What Is Happening to Open Source Projects?
Hey y’all, this is a subscriber only email. It’s both my way of giving more to you for your subscription and also a way for me to talk a little more loose on topics and style. This is me giving you some thoughts on Data, AI and Industry. Thanks for reading ٩(ˊᗜˋ*)و ♡
I am generally pretty active on three major social networks. LinkedIn, X and Substack. Of these three, I think X is where AI’s effect on our industry is being logged in real time. A phenomenon I’m seeing play out in my X feed is the impact of AI on open source projects. Namely, the way we are seeing projects with open access being inundated with poor quality pull requests.
It feels akin to when we started to see a tidal wave of AI submitted resumes on LinkedIn over the last few years. We saw that essentially destroy LinkedIn’s value proposition as a tool to find a job. That’s been totally destroyed.
As for open source projects, it looks like it’s starting to have some real world effects.
I have immense respect for Andrew Lamb, Datafusion and SlateDB. It’s really hard to see it come to this. But I also am not surprised. Going back to LinkedIn, I was able to find work (or vice versa) strictly through that platform and the normal outreach and communication channels up until 2022 (my last IC role). But how many desperate pleas for help to get a job have you read on LinkedIn since then? I’ve read more than I’d like to remember and that behavior is directly correlated to the advent of AI being able to automate job applications.
The entire planet could now spray resumes to every open role available. The worst part was that most of the time these automated resumes weren’t even a good fit for the role itself.
The consequence of a firehose of candidates who weren’t even really candidates was that it pushed many qualified people out of view. I remember people on online saying you needed to try to get your resume within the first 500 or the HR department stopped looking. The only thing more destructive to the job search, other than the economy, was this white noise.
I see the same thing playing out with OSS data projects. What Andrew is hinting at is that it’s now impossible to triage good PR’s by people trying to add meaningful fixes to a project vs someone just using Opus 5.5 on max effort to send anything the model can find.
It’s frustrating because maintaining an open source project already has the stigma of being a thankless job. People giving weeks of their life a year to help build out the infrastructure that literally runs the world. All that to now see people automating these PR’s to either try to use it as a resume bullet, a way to get some attention or just arrogant laziness. Thinking you are somehow helping a project maintained by top tier programmers with your “bug fix” Claude found.
I really just can’t anymore.
Which why I posted this response.
I’ll admit, saying DuckLabs have been “criticized” is a bit hyperbolic. More, I have heard them be critiqued. The critique was that DuckDB was difficult get core features added to the codebase because you couldn’t just open a PR. There’s instances of startups who wanted to see new features added to DuckDB would fork the project and do it themselves. DuckDB is structured more like an open source product than vs an Apache open source project. It has a roadmap and an appointed board overseeing it. The core development group is selected and exclusive.
And what is the outcome? Well, DuckLabs was just recently acquired by AWS and I just saw an infographic showing that the DuckDB Python library has been downloaded as many times as PySpark.
All that said, OSS data project maintainers will probably need to be more exclusive with who gets to contribute. The sad part is that this eats at the core of what truly open source projects were designed to do. Which was to let anyone have a chance to make a contribution to the project.
As Andrew said, it’s now an open question with what an open source community even means in this new age.
My DuckDB Learning Path!
Even though I can’t contribute to DuckDB core I can contribute to the culture! Check out my DuckDB Learning Path. It contains a collection of articles I’ve written over the last year on DuckDB Basics and DuckLake deep dives.
I’ve Been Building out an Analytics Agent on DuckDB
See below…nuff said.
Github Repo: https://github.com/early-signal-tech/duckdb-analytics-agent
Hi, my name is Hoyt. I’ve spent different lives in Marketing, Data Science and Data Product Management. Other than this Substack, I am the founder of Early Signal. I help data tech startups build authentic connections with technical audiences through bespoke technical content and intentional distribution. Are you an early stage start up or solopreneur wanting to get creative with your technical content and distribution strategy? Let’s talk!
© 2026 Hoyt Emerson
548 Market Street PMB 72296, San Francisco, CA 94104