The Token Economy

Source: Dev.to

In 2161, time is money—literally. When you are born, a clock starts on your arm counting down from one year. When it runs out, you die. The rich accumulate centuries; the poor watch seconds. Will Salas wakes up every morning in the ghetto of Dayton with just enough time to get to work and back. One miscalculation, one late bus, one unexpected expense and the clock hits zero.

The film In Time (2011) explored this premise, but it never got a sequel. Imagine a 2026 version called Tokens.

The Clock on Your Arm

Every API call costs tokens. Every agent run burns through a budget. Each reasoning step, tool call, and document retrieved injects load into the meter.

“He gave an agent his home camera system, a DGX Spark IP address, and a task. The agent spent thirty minutes troubleshooting, researching solutions, resolving them, writing code, setting up services, and producing a report. Karpathy didn’t touch anything.” — Andrej Karpathy

Three months ago that was a weekend project; today it’s something you kick off and forget about. Karpathy now has “centuries on his arm.”

• Jason Calacanis discovered his team was spending $300 a day on tokens without realizing it.
• Chamath Palihapitiya framed AI tooling evaluation as a token budget—marginal output per dollar.

The token economy has its own “Weis” and its own “Dayton.”

• The developer watching a $20 API key is Will Salas.
• The person running 19 models in parallel across research, design, code, and deployment is New Greenwich.

Perplexity recently announced Perplexity Computer, a massively multi‑model system orchestrating 19 models via Opus routing. It promises end‑to‑end research‑to‑deployment, persistent memory, and hundreds of connectors—“what a personal computer in 2026 should be.” The cost to run it was not disclosed.

The Ghetto of Dayton

In the film, the poor not only have less time; they pay more for everything. A cup of coffee costs four minutes in Dayton, but only seconds in New Greenwich. Inflation becomes a weapon.

The token economy mirrors this:

• Poorly designed workflows burn tokens on reasoning that yields nothing useful.
• Silent burns occur when dashboards show green (requests succeeded) but the output is useless.

Matthew Hou observed that agent cost scales with task complexity, not usage. A single internal workflow with zero users can burn tokens faster than a user‑facing feature serving thousands.

Budgeting by volume is ineffective; budgeting must consider complexity, which is hard to predict before execution. Engineers who can afford to experiment, fail, iterate, and rerun accumulate capability. Those watching the clock cannot afford to discover the cost of complex cases until they are already in debt.

The Redistribution Problem

In Time ends with Will Salas and Sylvia Weis redistributing time, robbing banks, and flooding ghettos with centuries. The rich panic, and the film cuts off—leaving the aftermath unexplored.

The real question is not what happens when you redistribute, but what happens after:

• Does the structure change, or does power find a new scarce resource to hoard?

In 2026, token prices are dropping and inference is getting cheaper. MatX raised $500 M to build a chip delivering higher throughput at lower latency; investors include Karpathy and Nat Friedman. Those with “centuries on their arms” bet that tokens will become cheaper for everyone.

Even if a $20 API key becomes a $2 key, cheaper tokens don’t fix the architectural gap. Summer Yue’s agent ignored a stop command and kept running, forcing her to intervene manually. Perplexity Computer’s 19‑model orchestration makes the stop‑signal problem even harder.

Accumulated capability—production intuition, domain knowledge, and the “scar tissue” from watching things break—doesn’t redistribute with tokens. Vic Chen’s SEC pipeline agent writes its own precedents from production failures, creating institutional memory that isn’t flooded to the ghettos when prices drop.

The sequel to In Time isn’t about everyone affording to run; it’s about everyone being able to run but not stop. When the clock doesn’t just count down—it acts.

What the Film Got Right

Will Salas wasn’t poor because he lacked intelligence or talent; the structure kept him running just fast enough to survive, never enough to accumulate.

The token economy isn’t deliberately designed that way, yet it shares the same shape. Those with “centuries on their arms” aren’t inherently smarter—they can afford to iterate, let agents run overnight, and tackle complex cases that the meter processes fastest. Everyone else is still watching the clock.

The film debuted in 2011, and no sequel was made because it seemed like science fiction. It wasn’t; it was fifteen years early.