Fortune Telling and Speculative Design

Open original

Maggie Appleton, a research engineer at GitHub Next, shares how her Labs team explores possible futures for software by building prototypes, and offers practical techniques for doing speculative design yourself. She argues design engineers are well placed to turn theories about where software is heading into tangible artefacts.

  • Her track record includes calling the rise of digital gardens (2020), the flood of AI-generated content on the open web (2022-23), and an explosion of home-cooked software and barefoot developers (2024).
  • GitHub Next works by posting rough prototypes daily on an internal platform, grouping team-wide weak signals on a shared board, and then pushing promising bets toward production, with Copilot, Spark, Workspaces and Agentic Workflows as past examples.
  • She frames design engineers as designers who make real decisions about state, data flow and architecture, and says agents now fill gaps when a prototype needs something the designer can't build alone.

Maggie Appleton, a Staff Research Engineer at GitHub Next (GitHub's R&D lab), gave a talk at Granola's Design Engineering night on speculative design and how design engineers can prototype possible futures of software. She argues the people best placed to do this are design engineers: designers who understand the material of software well enough to work with state, data flow, databases, APIs, auth and latency, who bridge design and engineering, and who make real decisions about functionality and architecture rather than just surface polish. Her case rests on the idea that the pace of change right now, with new models, harnesses and startups appearing constantly, makes it tempting to opt out of predicting anything. She thinks the better response is to adopt a scout's mindset, collect weak signals, form theories, and build artefacts that make those theories tangible. Agents, she notes, now fill gaps like databases or machine learning that a designer prototyping alone couldn't build. To justify her standing to make such predictions, Appleton points to three earlier calls. In 2020, during lockdown, she wrote about digital gardens, a slower, interlinked alternative to the chronological blog and the feed, a concept that peaked around 2021 and 2022. In late 2022 and early 2023, while working at Elicit, she predicted the open web would be flooded with AI-generated content, which she framed in her essay "The Expanding Dark Forest and Generative AI." In May 2024 she gave a talk on "home-cooked software" and "barefoot developers," predicting an explosion of personal software built by non-professionals for their communities, a prediction she says people dismissed at the time but which now looks like it is arriving. The practical sections describe how GitHub Next works. The team posts rough prototypes many times a day to an internal platform called Next Cloud, so that everyone can see, react to and riff on each other's experiments. Examples include Alex Gorischek's prototype that enforces a maintainer's guidelines on GitHub issues before a post is allowed, and Luke Edwards's rebuild of GitHub on Durable Objects, which explores a real-time multiplayer canvas for sketching ideas before turning them into agent work. The team also runs group theory-building sessions on a large FigJam board, where members add sticky notes on trends and directions they see in software and agentic development. Appleton's underlying point is that tangible, even silly, prototypes are the cheapest way to find out whether a direction is worth pursuing.

Ask about this article