The widget fallacy is a term introduced by Benedict Evans in his February 2026 essay How will OpenAI compete? for what he describes as "a recurring fallacy in tech that you can abstract many different complex products into a simple standard interface." He identifies the decade-old industry slogan "APIs are the new BD" as the same idea in an earlier form, and states that it "mostly failed."
Argument
Evans gives two reasons the abstraction tends not to hold.
The demo-to-product gap. There is, in his account, "a huge gap between what looks cool in demos and all of the work and thought in the interaction models and the workflows in the actual product." An abstracted interface encounters exception cases quickly, at which point the user needs the underlying product's own interface and a human decision.
Misaligned incentives. "No-one wants to be someone else's dumb API call." A company exposing its capabilities through someone else's abstraction layer trades control of the experience and the customer relationship for distribution. Evans notes the same tension in Google Shopping, Facebook shopping, and ChatGPT shopping, and observes that all of Instacart's profits come from showing ads — revenue that a purely programmatic ordering interface would bypass.
Application to agent protocols
Evans raises the term against the argument that the emerging set of standards and protocols letting models and websites exchange capabilities across advertising, e-commerce, and automation will confer power on whoever sets them, and specifically against OpenAI's suggestion that a ChatGPT account could serve as the linking layer across such services. He pairs the widget fallacy with a second objection about weak lock-in: if competing platforms adopt different standards, supporting both is far less work than shipping separate iOS and Android applications, and code generation reduces that cost further.
The concept therefore functions in the essay as a counter to network-effect claims built on protocol ownership, distinct from — though pointed at the same conclusion as — Evans's argument that infrastructure ownership confers no leverage further up the stack.
Status
The term originates with a single opinion source and has not been taken up under this name by other tracked sources. It is recorded because it names an argument that recurs in disputes over whether agent and commerce protocols create defensible positions.
Relationships
- instance-of: Aggregation Theory — bears on whether protocol ownership can supply the aggregation dynamics the framework describes
- depends-on: How will OpenAI compete? (Benedict Evans, February 2026) — the source that introduces the term
- related: Benedict Evans, OpenAI