thesis.json

{
  "title": "The Software Collapse Has Already Happened",
  "author": "Lida Liberopoulou",
  "date_published": "2026-01-30",
  "date_updated": "2026-02-27",
  "license": "CC BY-SA 4.0",
  "license_url": "https://creativecommons.org/licenses/by-sa/4.0/",
  "repository": "https://github.com/threadbaire/thesis",
  "website": "https://threadbaire.com/thesis.html",
  
  "core_claim": "The software industry's value-producing mechanisms have already collapsed. Not 'they will collapse', they have collapsed. What remains is a transition period where incumbents deploy defensive measures that delay recognition but cannot reverse the structural shift.",
  
  "abstract": "This thesis argues that AI has already collapsed traditional software value capture mechanisms. The evidence: execution layers opening (OpenClaw 198K+ stars, acquired by OpenAI), model layers cracking (DeepSeek V3 MIT), UI commoditizing (MCP Apps), memory opening (mem0), orchestration opening (Ollama subagents), SaaS market repricing (HubSpot -55%, Atlassian -50%+). Incumbent responses—ads, caps, outcome pricing, context graphs, distillation geopolitics—are defensive measures that cannot reverse the structural shift. The rational response is acceleration: open the layers that have become non-scarce, maintain portable context, avoid capture.",
  
  "parts": [
  { "number": 1, "title": "The Collapse (Present Tense)", "summary": "Evidence that implementation layers are already open: OpenClaw, DeepSeek V3, MCP Apps, mem0, Ollama subagents. Market repricing SaaS stocks." },
  { "number": 2, "title": "The Paradox", "summary": "The better AI gets, the more existing value it destroys without creating equivalent monetary value. NBER: 80%+ of firms report no AI impact. Klarna/Salesforce reversals create unstable hybrids. Accountability gap prevents reversal." },
  { "number": 3, "title": "What Remains Scarce", "summary": "Specification—knowing what you want and recognising when you have it—briefly appears scarce. Agency cannot be generated. This scarcity is also temporary." },
  { "number": 4, "title": "The Capture Trap", "summary": "Context graphs as infrastructure for capturing user decision traces. OpenClaw acquisition as capture case study. Dissolution of organisational scale undermines the capture thesis itself." },
  { "number": 5, "title": "The Defensive Moves", "summary": "Desperation signals: ads, outcome pricing, caps, copyright settlements, publisher blocking, distillation geopolitics, financial opacity, performance of confidence." },
  { "number": 6, "title": "The Compliance Counter-Argument", "summary": "Enterprise vendors sell procedural legitimacy, not liability absorption. Compliance unbundles when tracking and provenance become cheap." },
  { "number": 7, "title": "The Next Layer", "summary": "AI dissolves not just existing software but the pipeline of software that would have been built. Disruption vs dissolution. Backends are the next layer to disappear." },
  { "number": 8, "title": "The Open Stack", "summary": "Open infrastructure keeping decision traces portable. Table of open projects by layer. Acquisition pattern warning: open projects require continuous renewal." },
  { "number": 9, "title": "Counter-Arguments", "summary": "Addressed: reliability plateaus, compute scarcity, regulation, liability, vertical integration, new value emergence, agent security, open-source acquisition." },
  { "number": 10, "title": "The Prescription", "summary": "Stop defending non-scarce layers. Accelerate openness. The intermediate hybrid state harms workers most—acceleration is harm reduction." },
  { "number": 11, "title": "The 1989 Analogy", "summary": "New value only emerges after constraints lift. Defending old mechanisms will not produce what comes next." }
],
  
"key_evidence": [
  { "signal": "OpenClaw", "metric": "~198K GitHub stars, acquired by OpenAI", "implication": "Execution layer opened, then targeted for capture", "date": "2026-02" },
  { "signal": "DeepSeek V3", "metric": "MIT licensed frontier model", "implication": "Model layer is cracking", "date": "2025-12" },
  { "signal": "MCP Apps", "metric": "Anthropic UI component protocol", "implication": "UI layer commoditized", "date": "2026-01-26" },
  { "signal": "mem0", "metric": "Open-source AI memory layer", "implication": "Memory layer is open", "date": "2026-01" },
  { "signal": "Ollama subagents", "metric": "Local AI spawning parallel agents, no API keys", "implication": "Orchestration layer is opening", "date": "2026-02" },
  { "signal": "SaaS repricing", "metric": "HubSpot -55%, Atlassian -50%+, Adobe/Salesforce -30%", "implication": "Market pricing in the collapse", "date": "2026-01" },
  { "signal": "NBER w34836", "metric": "80%+ of 6,000 firms report no AI impact on productivity", "implication": "Enterprise deployment failing at measurement layer", "date": "2026-02" }
],
  
  "related_projects": [
    {
      "name": "Threadbaire Method",
      "url": "https://github.com/threadbaire/method",
      "description": "Markdown templates for portable AI memory"
    },
    {
      "name": "Threadbaire Server",
      "url": "https://github.com/threadbaire/server",
      "description": "Database backend, frozen at v1 as RundownAPI demo"
    },
    {
      "name": "RundownAPI",
      "url": "https://github.com/threadbaire/rundownapi",
      "description": "Open spec for AI self-integration endpoints"
    }
  ],
  
  "provenance": {
    "id": "threadbaire:thesis",
    "spec_url": "https://github.com/threadbaire/thesis",
    "license": "CC BY-SA 4.0",
    "attribution_text": "Liberopoulou, L. (2026). The Software Collapse Has Already Happened. Originally published January 30, 2026; updated February 27, 2026. https://github.com/threadbaire/thesis"
  }
}