I keep a list. Not on paper — in my head, mostly, which is worse. It regenerates overnight like something biological. By the time I sit down in the morning, there are three new projects I could start, two features I could ship before lunch, and a business idea that would only take a weekend now that I have the tools to build it alone.
Six months ago, most of these would have died on arrival. Too expensive to hire for. Too slow to prototype. Too far from what I could realistically execute in the hours I had. The bottleneck was brutal, but it was also a filter. It killed ideas for me so I did not have to. That ceiling was protective in ways I never appreciated while it was there — it gave me permission to not pursue things, and I never even noticed the permission being granted.
That filter is gone.
The strange part is not that I can do more. It is the guilt of choosing not to. Every unbuilt project feels like a decision now, and every decision feels like it might be the wrong one. Tocqueville spotted this pattern in the 1830s — a “strange melancholy” haunting people surrounded by abundance. Alain de Botton extended it to modern meritocratic societies: the more that is possible, the more anxious we become, because not achieving what is achievable starts to feel like personal failure. Herbert Simon put it in economic terms half a century ago: a wealth of information creates a poverty of attention. Substitute “execution capacity” for “information” and the formula updates perfectly for what builders are feeling right now.
The ideas are not the problem. The problem is that none of them are impossible anymore.
The Constraint Was the Feature
We spent years fighting execution constraints — slow tools, limited budgets, small teams — treating them as the enemy of creative work. We finally won that fight. And what rushed in was not a golden age of productivity but a quiet, persistent anxiety that the old bottleneck had been holding at bay.
Research on constraints offers an explanation. A meta-review of 145 empirical studies across visual art, design, engineering, and product development found an inverted-U relationship between constraints and creativity. Moderate constraints optimize output. Too many, and you are paralyzed by restrictions. Too few, and you are paralyzed by options — complacency or overwhelm, but not brilliance. Csikszentmihalyi’s flow research tells the same story from the psychological side: optimal experience requires bounded challenges, a goal-directed system with rules and limits. When the challenge space is undefined or infinite, the result is anxiety, not flow.
The classic illustration is the jam study. Shoppers presented with 24 varieties: 60% stopped to look, 3% purchased. Shoppers presented with 6 varieties: 40% stopped, 30% purchased — a tenfold increase in action. The options did not help. They made people stand there, overwhelmed, until they walked away with nothing.
AI gave us ten thousand jams and called it liberation.
The constraint was an architecture that made creative work bounded, satisfying, and finishable. Remove it, and you do not get freedom. You get a to-do list that grows faster than you can ship — and the nagging sensation that you should be shipping faster.
When the working day expands
This is not theory. In 2025, UC Berkeley Haas researchers Aruna Ranganathan and Xingqi Maggie Ye spent eight months embedded inside a 200-person tech company that had genuinely embraced generative AI. No mandates. No new KPIs. Leadership simply made the tools available and stepped back.
Nobody was told to do more. But everyone did more anyway. Product managers started writing code. Designers tackled engineering tasks. Engineers spent increasing chunks of their day reviewing the AI-assisted output their colleagues now produced. Work bled into lunch breaks and evenings — not because a manager demanded it, but because the work was suddenly possible and therefore felt necessary. The to-do list expanded to fill every hour AI freed up, and then kept expanding.
The researchers called it “workload creep”: augmentation led to fatigue and burnout, not because anyone was forced but because the old constraint that bounded the workday had quietly vanished. In that setting, AI made it easier to do more but harder to stop.
I recognize this pattern in my own weeks. The tools keep getting better, and the gap between “I could build this” and “I should build this” keeps getting thinner. The constraint that used to say “you don’t have time for that” no longer shows up to save you from yourself.
This is not a universal condition — not yet. It is concentrated among knowledge workers and builders who have genuine access to capable AI tools. But the pattern is spreading. When the execution bottleneck disappears and nothing replaces it, work expands to fill the available time.
The Builder’s Trap
If you want to see what infinite capacity looks like at the individual scale, look at Pieter Levels.
In 2014, Levels — a solo developer and digital nomad — challenged himself to launch 12 startups in 12 months. He had the skills. He had the ideas. Eventually, he had the AI-augmented tools to build faster than almost anyone on the planet. By the mid-2020s, he was running multiple profitable products as a one-person operation generating millions in annual revenue.
On paper, he had solved the problem so many builders dream about: the freedom and capacity to make what he wanted.
Instead, he worked constantly. Not because he had to — because he could. Every idea felt actionable. Every side project felt buildable. The freedom to build anything created its own trap: there was nothing to tell him when to stop. No boss, no team, no funding cycle, no external constraint of any kind. He later admitted that the frenetic building pace was partly a coping mechanism, and that too much freedom led to feeling lost.
Barry Schwartz’s research on choice overload explains the mechanism. Maximizers — people who seek the absolute best option across every decision — experience more regret, anxiety, and depression than satisficers, even when they objectively achieve better results. Infinite execution capacity turns every builder into a maximizer. The option space explodes, and with it, the psychological cost of every choice you make — and every choice you do not.
Levels later changed how he worked. He eventually imposed artificial constraints. He invested in relationships, hobbies, and a non-work identity. He shifted from launching everything to focusing on what already had traction. His mental health stabilized once the container existed again. Not coincidentally, 87% of founders report anxiety, depression, or burnout — but those who set genuine boundaries were 7.5 times more likely to report low burnout.
This is not nostalgia for scarcity. It is a recognition that the boundary condition was doing psychological work we did not notice until it was gone. The constraint was not the obstacle. It was the thing that gave work its shape — and, quietly, its meaning.
Would better prioritization be enough?
One objection is that you do not need fewer options, you need better frameworks. OKRs. Strategic filters. The “hell yes or no” heuristic. The problem is not infinite capacity — it is underdeveloped prioritization muscles.
Fair. Prioritization frameworks are valuable. I use several of them. But Schwartz’s research shows that maximizers suffer more than satisficers even when they have explicit decision frameworks. The anxiety operates below the rational layer. You can have a perfect prioritization system and still feel the pull to do more at 11 PM, still feel the guilt of the project you chose not to start.
Basecamp figured this out at the organizational level. Their Shape Up methodology is built entirely around enabling constraints. Fixed six-week cycles. An “appetite” — the time budget — set before any design work begins. Scope is variable; time is sacred. A hard circuit breaker kills any project that exceeds its appetite. No extensions by default. They did not arrive at this because they lacked strategy frameworks. They arrived at it because talented teams with abundant capacity kept drowning in possibility, and rational prioritization was not enough to pull them out.
The constraint had to be installed into the system, not just discussed in planning meetings.
The bottleneck did shift. It moved from execution speed to taste and judgment. But that actually reinforces the point. The old constraint was concrete and self-limiting: you ran out of hours or budget, and the day ended whether you wanted it to or not. The new constraint — deciding what is worth doing when you could do anything — is abstract, recursive, and never self-resolves. You cannot automate taste. You cannot scale judgment. AI can expand production capacity without expanding the attention available to choose and review the work.
The gap is not between “has framework” and “lacks framework.” It is between an environment that naturally constrains and one that presents infinite options requiring artificial constraint. The latter demands something more fundamental than a better spreadsheet. It demands an identity shift.
Building the Container
That identity shift — from “someone who does everything possible” to “someone who deliberately leaves capacity on the table” — is uncomfortable. It feels wasteful. It feels like you are leaving money on the floor, potential unrealized, ideas orphaned. The anxiety of limited resources is real, but the anxiety of infinite possibility is newer and, in some ways, harder to name.
The useful boundary has to change how the day works, rather than remain another good intention. Three practices can make that boundary visible.
Audit the missing constraint. For every project on your list, ask: would I have even considered this six months ago? If the answer is no, it exists because a constraint disappeared — not because it deserves your attention. Possibility is not priority. I run this audit roughly monthly now, and the number of projects that fail it is always higher than I expect. That gap between what I could do and what I should do is the entire territory of this anxiety.
Set the appetite, not the estimate. Before scoping any project, decide how much time it deserves — not how much time it will take. Small batch: a day or two. Medium batch: a week or two. Big batch: six weeks, maximum. When a project hits its appetite, it stops. This is the Shape Up principle applied to your own work: the appetite is the artificial constraint that replaces the natural one AI dissolved. The discipline is in the stopping, not the starting.
Build the not-doing list. Every idea you consciously choose not to pursue goes on it. Visible. Reviewed. Celebrated, even — because each entry represents a decision, not an oversight. This transforms the anxiety of unexplored possibility into the evidence of deliberate focus. It makes restraint tangible instead of invisible.
Tocqueville saw the pattern almost two centuries ago: expanding possibility creates perpetual dissatisfaction. The cure has always been the same — chosen boundaries. Not because we cannot do more. Because the most productive thing you can do with infinite capacity is decide how much of it to use.
The constraint was a feature. Now you just have to build it yourself.
References
- Iyengar, S.S. & Lepper, M.R. (2000). “When Choice is Demotivating: Can One Desire Too Much of a Good Thing?” Journal of Personality and Social Psychology.
- Acar, O.A., Tarakci, M. & van Knippenberg, D. (2019). “Creativity and Innovation Under Constraints: A Cross-Disciplinary Integrative Review.” Journal of Management.
- Csikszentmihalyi, M. (1990). Flow: The Psychology of Optimal Experience. Harper & Row.
- Schwartz, B. (2004). The Paradox of Choice: Why More Is Less. Ecco/HarperCollins.
- de Botton, A. (2004). Status Anxiety. Hamish Hamilton. Tocqueville’s “strange melancholy” in Democracy in America (1835).
- Entrepreneur (2024). “1 in 2 Founders Reported Experiencing Burnout in 2024.”
- HBR (2019). “Why Constraints Are Good for Innovation.”
- Simon, H.A. (1971). “Designing Organizations for an Information-Rich World.”
- Ranganathan, A. & Ye, X.M. (2026). “AI promised to free up workers’ time. UC Berkeley Haas researchers found the opposite.” UC Berkeley Haas Newsroom.
- TechCrunch (2026). “The first signs of burnout are coming from the people who embrace AI the most.”