Ep5: Provenance & Archival Slivers
Archivist and media theorist Christopher Stahl joins Michael for a conversation about provenance (the documented history of where something came from, who touched it, and how it changed hands) and what happens to that chain of custody when everyone can remix media and AI can generate it from scratch. They move from analog archives, where provenance meant physical objects with visible wear and institutional memory, to digital systems where it became timestamps and metadata, to today's AI landscape where the question isn't just "where did this come from?" but "what was trained on what, and who's responsible?"
They explore why traditional preservation is breaking down, what it means to save context instead of just files, and how concepts like archival slivers and governed layers might help meaning survive remixing. From art history's chain of custody to Ambistream's approach to traceable, layered media, it's a grounded look at how authenticity, responsibility, and memory travel through content that's increasingly automated and oversaturated.
Topics we cover:
00:45 – Christopher background: historiography & stage magic, Houdini
01:55 – From art history to library science (MLIS), preservation as a living practice
02:55 – “Born digital” vs digitized media and why the language around digital files borrows from biology
04:20 – Defining "Provenance", chain of custody, and how provenance differs for physical vs digital objects.
06:10 – Archival theory: "Respect des fonds" & original order
09:40 – The problem of scale: too much media, too little meaning, and why abundance devalues individual artifacts
10:40 – Archival slivers: Trevor Owens idea that we only ever save “a sliver of a sliver.”
12:10 – Dead Internet theory & disappearing culture - Why we assumed the web was permanent; it isn’t
13:40 – Digital humanities and studying thousands of cultural objects at once using computational methods
15:00 – Why “boring” and safe tech survives
16:30 – Fragility in digital systems, platform decay
17:50 – Accessibility, and of layered media: Subtitles, translations, sign language
19:30 – Ambistream’s core idea: layered, governed media, interactive controls, attribution, and device-agnostic playback
21:00 – Social TV explained (90s TV meets remix culture) Channels, looping, EPG-style playlists, and lean-back discovery
22:40 – Curation, remixing, and audience roles
24:30 – Provenance as a user experience: QR codes, attribution pages, looping media, and comfort through repetition.
27:30 – AI as a tagging and production layer, mood detection, computing costs, APIs vs local models.
29:00 – Ambiguity, folksonomy, and the many interpretations of “pink"
31:00 – Prompting, surveillance, and algorithmic intent vs. user agency and personalizaiton
33:00 – Nostalgia, Winamp, and retro interfaces
34:30 – The coming flood of AI content: Bots outnumbering humans, AI eclipsing human output, and the need for sub-curation
36:00 – Social TV as ambient media: House parties, background video, shared spaces, and non-main-character media
37:30 – Valuing AI vs human-created work: Aesthetic, cultural, monetary, human labor, and environmental costs
39:00 – AI as augmentation, Family photos, Harry Potter style animated memories, and ethical disclosure
40:20 – AI limits, copyright, and institutional caution
41:20 – Human-in-the-loop as future literacy: Bias, context, documentation, and why humans still matter
42:00 – The Mechanical Turk metaphor: the hidden human inside “intelligent” systems