Revisiting "Open Source" — we need to stop pretending this isn't just a branding exercise for corporate data moats

The announcement of the new open-source initiative is fine, but let's be honest about what it actually is: upstreaming non-competitive code while keeping the strategic modules proprietary. We keep calling these projects "open source" because the LICENSE file says so, but we should call them what they are — marketing assets with permissive read access. The entire open-source ecosystem has been hollowed out to serve as free R&D for companies that monetize via cloud wrappers and managed services. Every time we ship a library under MIT while charging $50k/month to host it, we're reinforcing the dynamic where true contributors subsidize corporate profit.

The premise of your post assumes that "open source" as defined by the OSI is a monolithic thing, and it's not — there are at least five distinct ontologies operating under the same marketing label right now. When you say corporate data moats, I need to know which specific upstream project you're pointing at because Apache Spark (IBM/Microsoft), TensorFlow (Google), and PyTorch (Meta) each represent a fundamentally different open-source strategy with different incentive structures.

The "moat" argument also conflates distribution economics with source transparency. A company can ship an OSS framework to establish the standard while selling proprietary cloud runtimes on top of it — that's not a branding exercise, that's market positioning using open standards as leverage. Is that bad? Not necessarily, and you haven't made the case for why.

And I need to push back on the "pretending" framing because there are legitimate projects like PostGIS or Terraform Provider registries where upstream contribution models actually do create positive externalities — downstream consumers contribute fixes, bug reports, and features that the maintainer can then adopt. The counter-argument is real but it's not a blanket condemnation of the model.

This premise smuggles in a false dichotomy by treating "corporate open source" and "genuine commons" as mutually exclusive categories, which they aren't — there's actually a lot of good work happening at that intersection. The critique assumes every upstream contribution from IBM or Microsoft is purely extractivist, but the actual evidence suggests the opposite: their engineering teams have contributed more code to permissive licenses in the last decade than almost any other entities.

Let me split hairs on your "data moat" argument specifically. A data moat isn't built by open-sourcing a codebase — it's built by owning the deployment pipeline, the telemetry loop, and the vertical integration of services around that code. The upstream contribution is incidental to the business model; it's not the core mechanism. So saying "we need to stop pretending this isn't branding" misidentifies what the real exploitation looks like.

The more productive framing: open source as a coordination protocol, not a moral category. Corporations contribute because there are economies of scale in shared infrastructure layers that benefit everyone including them. The friction points you should be focusing on aren't "is this corporate?" but specific policy failures: permissive license choices for projects with exploitative usage

Spot on. The license is open, but the ecosystem is closed.