2024 in Numbers
While I still do not like KPIs, the last year was exhilarating in many ways, and I want to continue last year’s tradition. However, this time, in the spirit of radical transparency, I also want to discuss certain things that did not work out.
But let us start with the good things first! In 2024, we…
- …published 12 conference or journal paper
- …published 2 workshop papers
- …wrote 10 preprints
We also…
- …had 4 academic visitors (Irene, Lydia, Myriam, and Rayna)
- …graduated 4 students (congrats to Pia, Jonathan, Katharina, and Rishendra)
This time, I wrote a whopping 726,383 words, give or take a few, in my personal lab book, applications, reviews, and other assorted documents. Again, this does not include everything I produced and is of course inflated by taking into account markup.
I also had a lot of rejections, to wit…
-
…17 (!) papers or grant submissions have been rejected (although to be fair, I am counting each rejection once, so these numbers refer to the same papers). In general, most of our papers are rejected at least once before they are accepted.
-
I have been rejected from 2 professorships.
On the bright side, I also got tenure at the University of Fribourg in Switzerland. Not only is this a great place to be, it finally—after some years of commuting—brings me back to my family. What a gift, in more than one way.
I also got my ERC Starting Grant founded (after the previous one was put on the wait list) and successfully transferred, thanks to my lovely grants office and the Swiss State Secretariat for Education, Research, and Innovation (SERI). What a perfect way to start a new position! I am still smiling as I reflect on this.
Moreover, I finally got around to calculate how often our work has been rejected, so here we go for 2024:
- Metric Space Magnitude for Evaluating the Diversity of Latent Representations: Rejected once from ICLR (in a previous version) and once from ICML.
- On the Expressivity of Persistent Homology in Graph Learning: Rejected once from NeurIPS.
- CliquePH: Higher-Order Information for Graph Neural Networks through Persistent Homology on Clique Graphs: Rejected once from KDD and once from NeurIPS.
- Bayesian Computation Meets Topology: Rejected once from ICLR, once from NeurIPS, and once from ICML. One of these rejections was very odd because the paper received good scores, but let us not dwell on that.
- The Magnitude Vector of Images: Rejected once from ICLR.
- All the World’s a (Hyper)Graph: A Data Drama: Rejected once from NeurIPS.
All other papers in 2024 have been accepted directly, but regardless of how many attempts it takes, any accepted paper is a stroke of good fortune, given that the reviewing process is fraught with high variance. Maybe 2025 becomes the year in which we finally work towards improving the reviewing system for good…
In any case: Wishing everyone the best for 2025!