Simon P. Couch

commons agents prefer to answer data questions by invoking code that data scientists already trust and maintain--R/Python packages, Quarto docs, Shiny apps. If they can't find relevant code, they can fall back to writing SQL/R/Python "by hand," and that decrease in trust is signaled to the user.

I made an AI video about genzplyr and I lowkey love it. It's also 100% slop. In this post I dig into what makes something slop, when it's ok to share it, and how to avoid it in work that matters. Read more at: substack.hadley.nz/p/what-is-slop

New on the @posit.co open source blog: if you step away from your coding agent long enough to make a coffee, your next message could cost ~16x as much☕ @sara-altman.bsky.social and I wrote about token caching and AI costs: opensource.posit.co/blog/2026-10...

an outdated system in a salmon farm using PHP was found to have been breached, allowing the attackers to perform remote cod execution

i maintain that MLE for "ML engineer/ing" was mostly meant to provoke statisticians

In case you weren't able to make it, @sara-altman.bsky.social and I wrote up the talk for the @posit.co OSS blog. We talked about the rise of "vibe analysis," the pitfalls of human-in-the-loop, and some new software from Posit. opensource.posit.co/blog/2026-10...

“You are infringing on the time and brainpower of people by pushing "your" content that little/no of your brainpower went into” is such a banger

Such an honor to keynote at posit::conf(2026). I still vividly remember the first conf keynote I ever saw, when @jennybryan.bsky.social spoke about debugging in San Francisco. Very grateful for the opportunity (especially to present alongside @sara-altman.bsky.social)!

Some reflections on one of the first #rstats packages I wrote: simonpcouch.com/blog/2026-10...

New in querychat: multiple tables, chat history, data-dict, and easy extraction of findings into Quarto, Shiny, Jupyter, etc. querychat is now much more capable beyond just querying "pre-defined" Shiny outputs. Available for both #rstats and #python. opensource.posit.co/blog/2026-09...

FYI Posit AI Pass subscribers; we wound down the instance serving Gemma 4 today. If you were a regular user, you might give GLM 5.3 Flash a try; it's vision capable, quite a bit larger and newer (and thus more capable!), and served at ~half the price that Gemma 4 was per-token!

Wicked psyched to share that shinychat v0.5.0 #RStats and v0.7.1 #Python are now available! Check out all the rad new features on the Posit Open Source blog. #PositConf2026 opensource.posit.co/blog/2026-09...

Every time @simonpcouch.com makes a great point in his #positconf2026 Keynote, his earring sparkles extra brightly in the stage lights. 🌟

We're excited to share commons, an R and Python package that helps data scientists build trustworthy data agents for their organizations! Read more on the @posit.co open source blog: opensource.posit.co/blog/2026-09...

In this edition of My Coworkers Are Awesome, @hadley.nz had a real-life version of the blob on the chores sticker made!

#PositConf2026 (via the #RPharmaSummit2026 pre-game) has already been worthwhile for introducing me to Kimi K3 in Posit AI Pass! So far it's good, fast, and relatively cheap. This is what I get for skipping any of @simonpcouch.com 's blog posts! opensource.posit.co/blog/2026-08...

What a lineup!

"For a while, I must admit, it looked as if software developer roles like mine were done for. [...] But our industry is slowly realizing that making truly cutting-edge software still requires humans to think and work together, to maximize their skill sets and to practice their respective crafts."

I AM SO EXCITED FOR POSIT CONF

Crossposting to the swarm artifactory instance

RUN ML MODELS DIRECTLY IN YOUR DATABASE

Can't make it to Houston? Virtual passes for posit::conf(2026) start at $99, with academic and nonprofit rates at $49 and a needs-based option at $0. Every keynote and session livestreamed, September 14–16. conf.posit.co

New on the @posit.co blog: In my experience, fine tuning LLMs is more difficult and more expensive than I've seen it made out to be. In most cases, I'd recommend focusing your efforts on experimenting with a bunch of prompts and existing models. opensource.posit.co/blog/2026-09...

5 months ago, models in the ~30B A3B range started to be able to do simple agentic coding tasks. I recently wondered how 8B (or even 4B!) models would fare: simonpcouch.com/blog/2026-09...

GLM 5.3 Flash is really fun. Super zippy and super cheap, and it’s great at getting things done. Pair it with a smart model or a smart human and you can get places fast!

How do you choose an LLM? A survey of the current model landscape, including results from the vibes-from-Sara-and-Simon eval (/s) New on the @posit.co open source blog: opensource.posit.co/blog/2026-08...

I've been really impressed with Kimi K3 and GLM 5.2 the last couple weeks. I'm excited to share that we've made the models available in Posit AI with strong privacy guarantees and a steep discount compared to the Claude models currently available in the service! opensource.posit.co/blog/2026-08...

i was looking up song lyrics and

Data analysis is often a branching and nonlinear process. We just shipped a feature in Posit Assistant to help with this; use /eda-log to keep track of loose ends in your analysis. @posit.co #databs #rstats opensource.posit.co/blog/2026-07...

this got old so fast

We're happy to announce our new #rstats package: lorax. If you fit tree-based models, lorax enables conversion to the party package's format (for making nice tree diagrams) and has methods to compute which features were used, the number of terminal nodes, etc. opensource.posit.co/blog/2026-07...

I am begging you to talk to other human beings directly, and not through a Claude layer. This goes doubly when I ask you, directly, to talk to me not through a Claude layer. Please.

A new release of mcptools, an #rstats package implementing the Model Context Protocol, is now on CRAN! It's a patch release with several security-oriented fixes. github.com/posit-dev/mc...

This week on my substack: a write up of my talk "y code when ai" (my thoughts on why knowing how to code is still relevant), along with a brief description of how I turned a video of my talk into this post. Read it at open.substack.com/pub/tidydesi... !

It’s Friday. Do a breakthrough.

I'm happy to share `ir`, a new command-line tool for running portable R scripts and Quarto documents. It's inspired by my two favorite parts of `uv`: self-describing scripts and tools you can run without installing first. opensource.posit.co/blog/2026-07...

Just ran today's Gemini 3.6 Flash release through #rstats bluffbench2. In the ballpark of Gemini 3.5 Flash on performance, slightly (5-10%) cheaper. More on the eval: posit-dev.github.io/bluffbench2/

One of the more interesting findings to me from this eval is that prematurely adding modeled results (like #rstats geom_smooth()) to plots drastically reduces the chances that models will catch the issue in the plot.

We're excited to share bluffbench2, a next-generation data science LLM evaluation. The eval measures how well models notice subtle data quality issues in plots. Even the most capable models today score in the mid-teens. opensource.posit.co/blog/2026-07...

This must mean that vibe-coded UIs were common in the human ancestral environment

httr2 1.3.0 is out now! Includes a bunch of improvements to OAuth token caching and summarises some of the biggest changes in the last few patch releases — opensource.posit.co/blog/2026-07.... #rstats