Covid 疫情让会议停摆「超过一年」，迫使数学家改用 Zoom，凸显实体交流的不可替代；而近年像 ChatGPT 这类工具，也可能让人以「快速聊天」取代敲门找同事，进一步稀释偶遇带来的创造力。

文中用时间线展示偶遇如何推动成果：Oberwolfach 在 2020 年初仍以有限 Wi‑Fi 与共餐促进讨论；2019 年 Kathryn Mann 与 Thomas Barthelmé 咖啡谈话引出多篇论文，并在 2023 年报导中呈现；一条未发表的想法从 2000 年代、2017 年生日派对口耳相传，直到 2024 年在多伦多吃饺子时接力，让原本想放弃的问题终于完成证明。

The piece argues that major mathematical advances are usually social: even “single-author” results often trace their key insights to visits, conference talks, or seemingly irrelevant conversations. Breakthroughs frequently happen in person wherever mathematicians gather, not in isolated late-night epiphanies.

This became clearer when Covid shut conferences for more than a year, forcing collaboration onto Zoom and away from shared chalkboards. Today, tools like ChatGPT could further reduce spontaneous hallway chats if researchers choose quick AI conversations over knocking on a colleague’s door.

A recurring catalyst is food and coffee. Oberwolfach’s early-2020 workshops highlight communal routines; a 2019 coffee chat between Kathryn Mann and Thomas Barthelmé sparked many papers, reported in 2023. Another proof depended on an unpublished early-2000s insight, passed along at a 2017 birthday party and then relayed over dumplings in Toronto in 2024—turning near-abandonment into completion.