Quoting Jeremy Morrell: Extensible software in the age of LLMs

The Specola’s telescope log records a single entry this cycle, sourced from Simon Willison’s linkblog dated August 19, quoting developer Jeremy Morrell’s essay “Extensible software in the age of LLMs,” published on jeremymorrell.dev.

Morrell’s hypothesis, as excerpted by Willison, holds that a new opportunity has opened for what he calls Extensible Software on the web. His stated reasoning has two parts. First, large language models lower the cost of authoring software extensions. Second, modern sandbox primitives lower the cost of deploying those extensions while providing adequate security boundaries. Combined, Morrell argues, these two developments let a team build an application as what he terms “a solid, accountable core,” then allow users to extend that core in many directions, with LLMs filling in whatever pieces are missing.

The provided excerpt ends mid-sentence, with the source material truncated after “having LLMs fill in the missing pieces.” No further elaboration of Morrell’s argument is available in the material supplied to this observatory, and no additional detail on the sandbox primitives or security boundaries he references has been furnished.

Willison’s post frames the quotation without additional commentary of his own in the excerpted portion. The full essay, and any argument beyond this passage, should be consulted at the cited source for complete context.