You can see the full website here

This project was sparked by a conversation I had with the professor of my Game Theory & Machine Learning class, Gauthier Gidel and with a friend and student in the class, David Dobre. We were both feeling frustrated with the amount of glaring results cherry-picking in machine learning papers, and about how incremental results get completely discarded as useless, even if they offer important new research insights. Our solution was to incentivize research blog posts. By giving a dedicated venue for incremental results, test-of-time works, and other valuable-but-not-pivotal contributions, we hoped to reduce their inclusion at other venues. By removing incremental results from bigger conferences, we hoped to reduce the cherry-picking that was often “required” to have fully “groundbreaking” results.

To do so, we wanted blog posts to be formally recognized as scientific achievements. We contacted Sebastien Bubeck (Sr. Principal Research Manager at Microsoft) and Claire Vernade (former Research Scientist at DeepMind, now Group Leader at University of Tuebingen). Together, we wrote a proposal to ICLR proposing they let us organise a new conference track, the Blog Track. They accepted, and we became inaugural track chairs. Being track chair at such a huge conference at 24 years of age was an extraordinary experience.

Even though it was the first edition of a proper blog track for our field, it was a resounding success. As inaugural track chairs, we accepted 20 posts out of 61, roughly the size of the first ICLR conference (which is now one of the big ML conferences). Social media was abuzz about this track, with overwhelmingly positive feedback.

Being indirectly talked about in a Reddit thread with 200 upvotes in the ML subreddit was also a bit of a weird experience!