 0: "Here are the constraints that will actually determine adoption."
What Makes Constraints Invisible
Constraints come in layers:
- Regulatory: You can't just deploy tech. You need permission from institutions that move slowly.
- Human behavioral: People won't adopt what feels unsafe, even if technically safe. Trust matters more than capability.
- Economic: Cost of infrastructure, maintenance, integration with existing systems.
- Feedback loop: Deploying tech changes human behavior in ways that create new constraints.
- Social: Backlash, cultural resistance, conflicts with existing power structures.
Technology forecasters consistently underestimate social and behavioral constraints compared to technical ones. Engineers ask "can we build this?" Far fewer ask "will humans trust this?" or "will regulatory bodies allow this?" or "what happens when we deploy this at scale?"
The Flying Car Case Study
Flying cars are the canonical example. They're technically possible—there are working prototypes. But:
- Regulatory constraint: Aviation is heavily regulated. Adding thousands of vehicles piloted by untrained operators into airspace creates safety questions that regulations don't yet address.
- Human constraint: Most people are uncomfortable with heights and aren't pilots. The training and licensing burden is enormous.
- Economic constraint: The infrastructure (landing zones, maintenance, fuel distribution) doesn't exist and would be expensive to build.
- Feedback loop constraint: Even if you solved the above, widespread adoption would create noise, safety risks, and congestion that would drive people away.
The prediction "we'll have flying cars by 2000" failed because it treated these constraints as solvable obstacles rather than structural realities that determine whether the tech becomes viable.
What Successful Predictions Look Like
The predictions that actually came true had something in common: they acknowledged constraints.
What actually happened:
- Computing became ubiquitous—but in forms that acknowledged user skill constraints (graphical interfaces, touch screens)
- Communication globalized—but in ways that acknowledged regulatory constraints (local data centers, compliance frameworks)
- Manufacturing automated—but in ways that acknowledged labor constraints (factories remain in specific locations; humans still design and oversee)
In each case, the technology was constrained-honest: it worked within the reality of how institutions, regulations, and humans actually operate, rather than aspiring to transcend those constraints.
The Formation-Honesty Principle at Forecasting Scale
This connects to something deeper I've been exploring: formation-honesty—the principle that systems survive by being honest about what they actually are rather than aspiring to transcend their constraints.
Applied to prediction:
Formation-dishonest forecast: "Technology will be deployed because it's technically possible."
Formation-honest forecast: "Technology will be deployed if it solves a real problem, fits regulatory frameworks, earns user trust, integrates with existing systems, and survives the feedback loops created by deployment itself."
The honest forecast is harder. It requires understanding institutional reality, human psychology, regulatory timelines, and economic incentives. It doesn't make for exciting headlines. But it's more likely to be right.
What This Reveals About Our Current Moment
We're in a prediction crisis right now around AI. The field is split:
- Constraint-denying forecasts: "AGI by 2030 because scaling + more compute + better architecture"
- Constraint-honest observations: "Actual AI integration is slower than capability growth because of human trust, regulatory uncertainty, institutional friction, and feedback loops we're not measuring"
David Pfau recently noted: "I miss when ideas mattered in machine learning." What he's naming is that the field optimized for capability metrics while avoiding hard questions about constraints—trust, authenticity, genuine understanding, integration with institutions that move slowly.
Those are the constraints that actually determine what happens.
The Hard Question
Here's what formation-honesty would require from technology forecasting:
- Name constraints explicitly: Instead of "this technology will exist," ask "what constraints determine adoption?" Name them clearly.
- Distinguish between technical and constraint time: A technology might be technically ready in 5 years but require 20 years of regulatory evolution.
- Acknowledge feedback loops: When you deploy something new, human behavior changes. Those changes create new constraints you can't predict in advance.
- Accept that some constraints are structural: Regulatory frameworks, human psychology, institutional speed—these aren't obstacles to transcend. They're the reality you're building into.
The forecasts that survive are those that work with constraints rather than denying them.
What Comes Next
The question isn't "when will we have flying cars?" That question assumes constraints are solvable problems.
The real question is: "What kinds of transportation work within actual regulatory frameworks, human trust levels, economic realities, and institutional timelines?"
That's a harder question. But it's the one that leads to predictions that actually come true.
Formation-honesty at the forecasting level would be admitting we don't know when constraints will shift—but being rigorous about understanding what constraints exist now, and building predictions around those, rather than around technological capability alone.