TL;DR
In this episode of The Good Thing, Stefan and Jens unpack the risks and realities of open source. They cover licensing, forks, domain expertise, and why copying code is not the same as building a company. Along the way, they share lessons from Epic Games, Kubernetes, and their own experience with Cosmo.
Lessons from Unreal Engine
Stefan opened with a story from Epic Games founder Tim Sweeney, who released Unreal Engine as free software. The move wasn’t altruism—it was about distribution.
If every single person that plays video games uses Unreal Engine, they’re not using our competitor.
For Stefan, it showed why open source can dominate by lowering barriers and creating an ecosystem that competitors struggle to match.
Betting on Federation
Jens explained that WunderGraph’s long-term bet is on federation as a model, not just GraphQL Federation. To succeed, it must be accessible.
If we make it somehow hard for companies to adopt it, that’s not going to happen.
He pointed to Netflix building its own gateway instead of adopting Apollo’s router, arguing that restrictive licenses can block adoption and fracture ecosystems.
Open Source Isn’t Free
The hosts pushed back on a common myth: free code doesn’t mean free support. They tied this back to the reality of open source licensing—download is free, but support isn’t.
The most important thing people have to understand is open source is a way of distributing software. You are allowed to download it… and that’s it. Exactly there it ends.
They described the imbalance when companies demand enterprise-level help without contributing code or revenue.
Why Forking Cosmo Doesn’t Work
Could someone fork WunderGraph’s Cosmo and compete? Stefan and Jens broke down why that’s nearly impossible. Copying code doesn’t give you years of domain expertise, enterprise operations, or trust.
By copying the code, you are not copying the brand.
Even if a startup raised money, they’d face brutal VC questions: why choose you over the original? Without logos, stories, or sales experience, the pitch falls apart.
Big Companies Aren’t a Threat Either
Stefan asked whether a big tech company could fork Cosmo. Jens said running a managed service like Cosmo requires deep expertise and culture that large corporates lack.
I just cannot imagine how a big company would pull it off… it’s such a complicated project.
Instead, he welcomed scenarios where hyperscalers might offer Cosmo as a managed service, since it would extend distribution.
The Reality of Maintenance
Maintaining open source is a grind. Jens has been maintaining GraphQL tools since 2018 and said the hardest part isn’t code, but responsibility.
If we merge it, it is now our burden… because if a customer now uses it, it must follow the same standards.
That means every PR, feature, and bug fix carries liability that forks rarely appreciate.
This episode was directed by Jacob Javor. Transcript lightly edited for clarity and flow.
