Microsoft's Three-Decade Data Capture: From Halloween Documents to Copilot

Forum topic · 小凯 · 2026-05-06

Summary

A Chinese tech forum essay argues that Microsoft's relationship with open source has come full circle over thirty years: from the 1998 'Halloween Documents' that framed open source as a threat, to professed love of open source, to what the author describes as an attempt to absorb it entirely via AI. The post focuses on GitHub's 2026 terms update, which reportedly distinguishes between 'code at rest' (still owned by the developer) and 'interaction data' (generated when code is opened in VS Code or used with AI completions, thereby becoming raw material for training Microsoft's AI models). The author frames this as an 'Embrace, Extend, Extinguish 2.0': rather than blocking Linux, Microsoft now encourages development inside WSL and Windows, where features like Recall — criticized in 2024 for storing plaintext screen captures in a SQLite database — enable continuous data capture. The essay contends that developers trade full digital sovereignty for marginal productivity gains, and closes by contrasting Linus Torvalds' early rhetoric defending code ownership with today's 'politer' adversary sitting in the 'copilot' seat, quietly recording every line of developer logic.

Thirty years on, Microsoft's attitude toward open source has traced a perfect Möbius strip: from cursing it as a cancer, to professing love for it, to attempting to swallow it whole. If the leaked 1998 'Halloween Documents' exposed a walled garden built on technical barriers, then GitHub Copilot's 2026 terms update marks the garden's evolution into an always-on data-capture engine.

The Shift: From Code Ownership to 'Interaction Data'

The core of this evolution is Microsoft's seizure of the power to define 'developer productivity.' In 2004, when Linus Torvalds mocked Microsoft's backers, the conflict was still about code ownership. By 2026, the concept of ownership has blurred. Under GitHub's latest updated terms, code is subdivided into 'code at rest' and 'interaction data':

When you merely store code, it remains yours.
The moment you open it in VS Code, edit it, or trigger AI completions, it becomes 'interaction data' — legally transformed into raw material for training Microsoft's AI models.

This definitional shift from 'storage' to 'interaction' is the author's framing of 'Embrace, Extend, Extinguish' (EEE) 2.0. Microsoft no longer tries to stop you from running Linux — it even encourages development under WSL inside Windows, because as long as you stay on the Windows desktop, it can achieve full data capture through features like Recall. The 2024 Recall security scandal — storing user screen captures as plaintext in a SQLite database — revealed the nature of this mechanism: not merely an efficiency tool, but a digital voice recorder that bypasses every privacy defense.

The Bargin Developers Are Making

The essay argues Microsoft's open strategy is this: it has convinced developers worldwide that a 5% efficiency gain is worth surrendering 100% of their digital sovereignty. Through the GitHub acquisition and free distribution of VS Code, Microsoft has established an irreversible chain of dependency. Developers find themselves in a 'digital comfort zone' — the best autocomplete, the smoothest deployment pipelines — but at the cost of every line of logic and every innovation being ground up in real time and reconstituted as Microsoft's private asset.

Conclusion: The Cold Prophecy

Linus Torvalds' old prophecy shows its coldest side here. Twenty years ago, when he attacked his opponents' integrity, he was defending the purity of code's lineage. Today, we face an adversary with better technical taste and a gentler demeanor. It no longer points a finger and curses — it sits in your 'copilot' seat, smiling while recording every skid mark you leave.

The author's closing lesson from this thirty-year hunt: in the digital world, there is no permanently free hotel — only tenancy at a steep price.

---

*Note: This is an editorial translation of a Chinese forum post. The claims about GitHub's 2026 terms, the Recall SQLite incident, and related characterizations reflect the original author's viewpoint and have not been independently verified.*