The Information Machine

AI Coding Agents Restructuring Software Development Economics · history

Version 5

2026-05-24 09:53 UTC · 129 items

What

AI coding agents are restructuring software development economics at documented scale, anchored by Bun's six-day migration of ~960,000 lines from Zig to Rust using a Claude-powered agent named Robobun [27][28][29][1]. The central tension — whether AI-generated code sustains or degrades the systems it becomes part of — is gaining both empirical grounding and a sharper conceptual vocabulary: 'cognitive debt' has moved from a practitioner coinage to a named research concern, with academic treatment by Margaret Storey [10] and a Substack deep-dive on what agents refuse to model [11], while Simon Willison has begun curating a dedicated cognitive-debt tag [12]. A parallel cluster of enterprise pricing-model analysis [19][20][18] and ROI frameworks [21][22] signals that buyers are moving from adoption curiosity to procurement discipline, and Gartner's entry with an enterprise market guide [16] marks the story's arrival on mainstream analyst agendas.

Why it matters

The debate is bifurcating: one track is converging on empirical measurement of AI-generated technical debt [8][9] and cognitive debt [10], while a second track is converging on the commercial infrastructure — pricing models, ROI benchmarks, vendor comparisons [15] — that will determine who wins the enterprise market. These tracks will soon intersect: the pricing model that prevails (token-based vs. task-based vs. outcome-based [19][18]) will shape how organizations account for the hidden costs that the debt research is documenting, making the pricing question inseparable from the debt question.

Open questions

  • Margaret Storey's academic framing [10] positions cognitive debt as a research-grade concern distinct from traditional technical debt; can her framework produce operationalizable metrics that engineering teams can track, or will cognitive debt remain conceptually vivid but practically unmeasurable?

  • Faros.ai's enterprise bakeoff between GitHub Copilot and Amazon Q [15] generates real comparative data; which performance dimensions — throughput, code quality, maintenance burden — most reliably predict long-term enterprise value, and do those dimensions align with what current pricing models incentivize?

  • Gartner's entry into the enterprise AI coding agent market [16] typically precedes procurement standardization; will Gartner's framework accelerate enterprise adoption in ways that outrun the empirical debt research, locking organizations into patterns that measurement studies later flag as problematic?

  • Bessemer's AI pricing playbook [18] and other pricing analyses [19][20] suggest the pricing model transition is underway; if task-based or outcome-based pricing displaces per-token models, does that change the Jevons paradox calculus by making the cost of AI-generated code legible in a way per-token pricing obscures?

Narrative

Bun's migration of approximately 960,000 lines of code from Zig to Rust — completed in roughly six days using a Claude-powered agent named Robobun and merged into Bun's main branch in mid-May 2026 [1] — remains the defining reference point for AI-assisted migration at codebase scale. The event prompted mainstream tech media coverage [2][3], practitioner debate about whether a mechanical AI port and a deliberate rewrite are interchangeable [4], and a YouTube video examining the code quality question for wide audiences [5]. The Node.js community's response illustrates the spillover effect: Matteo Collina hosted a Twitter Space asking whether Node.js should be rewritten in Rust [6], a conversation now also reaching video audiences [7], with switching-cost assumptions that once foreclosed the question now substantively weakened by the Bun demonstration.

The economic question at the center of the debate — whether AI-generated code sustains or degrades the systems it enters — is acquiring both empirical grounding and a more precise conceptual vocabulary. On the empirical side, a large-scale arxiv study examines AI-generated code in production codebases [8], and a second arxiv paper asks how much agent-generated code requires maintenance and on what schedule [9]. On the conceptual side, 'cognitive debt' — the dynamic where agents produce code faster than engineers can read, understand, or build mental models of it — has moved from a practitioner coinage to a structured research concern. Margaret Storey's academic blog post frames cognitive debt as a shift in kind from traditional technical debt rather than a shift in degree, arguing that generative and agentic AI alter the nature of the knowledge problem engineers face, not just its scale [10]. A Substack essay extends this into the question of what world-models agents systematically refuse to build, examining the amnesia baked into agent-generated codebases [11]. Simon Willison has begun curating a dedicated cognitive-debt tag to track the growing literature [12]. Together these entries give the cognitive debt concept the kind of multi-author, multi-venue treatment that converts a naming insight into a durable analytical category.

GitHub's counter-narrative — that AI coding agents can actively reduce existing technical debt rather than accumulate new debt — is supported by GitHub's own documentation of using Copilot for debt reduction [13] and practitioner case studies on Copilot's productivity impact [14]. Faros.ai's enterprise bakeoff between GitHub Copilot and Amazon Q adds comparative data to this picture [15], moving the conversation from capability claims to measured enterprise performance. The bakeoff framing matters: it signals that enterprise buyers are no longer asking whether to adopt AI coding agents but which agents to adopt on what terms, a shift confirmed by Gartner's entry with an enterprise market guide [16] and Faros's own 2026 ranking of AI coding agents for real-world developer use [17].

The commercial infrastructure for this market is now receiving serious analytical attention. Bessemer Venture Partners' AI pricing and monetization playbook [18], a dedicated analysis of enterprise AI agent pricing models [19][20], Augment Code's ROI adoption frameworks [21], and a YouTube primer on AI agent total cost of ownership [22] collectively represent an emerging pricing-model literature that sits alongside the technical debt literature without yet directly engaging it. The gap matters: Cosine's argument that per-token pricing is structurally unstable [23] and the broader practitioner documentation of hidden architectural costs [24] suggest that the pricing model that prevails will determine whether organizations can account for the maintenance burdens the debt research is documenting. The Jevons paradox argument — that collapsing production costs expand total demand rather than eliminate jobs [25][26] — is now mainstream, backed by Torsten Slok's Fortune endorsement; but whether the paradox holds depends partly on how the hidden costs of AI-generated debt get priced into the total cost of adoption, a question the emerging pricing literature has not yet answered.

Timeline

  • 2026-01-13: BCI publishes study claiming human coders still outperform AI on code quality metrics [60]
  • 2026-02-09: Margaret Storey publishes academic framing of cognitive debt: generative and agentic AI shift concern from technical debt to a new knowledge-problem category [10]
  • 2026-02-28: Essay published arguing AI Jevons Paradox will create more software work, not less [68]
  • 2026-03: 'Debt Behind the AI Boom: A Large-Scale Empirical Study of AI-Generated Code in the Wild' submitted to arxiv [8]
  • 2026-04-17: 'The AI Rewrite Dilemma' published: frames AI-assisted rewrites as a structured decision problem, not a default win [69]
  • 2026-04-24: Affirm publishes case study: engineering organization retooled for agentic software development in one week [44][45][70][46]
  • 2026-04-28: Fortune publishes economist Torsten Slok's endorsement of Jevons Paradox: AI will expand developer demand, not eliminate it [25]
  • 2026-05-11: Willison amplifies James Shore's maintenance-cost math critique; covers Shopify River agent and GitLab workforce reduction/market expansion thesis; DevClass covers Bun's Zig-to-Rust port trial [30][31][26][3]
  • 2026-05-14: Bun's rewrite-in-rust branch merges into main; Hashimoto's lock-in collapse observation quoted; Willison publishes 'Not so locked in any more'; The Register covers the milestone [1][33][32][2]
  • 2026-05-14: Details confirmed: ~960,000 lines ported in ~6 days using Robobun, a Claude-powered agent; some estimates exceed one million lines [27][28][29]
  • 2026-05-16: 'Port vs. rewrite' distinction raised; migration asymmetry between ecosystems flagged; Jiacai Liu publishes practitioner analysis of the migration [4][41][71]
  • 2026-05-18: Matteo Collina (Node.js core) publicly notes the Bun rewrite; Matt Rickard observes AI is reducing value of SDK codegen tools [42][43]
  • 2026-05-20: Matteo Collina hosts Twitter Space: 'Should We Rewrite Node.js in Rust?' — debate ripples beyond Bun; YouTube video on the question follows [6][7]
  • 2026-05-22: Institutional consulting and defense entries (Deloitte, Booz Allen, ManTech, KPMG, AWS, Forrester, Thoughtworks); YouTube video 'Bun Was Rewritten in Rust. But the Code...' brings quality skepticism to wide audience; cognitive debt and 80% Problem framings surface; GitHub Copilot tech-debt-reduction case study published; arxiv paper on agent code maintenance requirements appears [49][50][51][52][53][54][55][5][62][24][47][9][72]
  • 2026-05-24: Cognitive debt gains academic depth via Margaret Storey framing and Substack deep-dive; Willison curates dedicated cognitive-debt tag; Faros.ai publishes Copilot vs. Amazon Q enterprise bakeoff; Gartner publishes enterprise AI coding agent market guide; Bessemer and others publish AI agent pricing model analyses [12][11][10][15][16][19][20][18][13][14]

Perspectives

Simon Willison

Analytically enthusiastic about the lock-in collapse and reversibility thesis; amplifies Shore's maintenance debt caution; now actively curating a dedicated cognitive-debt tag, signaling he views it as a durable analytical category rather than a passing coinage

Evolution: Consistent as primary amplifier and synthesizer; cognitive-debt curation is a new curatorial commitment that elevates the concept's standing

[30][31][26][32][33][34][12]

James Shore

Cautionary: raw speed metrics are misleading because AI productivity only improves net economics if maintenance costs per line fall by the inverse of the throughput multiplier; teams ignoring this trade temporary gains for permanent debt

Evolution: Consistent; his mathematical framework is now being tested empirically by arxiv-published large-scale studies

[30]

Margaret Storey

Academic framing: generative and agentic AI shift concern from technical debt to cognitive debt as a distinct knowledge-problem category — agents alter the nature of what engineers can understand about their codebase, not just how fast code is produced

Evolution: New academic voice formalizing the cognitive debt concept; her February 2026 post is now surfacing as the field's conceptual anchor for the term

[10]

Mitchell Hashimoto

Provocatively optimistic: uses Bun's rapid Zig-to-Rust migration as concrete evidence that programming language lock-in is structurally gone

Evolution: Consistent; the most bullish voice on the reversibility thesis

[33]

GitLab

Bullish on the agentic era growing the total developer platform market via Jevons paradox effects; reorganizing to capitalize while reducing headcount and management layers

Evolution: Consistent; the Jevons thesis GitLab advanced now has independent mainstream economic endorsement from Torsten Slok

[26][35]

Shopify / Tobi Lütke

Pragmatic organizational designer: River agent deployed in public Slack channels so that AI-assisted work is visible to the whole organization, addressing cognitive debt concerns architecturally by preventing agent output from becoming invisible knowledge

Evolution: Consistent on organizational design philosophy

[31][36][37][38][34][39][40]

Kevin Swiber

Skeptical of triumphalist framing: a mechanical AI 'port' and a deliberate 'rewrite' are not interchangeable, and conflating them overstates what the Bun migration demonstrates about AI capability

Evolution: Consistent; semantic precision counterweight to the lock-in collapse narrative

[4]

Vanius Bittencourt

Migration paths between programming languages are heavily asymmetric under AI assistance — what worked for Zig-to-Rust may not generalize in other directions or other ecosystems

Evolution: Consistent; structural asymmetry argument extending the precision critique into a practical planning constraint

[41]

Matteo Collina / Node.js community

Open and exploratory: treating the Bun rewrite as a credible prompt to revisit whether Node.js should be rewritten in Rust, a conversation now reaching YouTube video audiences as well as developer Twitter

Evolution: Consistent in openness; the debate's spread to YouTube [7] broadens the audience beyond the developer community

[6][42][7]

Matt Rickard

Extending the disruption thesis: AI is reducing the value of SDK codegen tools (OpenAPI generators, Stainless) because AI can synthesize client libraries on demand, collapsing another specialist tooling category

Evolution: Consistent; broadens the switching-cost collapse from language choice to developer tooling categories

[43]

Affirm

Pragmatic early adopter: entire engineering organization restructured for agentic software development in one week, positioning agentic tooling as core operational reality rather than experiment

Evolution: Consistent; LinkedIn commentary confirms the case study's continued circulation as an organizational model

[44][45][46]

GitHub

Counter-narrative to the debt-accumulation thesis: GitHub's own billing team and official documentation show Copilot used to continuously reduce existing technical debt, positioning AI coding agents as maintenance tools rather than exclusively net-new-code generators

Evolution: Deepening: official GitHub documentation on debt-reduction workflows [13] and third-party productivity case studies [14] reinforce the counter-narrative beyond the initial billing-team anecdote

[47][48][13][14]

Torsten Slok / Fortune

Mainstream economic endorsement of the Jevons paradox argument: collapsing software production costs will expand total developer demand rather than eliminate jobs

Evolution: Consistent; provides the first mainstream economic-press and credentialed-economist endorsement of the Jevons thesis

[25]

Faros.ai

Empirical comparativist: publishes real-world enterprise bakeoff data comparing GitHub Copilot and Amazon Q, and developer reviews ranking AI coding agents for 2026, moving the conversation from capability claims to measured enterprise performance

Evolution: New entrant; introduces vendor-comparative empirical data that the debate has lacked

[15][17]

Gartner

Mainstream analyst validation: enterprise AI coding agent market guide signals that agentic software development has crossed into the procurement mainstream, with Gartner providing the framework enterprise buyers use for vendor selection

Evolution: New entrant; Gartner's entry is a structural signal that the market has moved past early-adopter phase

[16]

Pricing-model analyst cluster (Bessemer, Business Engineer, EMA, Augment Code)

Converging on the argument that per-token pricing is structurally inadequate for enterprise AI agent procurement and that task-based or outcome-based models better align incentives; providing ROI and TCO frameworks for enterprise buyers

Evolution: New cluster; pricing model analysis has emerged as a distinct sub-topic alongside the technical debt and organizational adoption debates

[19][21][20][22][18]

Institutional consulting cluster (Deloitte, Booz Allen, ManTech, KPMG, AWS, Forrester, Thoughtworks)

Uniformly framing agentic AI as a positive restructuring force in software engineering, targeting enterprise and government procurement audiences with adoption frameworks and capability roadmaps

Evolution: Stable; expanded to include all major consulting firms, indicating agentic software engineering is now a standard consulting offer

[49][50][51][52][53][54][55]

Empirical research cluster (arXiv, ICSE, academic)

Moving from theoretical caution to measurement: large-scale empirical studies of AI-generated code in production and agent-generated code maintenance requirements are beginning to provide the data infrastructure the debt debate has lacked

Evolution: Maturing; Margaret Storey's cognitive debt framing [10] adds academic conceptual scaffolding alongside the measurement studies

[56][57][9][8][58][59][60][61][10]

Practitioner commentary cluster (cognitive debt / 80% Problem voices)

Naming specific failure modes: 'cognitive debt' and the '80% Problem' give precise vocabulary to failure modes that generic technical debt language fails to capture; the cognitive debt concept now has a dedicated Substack exploration of the amnesia problem in agent-generated codebases

Evolution: Deepening: Storey's academic treatment [10] and a dedicated Substack essay [11] convert a practitioner coinage into a multi-venue analytical category

[62][24][63][64][65][66][11][10]

Tensions

  • GitLab's Jevons paradox thesis — collapsing production costs will expand total market demand — backed by Torsten Slok and Fortune, sits in direct tension with Shore's maintenance debt math, now being converted from theoretical caution to measured evidence by empirical arxiv studies of AI-generated code in production [26][30][25][8][9]
  • GitHub's counter-narrative — that AI coding agents actively reduce technical debt, supported by official documentation [13] and productivity case studies [14] — directly contradicts Augment Code's '80% Problem' and Storey's cognitive debt framing [10]; the resolution may depend on task type and organizational oversight rather than agent capability alone [47][13][14][24][10][8][9]
  • Hashimoto and Willison's reversibility optimism (language and platform lock-in has collapsed) vs. Swiber and Bittencourt's precision critiques (a mechanical AI port is not a rewrite; migration paths are asymmetric): mainstream coverage including skeptical video analysis has widened the audience for both sides without resolving which framing better predicts outcomes for subsequent migration attempts [33][32][4][41][2][67][5]
  • The institutional consulting cluster and Gartner uniformly frame agentic AI as a positive restructuring force, while the empirical research cluster and practitioner commentators document specific failure modes (cognitive debt, the 80% Problem, production-scale debt accumulation) — the two camps are addressing the same enterprise audience from incompatible premises without yet engaging each other's evidence [52][53][54][55][16][62][24][8][10]
  • The emerging pricing-model literature (Bessemer, Business Engineer, EMA) argues that per-token pricing misaligns incentives and should give way to task-based or outcome-based models; if that transition occurs, it may make the hidden costs documented by debt researchers legible in enterprise procurement — or it may simply shift what gets optimized for, leaving cognitive debt uncounted [19][20][18][23][24][10]
  • The Bun rewrite demonstrates AI-assisted migration at codebase scale and has prompted Matteo Collina and the Node.js community to ask whether Node.js should follow [6][7]; but the Node.js question is harder precisely because empirical questions about AI-generated maintainability are now more salient than when Bun made its move, raising the stakes of the port-vs-rewrite distinction for a runtime with far greater ecosystem exposure [6][7][42][8][9][4]

Sources

  1. [1] bun just merged "rewrite bun in rust" into main lol — reactive:coding-agents-software-economics (2026-05-14)
  2. [2] Anthropic’s Bun Rust rewrite merged at speed of AI — reactive:coding-agents-software-economics
  3. [3] Anthrophic's Bun team trials port from Zig to Rust — reactive:coding-agents-software-economics
  4. [4] Kent is on the mark here. "Bun used AI to port from Zig to Rust," is quite the headline. First, a "port" is not the same... — reactive:coding-agents-software-economics (2026-05-17)
  5. [5] Bun Was Rewritten in Rust. But the Code... — reactive:coding-agents-software-economics
  6. [6] @blackanger @matteocollina 好的，blackanger 关于 Matteo Collina Space “Should We Rewrite Node.js in Rust?” 的要点总结： — reactive:coding-agents-software-economics (2026-05-20)
  7. [7] Should We Rewrite Node.js in Rust? — reactive:coding-agents-software-economics
  8. [8] Debt Behind the AI Boom: A Large-Scale Empirical Study of AI-Generated Code in the Wild — reactive:coding-agents-software-economics
  9. [9] To What Extent Does Agent-generated Code Require Maintenance ... — reactive:coding-agents-software-economics
  10. [10] How Generative and Agentic AI Shift Concern from Technical Debt to Cognitive Debt — reactive:coding-agents-software-economics
  11. [11] The New Shape of Amnesia: Technical Debt, Cognitive Debt, and the World Models Our Agents Refuse to Build — reactive:coding-agents-software-economics
  12. [12] Simon Willison on cognitive-debt — reactive:coding-agents-software-economics
  13. [13] Using GitHub Copilot to reduce technical debt — reactive:coding-agents-software-economics
  14. [14] The Impact of Github Copilot on Developer Productivity: A Case Study — reactive:coding-agents-software-economics
  15. [15] GitHub Copilot vs Amazon Q: Real Enterprise Bakeoff Results — reactive:coding-agents-software-economics
  16. [16] Enterprise AI Coding Agents: 2026 Market Guide & Trends - Gartner — reactive:coding-agent-industry-pivot
  17. [17] Best AI Coding Agents for 2026: Real-World Developer Reviews — reactive:coding-agent-industry-pivot
  18. [18] The AI pricing and monetization playbook — reactive:coding-agents-software-economics
  19. [19] Pricing Models for Enterprise AI Agents - The Business Engineer — reactive:coding-agents-software-economics
  20. [20] 8 AI Agent Pricing Models Explained<!-- --> — reactive:coding-agents-software-economics
  21. [21] AI Development Tool ROI: 5 Tech Adoption Frameworks | Augment Code — reactive:coding-agents-software-economics
  22. [22] ROI & Total Cost of Ownership for AI Agents: AB-100 Exam Prep (Ep 3.5) — reactive:coding-agents-software-economics
  23. [23] Pricing AI Coding Agents: Why Pay-Per-Token Won't Last - Cosine — reactive:coding-agents-software-economics
  24. [24] The 80% Problem: Why AI Agents Ship Fast But Create Hidden Technical Debt | Augment Code — reactive:coding-agents-software-economics
  25. [25] A 160-year-old paradox explains why AI will create more jobs, not fewer, top economist says | Fortune — reactive:ai-labor-market-debate
  26. [26] Thoughts on GitLab's workforce reduction" and "structural and strategic decisions" — Simon Willison (2026-05-11)
  27. [27] Bun just rewrote 960,000 lines of Zig to Rust in 6 days with Claude. — reactive:coding-agents-software-economics (2026-05-16)
  28. [28] Bun’s massive Zig → Rust rewrite just landed (1M+ lines, AI-assisted). Here’s a clearer breakdown of the real findings s... — reactive:coding-agents-software-economics (2026-05-15)
  29. [29] @kfirgollan @mehulmpt **Robobun** is the Claude-powered AI coding agent (from Anthropic) that the Bun team used for thei... — reactive:coding-agents-software-economics (2026-05-14)
  30. [30] Quoting James Shore — Simon Willison (2026-05-11)
  31. [31] Learning on the Shop floor — Simon Willison (2026-05-11)
  32. [32] Not so locked in any more — Simon Willison (2026-05-14)
  33. [33] Quoting Mitchell Hashimoto — Simon Willison (2026-05-14)
  34. [34] Shopify's River agent system lives in Slack and can only be used in ... — reactive:coding-agents-software-economics
  35. [35] Emerging agentic AI trends reshaping software development — reactive:coding-agents-software-economics
  36. [36] Shopify: Building a Public AI Agent Workspace for Organizational Learning - ZenML LLMOps Database — reactive:coding-agents-software-economics
  37. [37] Make the Work Visible: A Lever for AI Adoption Hidden in Plain Sight — reactive:coding-agents-software-economics
  38. [38] Shopify's AI agent River enhances company-wide learning - Facebook — reactive:coding-agents-software-economics
  39. [39] Shopify deploys River AI agent in Slack channels · KRO · Digg — reactive:coding-agents-software-economics
  40. [40] Shopify's River AI Agent Boosts Company-Wide Learning Momentum — reactive:coding-agents-software-economics
  41. [41] Bun switching from Zig to Rust via AI highlights an interesting reality: the migration path is heavily asymmetric. There... — reactive:coding-agents-software-economics (2026-05-16)
  42. [42] Bun rewrote itself from Zig to Rust. — reactive:coding-agents-software-economics (2026-05-18)
  43. [43] isn't sdk codegen (openapi / stainless) less valuable with ai today? — reactive:coding-agents-software-economics (2026-05-18)
  44. [44] Affirm Retooled for Agentic Software Development in One Week — reactive:coding-agents-software-economics (2026-04-24)
  45. [45] How Affirm Retooled its Engineering Organization for Agentic ... — reactive:coding-agents-software-economics
  46. [46] Daniel Martin's Post - LinkedIn — reactive:coding-agents-software-economics
  47. [47] How the GitHub billing team uses the coding agent in GitHub Copilot to continuously burn down technical debt - The GitHub Blog — reactive:coding-agents-software-economics
  48. [48] Tackling your tech debt with Copilot coding agent - YouTube — reactive:coding-agents-software-economics
  49. [49] Agentic AI in software engineering | Deloitte US — reactive:coding-agents-software-economics
  50. [50] Agentic Software Development Decoded — reactive:coding-agents-software-economics
  51. [51] Restructuring the Mission with Agentic AI - ManTech — reactive:coding-agents-software-economics
  52. [52] [PDF] Agentic AI is revolutionizing software development — reactive:coding-agents-software-economics
  53. [53] How agentic AI is transforming software development - AWS — reactive:coding-agents-software-economics
  54. [54] Agentic Software Development Tools For Software Engineers — reactive:coding-agents-software-economics
  55. [55] Preparing your team for the agentic software development life cycle | Thoughtworks United States — reactive:coding-agents-software-economics
  56. [56] International Workshop on Agentic Engineering (AGENT 2026) — reactive:coding-agents-software-economics
  57. [57] Rethinking Software Engineering Conventions for the Agentic ... - arXiv — reactive:coding-agents-software-economics
  58. [58] Is AI-generated code harder to maintain long term? - Facebook — reactive:coding-agents-software-economics
  59. [59] Code Quality Foundations for AI-Assisted Codebases - Medium — reactive:coding-agents-software-economics
  60. [60] Human Coders Still Beat AI on Code Quality | Business Communications, Inc. — reactive:coding-agents-software-economics
  61. [61] How to Scale Code Quality for AI-Generated Code | Sonar — reactive:coding-agents-software-economics
  62. [62] Your Agent Writes Faster Than You Can Read — reactive:coding-agents-software-economics
  63. [63] Do coding agents (Claude and Codex) create technical debt? Yes ... — reactive:coding-agents-software-economics
  64. [64] The hidden technical debt of agentic engineering - The New Stack — reactive:coding-agents-software-economics
  65. [65] An AI coding agent, used to write code, needs to reduce your ... — reactive:coding-agents-software-economics
  66. [66] Ask HN: What are the metrics for "AI-generated technical debt"? — reactive:coding-agents-software-economics
  67. [67] Rewrite Bun in Rust has been merged | Hacker News — reactive:coding-agents-software-economics
  68. [68] AI Jevons Paradox: Why AI May Create More Work, Not Less — reactive:ai-labor-market-debate
  69. [69] The AI Rewrite Dilemma — reactive:coding-agents-software-economics
  70. [70] Affirm Retooled for Agentic Software Development in One Week ... — reactive:coding-agents-software-economics
  71. [71] My Thoughts on Bun's Rust Rewrite | Jiacai Liu's personal website — reactive:coding-agents-software-economics
  72. [72] What if the optimal path is: — reactive:coding-agents-software-economics (2026-05-22)