In 2013, the SMT-COMP organizers published a retrospective on the first eight years of the competition. At the time, "the number of solvers participating each year has been fairly constant, ranging from 9-13, with a median value of 11 participants." I thought it would be fun to look at these numbers again in 2026 to see how the community has changed over time. I also thought it would be fun to include the SAT Competition since the two communities are tightly knit.
Claude Code did the bulk of the work. I asked it to comb through the last 20 years of both competitions (21 SAT, 20 SMT-COMP) and collect every participating solver, every author of each solver, and every affiliation of each author of each solver. Both competitions do a good job of making this information available, although mostly through individual PDFs. Claude collected all the data and created a CSV file that I manually spot-checked and iterated on. The data is not perfect but should be good enough to track general trends. One important caveat: "country" here means the country of an author's listed institutional affiliation for a given submission, not their nationality.
Overall, we found 800 unique authors from 319 unique institutions across 39 unique countries. I focused on authors instead of solvers because this is a community of people and not tools. Also, I think it is a less noisy metric. For example, in 2026, Aina Niemetz and Mathias Preiner contributed to at least seven competing and non-competing solvers. I do not know if Bitwuzla and Bitwuzla-fixed should be counted as the same solver or if we should even count non-competing solvers (we do not). But it does not matter because I do know that Aina and Mathias are outstanding members of the community.
Every year there are about 5.39 more authors. In 2006, there were 55 authors across both competitions; in 2026, there were 147. Barring some minor blips, like 2011, 2017, and 2024, the growth appears to be quite steady.
A huge chunk of the growth is coming from authors at Chinese institutions—approximately 2.10 more authors per year. Other countries, like Germany, are growing but not quite as fast. About 0.88 authors from German institutions are added per year. On the other hand, participation from Canada and Italy has been roughly flat to slightly declining. There are approximately 0.08 fewer authors with Canadian affiliations per year; there are approximately 0.30 fewer authors with Italian affiliations per year.
Canada is an interesting case. In 2020, Canada and China both contributed 11 authors. Since then, Canada lost three major researchers and China tripled its participants. Fahiem Bacchus passed in 2022, Vijay Ganesh moved from the University of Waterloo to Georgia Tech in 2023, and Kuldeep Meel moved from the University of Toronto to Georgia Tech in 2024. All three submitted a solver with their students in 2020 (although Kuldeep was affiliated with NUS at the time). A very small number of researchers can have an outsized impact on the composition of the community.
Visualizing the share of unique authors per country per year makes these trends more obvious. China went from having 0% of the solver authors in 2006 to 28% of the solver authors in 2026. Germany, France, and the USA seem to be relatively stable participants while Canada, Italy, and Spain are contributing smaller and smaller shares.