*Disclaimer: the research behind this page was AI-generated: gathered and written by Claude using [swamp](https://github.com/swamp-club/swamp) workflows and [Exa](https://exa.ai) search. Check the linked sources before relying on a summary.* The desktop has hardly changed in twenty years. Files, folders, overlapping windows and apps are commodities, and they carry inertia. The open problem sits higher in the value chain: getting back into the work you were doing. Scott Jenson calls this *working memory*: the OS should remember what you were doing, so the user stops being the integrator of last resort. ## The map ```mermaid wardley-beta title Desktop UX for knowledge work size [1100, 720] anchor Knowledge worker [0.97, 0.55] component Doing creative work [0.88, 0.55] component Resuming context [0.78, 0.12] component Finding past work [0.76, 0.42] component Apps [0.66, 0.78] (inertia) component Activity timeline [0.62, 0.22] component Spatial canvas [0.56, 0.18] component Project clipboard drawer [0.52, 0.10] component Overlapping windows [0.50, 0.93] label [-150, 4] component Desktop search [0.45, 0.80] component Files and folders [0.42, 0.90] (inertia) component Activity capture [0.36, 0.38] component Semantic metadata store [0.30, 0.16] component Local-first sync [0.26, 0.30] component Cloud storage [0.24, 0.84] component Local LLM [0.20, 0.42] component OS and hardware [0.06, 0.96] label [-125, 4] Knowledge worker -> Doing creative work Doing creative work -> Resuming context Doing creative work -> Finding past work Doing creative work -> Apps Doing creative work -> Overlapping windows Resuming context -> Activity timeline Resuming context -> Spatial canvas Resuming context -> Project clipboard drawer Finding past work -> Activity timeline Finding past work -> Desktop search Apps -> Files and folders Activity timeline -> Activity capture Activity timeline -> Semantic metadata store Desktop search -> Files and folders Activity capture -> Local LLM Semantic metadata store -> Local-first sync Files and folders -> Cloud storage Local-first sync -> OS and hardware Cloud storage -> OS and hardware Local LLM -> OS and hardware evolve Activity timeline 0.45 evolve Activity capture 0.62 evolve Local LLM 0.72 note "Lifestreams 1996 / Scopeware 2001" [0.70, 0.05] note "Windows Timeline, Zeitgeist: built then removed" [0.68, 0.30] note "Recall-style tools" [0.40, 0.48] ``` How to read it: - **Resuming context** is in genesis. No shipping desktop does it. - **Activity timeline** has been built several times: Lifestreams and Scopeware, then GNOME Zeitgeist and Windows Timeline. Each was later abandoned. It is still custom-built. - **Activity capture** is moving fast because of the Recall-style tools. These tools record your activity but do not change the document model, so the timeline stays a search layer over files and folders. - **Local LLM** is moving toward commodity. That makes grouping raw activity into tasks cheap, and that grouping is the hard part of a timeline. - **Files and folders** and **Apps** carry inertia. Each new idea gets added beside them. None has replaced them. ## Scott Jenson - [Are we really going to use the same Desktop UX forever?](https://www.youtube.com/watch?v=V7AfAcQwLW0) (talk, 2026) - [Akademy 2026 recording](https://media.ccc.de/v/kde2026-7-are_we_really_going_to_use_the_same_desktop_ux_forever) (KDE) - [How the Desktop UX needs to evolve to keep up with Local-first](https://www.youtube.com/watch?v=-IOLRcFC6OY) (talk) - [Evolving Desktop OS, Local-First, & Agentic UX](https://www.infoq.com/podcasts/evolving-desktop-agentic-ux/) (InfoQ podcast, with transcript, 2026) - [Ubuntu Summit 25.10 session](https://discourse.ubuntu.com/t/are-we-stuck-with-the-same-desktop-ux-forever/67253) - [The Future Needs Files](https://jenson.org/files/) and [The Present Needs Files](https://jenson.org/files2/) (essays, 2021) - [Small is Beautiful: Why Desktop UX still has something to teach Mobile](https://jenson.org/small/) (2016) - [Old is New: Why Desktop UX still inspires](https://jenson.org/oldnew/) (2017) - Prototypes: [FocusStream](https://github.com/scottjenson/FocusStream) (a Lifestreams-style browsing timeline), [DesktopUX0](https://github.com/scottjenson/DesktopUX0) (files and clipboard as one model), [DesktopUX1](https://github.com/scottjenson/DesktopUX1) (discovering intent from history), [Glance](https://github.com/scottjenson/Glance) (KWin edges for ultrawide screens) ## Lifestreams and its line - Freeman & Gelernter, [Lifestreams: A Storage Model for Personal Data](https://sigmodrecord.org/publications/sigmodRecord/9603/pdfs/381854.381893.pdf) (SIGMOD Record, 1996) - Freeman, [The Lifestreams Software Architecture](http://www.cs.yale.edu/homes/freeman/dissertation/etf.pdf) (Yale PhD dissertation, 1997) - [Lifestreams: An Alternative to the Desktop Metaphor](https://www.cs.yale.edu/homes/freeman/ericchi96.html) (CHI 1996) - Fertig, Freeman & Gelernter, ["Finding and Reminding" Reconsidered](https://homepages.cwi.nl/~steven/sigchi/bulletin/1996.1/fertig.html) (SIGCHI Bulletin, 1996) - Gelernter, [Beyond Lifestreams: The Inevitable Demise of the Desktop Metaphor](https://doi.org/10.7551/mitpress/1584.003.0005) (MIT Press, 2007) - Rekimoto, [Time-machine computing](https://doi.org/10.1145/320719.322582) (UIST 1999) - Harrison, [Kronosphere: A Replacement for the Ailing Desktop Metaphor](https://cs.nyu.edu/web/Research/MsTheses/harrison_chris.pdf) (NYU MS thesis) - [Mirror Worlds](https://en.wikipedia.org/wiki/Mirror_Worlds) and [Scopeware](https://www.computerworld.com/article/1441619/scopeware-aims-to-unite-corporate-information.html) (2001), the commercial attempt ## Implementations, shipped and abandoned - [GNOME Activity Journal / Zeitgeist](https://en.wikipedia.org/wiki/GNOME_Activity_Journal) - [Windows Timeline](https://support.microsoft.com/en-us/windows/ai/ai-features/get-help-with-timeline) (removed in Windows 11) - [OpenRecall](https://github.com/openrecall/openrecall/), [Pensieve](https://github.com/arkohut/pensieve): open-source Recall-style capture ## Adjacent - Jef Raskin's [Archy](https://raskincenter.org/rchi/archy-how-it-worked/): no applications, one content space - Ink & Switch, [Malleable software](https://www.inkandswitch.com/essay/malleable-software/) (2025)