RM

Ron Minsky

Things Ron Says on Podcasts

Jane Street engineer who hosts Signals and Threads, the firm's engineering podcast, where he interviews colleagues and outside guests about programming, tooling and systems. Recent episodes cover agentic coding platforms, testing, build systems, networks as programs and teaching in the age of AI. Has long written for the Jane Street tech blog under the name Yaron Minsky, on OCaml and the firm's engineering practice.

“The agents are kind of amazing, and one of the things that makes them amazing is also one of the deficiencies, which is they are perpetual beginners, right? They never learn how to use the tools. They rediscover every time how to use the tools. And so you kind of have to build your whole system around them being these kind of perpetual beginners.” — Ron Minsky, Signals and Threads “The whole existence of AI agents has kind of broken the social contract around code review. It used to be when someone would hand you a feature, it made more sense for you to help get that thing over the line, because they put in a lot of sweat equity. And now all you need is a prompt and a dream, and you can for free create a plausible-looking problem, but it's still hard work to verify that it makes sense.” — Ron Minsky, Signals and Threads “The overlap of problems that are easy enough that they are suitable for someone who is learning to do, and hard enough that an LLM can't just one-shot it, is close to the empty set now.” — Ron Minsky, Signals and Threads

Where to Find Them

Ron Minsky hosts Signals and Threads .

Recently: “The Changing Craft of Programming with John Crepezzi” on Signals and Threads (October 2026); “Learning Goals and the Goals of Learning: Teaching in the Age of AI with Aaron Bauer” on Signals and Threads (September 2026); “Wrestling the world into rows with Eric Mannes” on Signals and Threads (September 2026); “The Network as a Program with Nate Foster” on Signals and Threads (June 2026); “Why Testing is Hard and How to Fix it with Will Wilson” on Signals and Threads (March 2026); “State Machine Replication, and Why You Should Care with Doug Patti” on Signals and Threads (April 2022).

What They Said

“The agents are kind of amazing, and one of the things that makes them amazing is also one of the deficiencies, which is they are perpetual beginners, right? They never learn how to use the tools. They rediscover every time how to use the tools. And so you kind of have to build your whole system around them being these kind of perpetual beginners.” — Ron Minsky, Signals and Threads

Minsky describes how a small web app that takes an agent 30 seconds as a single HTML file takes about 10 minutes inside his firm's internal build and code review tooling. His conclusion is to pre-build deterministic setup so the agent doesn't have to relearn the tools every time.

Signals and Threads · 2026-10-16 Permalink → Listen →
Signals and Threads Around 1:13:16 into the episode
Ron Minsky

…is our code review system that has to be created and a build process that has to be kicked off and there's just like more steps and tool calls and the thing that actually represents there's like 30 seconds of like actual work yeah to go and like generate the code you probably want to grab like

John Crepezzi

the bonsai starter template and like get your code and off to a good start and

Ron Minsky

so like the whole thing in reality takes 10 minutes what it would have taken 30 seconds in the pure index.html world and i think at first we were like oh maybe this just like the cost we're paying because of like internal languages and libraries like no actually this is maybe stuff that we can fix by exactly as you're suggesting doing like more like hard deterministic setup stuff so that like things are just there and ready and like the agent doesn't have to go and discover again how to use all the tools that's again the kind of like the the agents are kind of like amazing and one of their thing that makes them amazing is also one of the deficiencies which is they are perpetual beginners right they never learn how to use the tools right it's just they rediscover every time how to use the tools and so you kind of have to build your whole system around them being these kind of perpetual beginners

John Crepezzi

right yeah i think another part of this is you can maybe like instill them with a little bit of repeated knowledge by like writing skills right but then there's a question uh so skills for anyone that doesn't know are these like little documents that you write to kind of guide the model on things…

Ron Minsky

yeah actually we've talked a little bit about things like skills we talked about function calling there's like we built our own apis for it there's apis like mcp which have come out in the outside world like which became public and popular after we'd already built our own version of it skills is…

John Crepezzi

nice part about having our own tools right is that we can like if we see a need for something we can go make it um but then there's also this thing you have to do sometimes of being like okay well the outside world has really kind of set their standard against this for example skills and like…

Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.
“The whole existence of AI agents has kind of broken the social contract around code review. It used to be when someone would hand you a feature, it made more sense for you to help get that thing over the line, because they put in a lot of sweat equity. And now all you need is a prompt and a dream, and you can for free create a plausible-looking problem, but it's still hard work to verify that it makes sense.” — Ron Minsky, Signals and Threads

Minsky and Crepezzi are discussing how coding agents change code review. Users now arrive with finished code instead of feature requests, which makes it socially harder to say no to a change that shouldn't exist.

Signals and Threads · 2026-10-16 Permalink → Listen →
Signals and Threads Around 1:46:08 into the episode
Ron Minsky

…even though it's not in an ideal shape. But if it's much easier to take those changes and improvements and turn them into a reality, then you can end up with higher quality code that comes out of it. So I do think a lot of what comes out of all this technology depends on what we make out of it.

John Crepezzi

…know maybe your user wants a thing that you think shouldn't exist but now they're not coming to you with like a request they're coming to you with with code and it feels like a little bit harder socially to be like no we don't want it when the thing they're presenting is kind of a finished product

Ron Minsky

that's right like there's a way in which the whole like existence of AI agents has kind of broken the social contract around code review where it used to be when someone would hand you a feature like you were you know it made more sense for you to like help get that thing over the line because like they put in a lot of work a lot of sweat equity that went into creating that and now like all you need is like a prompt in a dream and like you can like for free create a plausible looking problem but it's still hard work to verify that it makes sense and like you don't want to like add lots of things to a system that will make it overall like messy and complicated even if the code is good it might be that it makes the system that you're building conceptually complicated for the end user right if everyone gets every knob that they want like you just get too many knobs yep and so yeah figuring out like what's the appropriate kind of pushback right and it's just like the shape of it is different like you've always had to deal with these kind of questions prioritization questions have always been hard but they now just present themselves in different ways and i think it's just something that we have…

John Crepezzi

yeah i think that's right one framing i've found helpful that i've heard recently is just think of like they are coming to you with code they probably put some time into it think of it as a prototype and then just still have the discussion about whether or not the thing is warranted yeah it's like…

Ron Minsky

right and the prototypes can be amazing right it can be like someone has an idea of what to do and there's like the prototypes they hand you but the amazing thing is all the prototypes they threw away like they tried a thing and you're like yeah this sucks and they get rid of it it's like you know…

John Crepezzi

thank you for having me

Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.
“The overlap of problems that are easy enough that they are suitable for someone who is learning to do, and hard enough that an LLM can't just one-shot it, is close to the empty set now.” — Ron Minsky, Signals and Threads

Minsky and Aaron Bauer were discussing how Jane Street teaches early-career engineers now that an LLM can write the scaffolding for any exercise they set. Minsky's worry is that struggling with the material is a core part of learning, and that the exercises which are both small enough to teach with and hard enough to resist a model have nearly vanished. His tentative answer is that education sometimes works by withholding affordances — like making people do it on paper.

Signals and Threads · 2026-09-23 Permalink → Listen →
Signals and Threads Around 1:07:09 into the episode
Ron Minsky

The initial scaffolding can be made really easy, right? The models can do a decent job of that. But what's the understanding that you have to give to people?

Aaron Bauer

…these tools effectively, about how to integrate them with their workflow, about how to understand the ways in which they can be effective or lead you astray. And I think this is still in very early days. I don't think we have sort of settled answers to this. And the tools are changing very quickly.

Ron Minsky

But you can already see problems, right? There are definitely people who run into problems where they just like kind of vibe it up too much and like use the LLMs to generate stuff and do not think. And like that's a real problem. Like you still need to think. I actually worry more generally about what it means for the overall education system. Like I think a kind of critical component of learning is struggling with the material. And the world where you always have a genie at your beck and call that you can just like summon up to solve the problem is itself like a real problem. And the overlap of problems that are easy enough that they are suitable for someone who is learning to do and hard enough that an LLM can't just one-shot it is like close to the empty set now. So it's like a really challenging problem. I mean, it really does make me just think back to the like doing stuff on paper without a computer. Sometimes an important thing about making the educational…

Aaron Bauer

Yeah, I think you and pretty much everyone else involved in education is deeply worried about this. To your first point about like, how do you get something that's small enough to be tractable for students and large enough to not be trivial with an LLM? Like we have run into this exact problem…

Ron Minsky

Yeah, and I think it'll be interesting over the next few years as we're doing this increasingly with cohorts of people who have not spent that much time. Much time typing code themselves. They'll just be like, there's like a core learning mechanism that maybe they haven't had enough experience…

Aaron Bauer

Yeah, there's been a lot of conversation about things along the lines of that cloud code learning mode, where what LLM tools, if any, should we provide to folks going through OCaml boot camp, learning OCaml for the first time. And if they're just having LMs write all the code, well, it doesn't seem…

Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.

Collections They Appear In

Hands-On Engineering Podcasts Software Engineering & Infra