The Open Science Framework (OSF) announced they are winding down core parts of the platform over the next twelve months. If you use OSF as a key part of your research workflow, you will be forced to change how you work. No new projects after November 16th (2026) and all projects frozen on February 19th (2027).
Projects are the fundamental part that made the OSF, so that's a major shift. It's not surprising this happens rather unexpectedly, as the OSF has previously had the tendency to spring some of these announcements on people.1 I highly recommend you read the announcements of what is changing and why on their blog.
Below are some of my honest, initial reflections.
The OSF was pivotal to the early days of reforming research around 2012-2017. I used it a lot for a while, and ultimately used it through its integrations only. It added some great services and some not so great ones. I did not always agree with their approach, but I was often glad they existed.
The OSF has had many failings over the years and I think it is great they're making a decision on the role it has in 2026 and beyond. To me, the decision to unwind projects is a reflection of the fact that they did not really connect with the userbase to understand the needs of researchers, while riding on the tide that they served the researcher's needs. They gobbled up tens of millions of dollars for their operations. The fact that this announcement came as a surprise, is a highlight of how disconnected the development was from the people they claim to serve.
There's also plenty of detail questions that are glossed over. Take for example their projected $4-5 million operating expenses – I would have loved to have seen these broken down. What exactly is the expense? Their 2023 financials2 say most of the money goes to payroll, but they argue storing petabytes plus the bandwidth to serve it is the motivating factor. Which is it?
Sustainability of open infrastructure is a big question, I agree – and we now primarily learn the outcome. I know other organizations struggle with sustainability too, and it would've been great to know more about the ongoing struggle. Part of the open innovation we need to do collectively is sharing those struggles so we can learn from one another, and I wish we could be more honest with each other about this while things are still uncertain. The OSF dropped the ball by springing this in an announcement, when tons of people are out on holidays.
Another thing: Why has the development momentum never really improved? OSF today is so similar to OSF ten years ago, minus the redesign that honestly made it so much worse to use. I won't berate strategic choices, as there's always something to remark, but I fear that they did not keep improving the OSF at a steady pace to keep people excited enough. The OSF did not live up to what it could have been with the amount of money that went to it. In that sense, good choice to unwind parts now.
But honestly, Center for Open Science started pivoting their strategy many years ago, focusing on positioning themselves as doing meta-research projects. The writing was on the wall they would change the OSF down the line – I am glad to see there's at least a small offramp now. It would have been nice for that conversation to happen over the years too – a good strategy does not happen over the course of a few weeks. It makes the pill all the harder to swallow now.
I hope they stay committed to maintaining link stability, but honestly, I fear this is the start of unwinding other parts of the OSF in the next decade too. The cost problem will keep on existing for them, and what is still okay now will not be tomorrow. They're already laying the groundwork for deleting private projects in the future, which may well lead to removing other things too.
Having started developing open infrastructure myself since 2019, I increasingly saw how bloated and inefficient the OSF platform was. It is slow and not really joyful to use. The spam measures created false positives left and right. The platform completely stagnated due to its own bloat in early technical design decisions3, that dragged on and on and on. It is good they finally made a choice, as clarity is needed. I am afraid they are creating a drip of decisions that will unwind trust in other services too. OSF is more than the OSF in that way, because it was the first. If it is supposed to be a role model, what does this change role model now?
Finally: If the whole theory of change for the Center for Open Science is to make things possible with infrastructure, and behavior change possible by UX, what remains now? That is a question that remains unanswered, and honestly, highlights that the theory of change is increasingly distant from the COS' practice. They did great work for a while, and we will have to see what we can do when the pieces fall into place.