{"slug": "when-the-hard-part-stops-being-hard", "title": "When the Hard Part Stops Being Hard", "summary": "A paper co-authored by Ilya Sergey, 'Tracking Borrows with Regular Expressions,' was accepted to OOPSLA'26, presenting a new type system for the Move smart contract language that uses regular expressions to capture heap reachability. Sergey reports that the formal soundness proofs, typically 80-90% of the effort in PL papers, were completed in about four weeks using a frontier LLM, and he notes that POPL submissions nearly doubled from ~350 to 600 in 2026, with most being competent research produced at ten times the usual pace. He argues that the publication culture in programming language research is shifting as AI tools reduce the need for visible human struggle, potentially making multiple single-author papers at top venues unremarkable within a year or two.", "body_md": "# When the Hard Part Stops Being Hard\n\nA few days ago, a paper I co-authored, [Tracking Borrows with Regular\nExpressions](https://verse-lab.org/papers/regex-borrows-oopsla26.pdf), was\naccepted to OOPSLA’26. It presents a new type system for Move, a Rust-style\nsmart contract language, built on a rather cute idea: using regular expressions\nto capture heap reachability. I won’t go into the technical details here. What I want to\ntalk about instead is how the paper was made and how the publication culture in\nprogramming language research has changed in the past five months.\n\nFor this paper, the part that usually consumes about 80-90% of the effort in a\nprogramming language (PL) design paper, the mechanisation of its meta-theory and\nthe formal soundness proofs, was done by one person (me) in Lean, using a\nfrontier LLM, in about four weeks, end-to-end, at the scale of a production\ncompiler rather than a toy calculus. I wrote about that experiment in [a blog\npost back in March](/2026/03/18/move-borrow-checker-lean/). 1 Back then,\nI’ve given a several talks on this effort and discussed it with a a couple of\ndozen of prominent members of the PL research community. While I received a fair\nshare of curiosity and enthusiasm, many reactions at the time ranged from\nskepticism (“the formalisation must not have been hard enough”) to outright\nrejection (“this is not how PL theory should be done”).\n\nI think those reactions say something about our publication culture. Unless research embodies a visible amount of human effort, or even struggle, it is unlikely to be taken seriously. We do love elegant ideas in PL, but we prefer them wrapped in eight to ten person-months of labour: a large implementation, extensive benchmarking, or machine-checked proofs pushed through by hand. That wrapper is now gone. An experienced researcher with a good idea can turn it into a competent PL publication in about a month.\n\nWe used to raise eyebrows at the AI/ML world, where a month-long sprint from an\nidea to a solid NeurIPS submission is, as I am told, routine. That is now our\nreality too. When I described my experiment to [Martin\nRinard](https://people.csail.mit.edu/rinard/), an MIT professor who was visiting\nNUS at the time, it took him roughly a month to fully formalise in Lean an\noptimising compiler he was working on, write a\n[paper](https://arxiv.org/abs/2607.23033) about it, and submit it to a\nconference, having never touched Lean before.\n\nI don’t think we have fully absorbed what this does to our field, but the first effects are already visible. What was surprising in March 2026 became common knowledge by July 2026: POPL submissions nearly doubled this year, from ~350 to 600. The unexpected part is that the fraction of outright AI-generated slop among them is relatively small. Most are pieces of competent research, produced at ten times the usual pace, with the tedium of proofs, implementations, and evaluation now largely automated. Having more than two papers at a single POPL, PLDI, OOPSLA, or ICFP used to signal a strong vision, a prolific group, and a wide network of collaborators. Now the same amount of research can be done by a single PhD graduate with good ideas. Therefore, I would not be surprised if three or four single-author papers at a top PL venue becomes unremarkable within a year or two.\n\nWhy only a year or two? Because the community will inevitably raise the bar,\nonce we realise that producing old-style papers (“look, I defined a toy calculus\nand proved it confluent!”) with modern tools is not a good use of anyone’s time.\nIn an age when anyone can build a\n[CompCert](https://en.wikipedia.org/wiki/CompCert) or a [seL4\nmicrokernel](https://sel4.systems/) from scratch in the proof assistant of their\nchoice, we will have to become more ambitious, and take on challenges we could\nnot have imagined a couple of years ago.\n\n-\nIn case you’re curious, the Lean development is available at\n\n[https://github.com/ilyasergey/lean-move](https://github.com/ilyasergey/lean-move).[↩](#fnref:repo)", "url": "https://wpnews.pro/news/when-the-hard-part-stops-being-hard", "canonical_source": "https://proofsandintuitions.net/2026/08/14/when-the-hard-part-stops-being-hard/", "published_at": "2026-08-14 15:05:15+00:00", "updated_at": "2026-08-14 15:13:24.280861+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-research", "ai-tools", "large-language-models"], "entities": ["Ilya Sergey", "OOPSLA'26", "Move", "Lean", "Martin Rinard", "MIT", "NUS", "POPL"], "alternates": {"html": "https://wpnews.pro/news/when-the-hard-part-stops-being-hard", "markdown": "https://wpnews.pro/news/when-the-hard-part-stops-being-hard.md", "text": "https://wpnews.pro/news/when-the-hard-part-stops-being-hard.txt", "jsonld": "https://wpnews.pro/news/when-the-hard-part-stops-being-hard.jsonld"}}