Welcome to Open Channels News, your go-to source for the latest in open source and digital change. In today’s episode, Robert Jacobi examines why the gap between what technology can do and what it actually delivers matters more than ever:
AI’s Impact on Open Source Maintenance:
Generative AI cuts code writing time by more than half, but maintainers face a surge in low-quality submissions. The Linux Foundation finds that while AI increases participation, it also creates fresh challenges around quality control, trust, and project sustainability.
Denmark’s Open Source Office Pilot Falters:
A Danish government ministry’s attempt to replace Microsoft Office with open source Collabora ended in under six months. Everyday workflow issues like formatting and document compatibility proved tougher than expected, highlighting that software migrations depend on more than just ideals or procurement.
The Changing Shape of the Open Internet:
Ben Thompson argues the era of the wide-open public internet is fading. As AI and new economics reshape who publishes and how, the future may belong to small, trusted communities rather than massive, universal platforms.
Our sponsors keeps our network independent and actively publishing.
Since 2005, Automattic has built tools for the open web including WordPress.com, WooCommerce, and Jetpack that are used by millions of people to create, sell, and publish online. They believe in ownership, flexibility, and open source and we’re grateful for their support. Learn more at automattic.com.
In five days, Omnisend moves every flow, list, and template off your current platform, and you could be paying up to 35% less without lifting a finger. You just show up when it’s done. We’re glad to have them supporting the show and the community we’re building around it. Use the code OpenChannels and get 30% off your first 3 months of any paid plan.
Takeaways
- AI Accelerates Code Creation, Not Maintenance: Robert discusses how generative AI can cut code authoring time by about 52%, but maintainers report a significant increase in low-quality AI-generated code submissions, with larger projects seeing up to 31%. The productivity gain for code contributors may result in a productivity loss for maintainers tasked with reviewing submissions 00:52.
- Governance Gap in AI Use: Only 35% of open source projects surveyed have formal policies regarding generative AI, highlighting a major gap in governance and the need for clear contributions that include tests, context, and disclosure 02:04.
- Open Source Adoption Faces Workflow Challenges: Robert details a Danish ministry’s failed attempt to replace Microsoft Office with an open source alternative due to workflow friction—issues like styles, formatting, and change tracking impeded real-world adoption. This demonstrates that success depends on more than just switching software; it requires rethinking workflows, interoperability, and change management 02:35.
- Digital Sovereignty Has Hidden Costs: Replacing entrenched software platforms involves costs beyond licenses, including training, migration issues, and workflow disruptions. Simply installing a replacement does not guarantee a smooth transition or user satisfaction 03:40.
- The Open Internet Is Evolving, Not Ending: Ben argues that while the truly open and effortless web is harder to access due to security and scale challenges, AI now enables building valuable tools for smaller, specialized communities rather than mass platforms 05:15.
- Sustainability Over Possibility: Robert observes that making something possible—be it generating code or providing alternatives—does not guarantee sustainability, quality, or ease of maintenance. The core challenge remains ensuring quality, trust, compatibility, and ongoing stewardship 06:13.
- Responsibility Persists Despite Easier Tools: While tools for code and content creation have become more powerful and accessible, the underlying responsibilities of maintenance, governance, and building trust are as important as ever 06:51.
Questions This Episode Answers
Q: How is generative AI affecting open source development and code maintenance?
A: Generative AI dramatically speeds up writing open source code, reducing authoring time by around 52%. However, maintainers are facing more work, as up to 31% of incoming pull requests on large projects are low-quality, AI-generated submissions, making code easier to produce but not easier to maintain (00:40).
Q: What challenges do maintainers face with AI-generated contributions?
A: Maintainers must review, understand, and test incoming pull requests, and determine if the project should own the code long-term, turning what seems to be a “gift” into a potential maintenance burden. Most projects lack formal policies for generative AI use, leading to governance gaps (01:33 and 02:04).
Q: Why did the Danish government’s attempt to switch to open source office software fail?
A: Denmark’s pilot to replace Microsoft Office with Collabora failed due to issues like incompatible document formats, problematic styles, and unreliable change tracking—problems that directly impacted employees’ ability to complete everyday tasks efficiently (02:35).
Q: What does the Danish experiment teach about migrating to open source solutions?
A: The experiment shows that success isn’t about simply installing new software, but designing migration around real workflows, interoperability, training, and reliable results; otherwise, costs from lost time and productivity can outweigh the benefits (03:50).
Q: How does Ben Thompson describe the changing nature of the internet?
A: Ben Thompson argues that the open, effortless internet of the past is fading due to new gatekeepers, abundance of content, and AI’s impact. He suggests that the future may be smaller, private groups and specialized communities rather than massive public platforms ( 04:12).
Q: Is running your own website or server still viable today?
A: While the freedom to self-host still exists, it now comes with greater complexity and security risks, making it feel less viable or attractive for casual users than in the early 2000s (05:46).
Q: What is the main lesson from the three stories discussed in this episode?
A: The episodes highlight that making something possible—whether producing code, deploying alternative software, or publishing online—doesn’t guarantee it will be sustainable or effortless; maintaining quality, trust, and compatibility still requires meaningful, ongoing work (Robert Jacobi at 06:32).
Q: What responsibilities remain despite advances in technology and AI assistance?
A: Even as tools make creating and sharing software easier, the work of ensuring quality, trust, and ongoing maintenance doesn’t disappear—these fundamental responsibilities remain in the hands of people and communities (Robert Jacobi at 06:51).
Mentioned Links and Resource
- The Linux Foundation Research (Generative AI and Open Source Development Study) – Highlights AI’s impact on code authoring and open source maintenance. (Note: the report is “Navigating Open Source Software Development in the Generative AI Era,” published October 20, 2026.)
🔗 https://www.linuxfoundation.org/research/oss-development-gen-ai-era-2026 - Collabora Office (Open Source Office Suite) – The Danish government pilot tested Collabora as an alternative to Microsoft Office.
🔗 https://www.tomshardware.com/software/microsoft-office/denmarks-bid-to-ditch-microsoft-office-for-open-source-alternative-collabora-collapses-in-under-six-months-employees-cite-issues-with-style-formatting-and-tracking-changes - Ben Thompson’s Stratechery: “The End of the Internet” – Explores the shifting economics and security of the open internet.
🔗 https://stratechery.com/2026/the-end-of-the-internet/
Episode Transcript
Robert Jacobi: Welcome to OCN, the Open Channels News Broadcast. I’m your host, Robert Jacobi. Today we have 3 stories about the gap between what technology promises and what it actually delivers. AI is making it dramatically faster to write open source code while leaving maintainers with more work. Denmark tried to replace Microsoft Office with an open source alternative and ran straight into the reality of everyday workflows. And Ben Thompson argues that the open internet we grew up with is coming to an end. The common thread? Making something possible is not the same as making it sustainable.
Our first story. The Linux Foundation has released research on generative AI and open source development, and the headline is almost comically perfect for 2026. AI can cut code authoring time by about 52%. Awesome. Meanwhile, maintainers estimate that roughly one quarter of incoming pull requests are low-quality AI-generated submissions. For larger projects, that estimate rises to 31%. So we’ve made it easier to produce code. We haven’t made it easier to maintain code. And those are 2 very, very different things.
The Linux Foundation says 59% of maintainers are seeing more contributions. 71% believe AI lowers the barrier to participation, but 53% disagree that AI has improved trust in their communities. Think about the economics here. A contributor can generate a pull request in minutes. The maintainer still has to understand it, test it, check whether it fits their architecture, and decide whether the project should own that code for years. And that last part matters. A pull request is not a gift if it comes with a maintenance bill nobody agreed to pay.
I don’t want to dismiss the upside at all. The survey says developers are putting saved time into technical debt, new features, and learning. Those are real benefits. The issue is whether the productivity gain for the person submitting code turns into a productivity loss for everyone reviewing it. Only 35% of projects surveyed have a formal policy permitting generative AI use. That leaves an enormous governance gap. Takeaway is simple. AI-assisted contributions should arrive with tests, context, and disclosure. Don’t make a volunteer maintainer do the homework your AI skipped. Open source has always depended on people. Faster code generation does not remove that fact.
Let’s move from creating software to actually using it. Tom’s Hardware reports that a Danish government ministry has ended a pilot to replace Microsoft Office with the open source Collabora suite. The experiment began in 2025, involved 38 employees, and reportedly lasted less than 6 months. Sticking points included styles, formatting, and tracking changes. Not glamorous problems, but these are the problems that determine whether somebody can finish their work or not, at least on time.
And this is where conversations about digital sovereignty get complicated. On paper, replacing proprietary office software with an open source alternative sounds straightforward and awesome. You control more of your stack, you reduce dependence on one vendor, you gain flexibilities. And while these are worthwhile goals, an organization doesn’t run on license philosophy alone. It runs on documents that need to look right, revisions that need to survive, and people who need to exchange files with everyone else. If a document breaks every time it moves between teams, that isn’t a minor inconvenience, that’s a cost. Training is a cost. Lost time is a cost. A migration that employees can’t reliably use is a cost.
This pilot doesn’t prove open-source software can’t work. It shows that a successful migration has to be designed around actual workflows, interoperability, and change management, not just the software procurement line items. The lesson isn’t that Microsoft is unbeatable, it’s that replacing an entrenched platform takes more than installing a substitute. Freedom matters. So does getting the document out the door.
And finally, Ben Thompson at Stratechery has published an essay called “The End of the Internet.” And if that title sounds a little dramatic, well, it’s a good way to get my attention. Thompson starts by remembering installing WordPress back in 2003, when putting a website online meant figuring out PHP, MySQL, and hosting it yourself. Anyone could publish, anyone could run a server. That openness was the point.
His argument is that the environment has changed. First, search and social platforms became the gatekeepers of discovery. Then the economics of unlimited content made it harder to stand out. And finally, AI adds another layer, more automated content, automated attacks, and less certainty that a publisher gets any traffic back when machines consume its work. Thompson points to the growing security burden as one reason the old model of casually running your own public server feels less viable.
But there’s a turn. He isn’t saying creation is finished. He’s saying the next wave may be smaller. Private groups, specialized communities, software built for one team, one household, or one very particular problem. AI lowers the cost of making something useful for a tiny audience without needing to turn it into the next giant platform. It’s a fascinating reversal of the last 20 years. We spent decades assuming the goal of internet software was to scale to everyone. Now the opportunity may be to build exactly what a small group needs.
I’d push back on one thing. The open internet isn’t literally dead. People still publish, host, collaborate, and build in public. But the economics and security assumptions that made that early era feel effortless have changed. And the WordPress example is a useful and perfect reminder. The freedom to run your own site still exists. The question is how much complexity and risk you want to own yourself. Maybe the future isn’t one enormous public square. Maybe it’s millions of smaller rooms connected by open tech with enough trust to make those connections useful.
So let’s see how today’s stories fit together. The Linux Foundation shows us that making code cheap doesn’t make maintenance cheap. That making an alternative available doesn’t make adoption effortless. And Ben Thompson argues that making publication and software creation accessible to everyone doesn’t automatically make the public internet economically or technically sustainable. 3 stories, 1 recurring mistake: confusing the ability to create something with the work required to keep it useful.
AI will help us make more software, open source will continue to offer alternatives, and smaller communities may find new ways to build exactly what they need. But quality, trust, compatibility, and maintenance still have to come from somewhere. The tools are getting easier. The responsibility isn’t going away. And that’s OCN. I’m your host, Robert Jacobi. Be open and stay secure.
