ChatGPT predicts which STOC 2027 paper submissions will be accepted (chatgpt.com)

ChatGPT has leveraged its predictive capabilities to analyze the 1,665 paper submissions for the upcoming STOC 2027 conference, focusing on authors' publication histories to forecast which papers are likely to be accepted. This year is particularly noteworthy as STOC implements non-anonymous submissions, publicly posting papers and allowing a maximum of five submissions per author, with acceptance decisions due by February 1, 2027. ChatGPT emphasizes the importance of authors' track records while acknowledging that a strong history alone does not guarantee the novelty or significance of their contributions. Among the standout predictions are three promising submissions: first, paper #1467 about the k-server conjecture, co-authored by seasoned researchers known for their prior success at STOC; second, paper #817, which claims groundbreaking results on splay trees, featuring a team linked to the renowned Robert E. Tarjan; and third, paper #29, which presents a near-optimal algorithm for the Subset Sum problem. The predictions highlight the relationship between an author's experience and their project's potential impact, although ChatGPT notes that clarity and verifiability of proofs—key criteria emphasized by STOC—remain unpredictable elements outside of author histories.