Let’s stop writing papers…

Patrick Morris, Ramón y Cajal Fellow (Post-doc) at Universitat Politècnica de Catalunya (UPC), Barcelona

SERIES · AI is making papers cheap to produce but they remain costly to referee. Proofs and Prompts is publishing a series of posts from 26 October on the future of publishing, papers and credit. Questions such as how to disclose AI use, who bears the cost of checking, what authorship means and what papers of the future might look like will be addressed. This post is part of the series. Each is written independently and reflects only its authors’ views. All posts in the series

AI Disclosure: All sentiments and words are my own. I used Chat GPT 6 medium to suggest other essays discussing changes in mathematical culture and to inform myself about initiatives. Another chat was used to understand the process of atlasing, after Google AI suggested it as a term that is ‘like mapping out but starting with an a’ (which admittedly I probably could have got to myself if I had thought for a moment).

In this essay, I put forward some ideas on what the future of mathematical culture could look like as well as thoughts and concrete suggestions for how we get there. My focus is centred around a more diverse ecosystem of mathematical output, moving away from traditional practices of writing and submitting papers to journals. My vision is aspirational, formed through a fair amount of dreaming, as well as insights from others. Through my research, it became clear to me that as a community we have vastly different views on how we approach the use of AI in maths as well as its current and predicted capabilities. As much as possible, I have tried to envisage a robust community that can adapt to developments in AI whilst making space for people to contribute in the way they feel most comfortable.

Stop writing papers, really?!

My main aim is to provide some direction and cultivate positivity around the future of mathematics. Before that though, let me address what I believe we should not do. The title of this essay is purposefully incendiary and stems from a deep feeling that we cannot simply try to do business as usual. The pressure to publish has been the prominent driving force in most mathematicians’ careers, often drowning out the rest of what we do. Arguably the community was long overdue some introspection but the advent of AI gives us no choice.

The situation requires a monumental shift in how we approach mathematics as well as a complete reform of our value system, rather than clunky attempts to try and accommodate new advances into the status quo. (I realise that this view is not held by everyone, but it is beyond the scope of this essay to convince people of this ¹). This is why we urgently need to stop writing papers and focus on other activities.

In reality, I know that this is idealistic and that there are many obstacles to change. Our institutions are built around the current system and will be slow to adjust, a problem that disproportionately affects junior academics and especially students. I am a post-doc and am acutely aware that my publication list will remain the most influential factor in obtaining a permanent job and grants in the coming years. If I do get tenure at my current university, as I hope I will, my pay is affected by how many papers I write.

There is another issue which is simply that this is what we know. I myself will contradict my own advice. I have a pile of unfinished projects that I plan to complete for a variety of reasons, some because I am invested, some because I feel that I owe it to younger colleagues who will benefit from the experience and just one that I believe provides new ideas that cannot (currently) be obtained by pushing a button. Nonetheless I believe I am not alone in at times feeling quite lost² over the last months and I hope that my suggestions can provide some purpose moving forward.

Ok, so let’s at least write fewer papers. What should we do instead?

Below I lay out different axes of activities (in no particular order), briefly explaining how we currently approach them and some ideas of what we can do differently. All of them are broadly aimed at advancing and facilitating human understanding of mathematics.

1. Atlasing: In neuroscience³, atlasing refers to creating models of the brain that provide a standardised framework for many individual brain maps. I like this as a metaphor for the process of collating and canonising different mathematical works. This has typically been done through books and surveys, which remain incredibly valuable resources. One particularly ambitious example of this was The Princeton Companion to Mathematics which aimed to overview all of modern mathematics. Moving forward we should be more systematic and dynamic in our efforts to atlas. I don’t think it is so crazy to aim for a comprehensive online mathematical encyclopaedia that catalogues topics, theorems, concepts, proof methods and their interconnections and which is continually being expanded and refined. On the formal level, this is already happening⁴.
2. Big Mathematics: This is a term used by Tao to mean⁵ ‘large-scale decentralised collaboration between humans and machines’. I would like to slightly broaden this definition by not imposing the use of machines. Indeed, I see the crowdsourcing Polymath project as an innovative example of big mathematics. Once we move away from a system that lauds individual ownership of results, we can become more open and collaborative on a large scale. I would love to see more accounts of failed attempts or half-proven results, as well as logs of what cannot be currently achieved by AI. In this way, I think we will save a lot of duplication by people working in secrecy.
3. Curation: By definition, to curate is to select, organise and present. Whilst similar to A, I would contrast them by placing A as an attempt to create permanence, whilst C is concerned with processing the torrent of brand-new mathematics. Good curation picks out results/progress, places it in a wider context and tells you why you should care about it. This is a role that has been underserved with the exceptions I know of being some blogs⁶, Quanta magazine and the Bourbaki seminar (and to some extent MathReviews and workshops on current trends). With more maths being generated and the fall of many guiding conjectures, we will need more curation. This can take the form of podcasts, blogs, magazines and newsletters both for maths as a whole and for specialists in its many corners.
4. Dialogue: Mathematicians talk to each other all the time, whether sharing ideas at a blackboard, debating our culture over coffee or catching up at conferences. These ongoing conversations are deeply useful but are mostly localised. Our global community also needs spaces to discuss and debate more widely and they are lacking⁷. Comments sections on blogs such as Bloom’s Erdős problem website have almost accidentally become forums. We should create avenues for meaningful conversations, both about maths itself and about the culture. For example, a repository where people can share feedback, questions, answers and comments on papers/preprints would surely be useful. Maybe even a mathematics focused social media or a shared discord or slack. I don’t mean that we should move completely online. On the contrary, I think we should also create more seminars, reading groups and round-table discussions. There is space for both to flourish.
5. Exposition: This one has been grabbing a lot of attention⁸ due to AI’s current inability to present ideas well. If we’re honest with ourselves, we were also lacking in this as humans. I know plenty of highly influential papers that are terribly written and a bunch more that contain some beautiful idea that is completely obscured by the technicalities needed to get the proof over the line. The situation with talks is even worse, with the plenary lectures at our biggest conferences essentially being awards for research and with no expectation on the speaker or from the audience to understand after 5 minutes in. On the flip side, there are certainly some amazing storytellers in our community and reading their work or seeing them speak has made me realise what an art it is. It is exciting to see what we can achieve in terms of exposition when we give it the real weight and respect it deserves. It can and should come in many different forms and for the full spectrum of audiences, from experts, to students and researchers in other areas, to the general public.

Ok but what should we actually do now?

The areas outlined above are not new but need much more attention and, most importantly, the development of platforms and infrastructure to highlight and reward them. This is the time to innovate and it is happening already⁹. Proofs and Prompts, for example, is a great example facilitating D and the new video journal Mathematical Discourse feels like an important move towards developing E. Further initiatives include the Mathathon coding project to develop new proofs and expositions of known results (B,E), the open-source mathematical formalisation project Mathlib (A,B) and Transformation, a new weekly digest from Quanta tracking news in maths and AI (C).

I hope to see many more exciting projects come to fruition and I believe we all have a part to play in revolutionising what modern mathematics looks like. In the past, we have placed a great deal of responsibility on relatively few, those that have established themselves to the extent that they can take time away from proving theorems to serve as editors, give colloquia and write books. These people have an important role to play in providing leadership, expertise and stature to new initiatives but we cannot rely solely on them to build everything, there is simply too much to do. Moreover, we should take this opportunity to diversify the voices at the table. In particular, young researchers as well as groups that are underrepresented in our current community will bring vital new perspectives to the conversation. I encourage everyone to think about how the future can look and how they want to contribute to it. Apply for grants to help realise your ideas and support and contribute to the ideas of others.

We should think with longevity and establish long-term alternatives to traditional journals. We can borrow from what worked before and aim to incentivise contribution by creating venues for other types of mathematical output that are competitive and prestigious. However, we should also explore new ways of rewarding output. We can look at measuring forms of engagement from the community that go beyond citations, maybe through views or voting systems (as used by Math Overflow, for example). Learning also from our mistakes in the past, we should avoid outsourcing ownership of our infrastructure and should celebrate community-led alternatives.

It is incredibly difficult to predict the future of maths but for now, developing along axes A-E represents a tremendous workload of things that we can be focusing on, all of which I can only see as being beneficial.

Acknowledgements: I am very grateful to my many colleagues whom I have spoken with at length about this topic. In particular I thank Richard Lang, Miquel Ortega and Tibor Szabó for their insights and Tássio Naia and my dad, Tim Morris, for their comments on an earlier draft. I also thank my wife Jiselle for her patience and help in processing my thoughts and writing in a format I am not so used to.

1. Many eloquent arguments can be found elsewhere, see e.g. this essay of Bryna Kra, Daniel Litt’s The end of mathematics and Amit Harlev’s Dear conferences, journals, and hiring committees: don’t wait for everyone else. ↩
2. Kirwin Hampshire’s The Dark Night of Mathematics captures the feeling beautifully. ↩
3. See, for example, A. W. Toga and P. M. Thompson, Maps of the Brain, The Anatomical Record, 265(2), 2002. The notion also appears in topology and differential geometry where an atlas is a collection of coordinate charts that covers a geometric space. ↩
4. The mathlib Community, The Lean mathematical library, arXiv:1910.09336, 2019. ↩
5. According to Terence Tao on AI in mathematics (and beyond) at the time of writing. He also talks of Math 2.0 and highlights many interesting examples and ideas for rethinking mathematical research. ↩
6. Terence Tao’s and Gil Kalai’s spring to mind for me. ↩
7. Maths Stack Exchange and Math Overflow are notable exceptions but they seem to be in decline, sadly. ↩
8. See e.g. Grant Sanderson’s essay If math is more than proof, we need to better celebrate the rest of it ↩
9. There are also some examples of older initiatives that were ahead of their time and have sadly stunted. Tricki, for example, aimed to create a database for problem solving techniques (A,B), see also the incredibly ambitious The Mathematics Survey project of David Aldous and Jim Pitman (A,E). Maybe it is time to revisit these ideas. ↩

---

Received 19 October 2026.