Alan Kay's talk at UCLA – Feb 2024 [video]

Original Article ↗ Hacker News Discussion ↗

Alan Kay’s February 2024 UCLA talk prompts renewed interest in his decades-long work on Smalltalk, user interfaces, education, and the still-unrealized “computer revolution,” with commenters sharing extensive pointers to his talks, papers, and related research projects like STEPS and Dynamicland. Many emphasize his ideas about live, fully malleable computing environments—contrasting them with today’s static, app-centric systems—and lament that mainstream operating systems never adopted such paradigms. Others debate why Smalltalk and similar approaches failed to gain widespread use, how newer trends like LLMs, formal methods, and functional programming fit into his vision, and what a truly human-amplifying computing medium might look like.

Recommended talks, papers, and books

  • Multiple links to comprehensive lists of talks, papers, and reading lists were shared, including curated “menus” of Kay talks and Viewpoints Research Institute papers.
  • Specific favorites: “The Computer Revolution Hasn’t Happened Yet”, “Normal Considered Harmful”, “Back to the Future of Software Development”, talks on simplicity, and education-focused talks (including a 1995 US House hearing with Papert).
  • Highly recommended writings include the “Early History of Smalltalk”, with one commenter preferring an Ingalls paper as less self-promotional.
  • Books on PARC history were discussed; one was viewed as “the only accurate and good book”, another still considered worthwhile but somewhat confusing.

Reflections on Kay’s ideas and this talk

  • Strong positive reactions to the framing of computing as a “shared stage” for reflecting and amplifying human intelligence.
  • Several highlight his critique of schooling and human tendency toward belief over thinking; some express frustration that such ideas are poorly received in typical online forums.
  • One person notes that long-time followers may find this talk repetitive; others say it made them want to binge-watch more.

Live systems, Smalltalk, and operating systems

  • Desire for a “live all the way down” OS where everything is inspectable and editable at runtime; current mainstream systems are seen as static and closed.
  • Smalltalk, Pharo, Emacs, and various live-programming environments are cited as partial realizations of this vision.
  • Some describe real projects where Smalltalk was used down to near-hardware level; others discuss repurposing Pharo or building new live OSes.
  • There is lament over the disappearance or difficulty of accessing artifacts from Kay-related projects like STEPS and Frank, though some individuals still have running code.

Natural language, LLMs, and human–computer interaction

  • One thread argues the key missing piece for end users is a first-class language to talk to the OS; today’s GUIs are likened to animals pushing predefined buttons.
  • Counterarguments stress that natural language is ambiguous, optimized for human relationships, and poorly suited to precise, task-specific computing.
  • Others respond that newer systems (LLMs) can, in principle, absorb ambiguity and that “prompt engineering” exists precisely because of this tension.
  • There is a split between those who think LLMs could enable Kay-like dynamic, just-in-time interfaces and those who suspect he would be unimpressed or concerned.

Smalltalk, OOP, and alternative paradigms

  • One critical view: no widely used system is written in Smalltalk for “sound reasons”; object orientation is only one organizational tool among many.
  • Others reply that later Kay-led work explicitly explored non-OOP paradigms (array languages, spatially replicated computation, tuple spaces, solver communities).
  • Formal methods, functional programming, static type systems, and modern verification tools are cited as important advances that go beyond classic OOP.
  • Influence of Smalltalk is traced into other languages and tools (Objective-C, Ruby, Java-era tooling), and newer Smalltalk-like languages are mentioned.

Related tools, demos, and historical systems

  • Strong interest in the Sketchpad emulator shown in the talk; several links to source, live demos, and Sutherland’s original thesis are provided.
  • Emacs is analyzed as an early, deeply extensible “live” environment, with emphasis on its design for runtime modification, hooks, and user empowerment.
  • DynamicLand and newer projects inspired by Kay/Engelbart are mentioned; upcoming large releases of research materials (but not clearly source code) are anticipated.

Meta and community commentary

  • Some lament the “negative energy” of typical online discussion and its impact on ambitious projects.
  • Others nostalgically recall early-internet optimism, contrasting it with today’s commercialized, polarized web, while noting that low-quality communication has always been common offline as well.