October 2026
- Posit Conf 2026 took place in October. From the times when it was RStudio Conf, it happened to become the major global R syncing event. Now is a good moment to catch up on announcements and talks. https://lnkd.in/p/eEYfAN9a
- The R ecosystem saw an explosion of package & environment managers in the last years: {renv}, {}rix}, rv, uvr and ir – just to name the most notable ones. Isabella Velásquez overviewed them in a nice systemizing blog post https://rworks.dev/posts/r-managers
- Polars 2.0 is out. The Rust-based dataframe library hits a major milestone. Curiously (and this links to our previous talk in {qqsl}), one of the major news is that now SQL is a first-class citizen on this hype train https://pola.rs/posts/release-polars-2
- Some (regular, already) magic from James Balamuta – turn an R project into a standalone website that runs R, serverless. {webrarian} packs an R project so the analysis runs live in the browser — no server needed. https://lnkd.in/p/e4X9XBNK
- Quarto is being rebuilt in Rust. So far, this is an experimental Quarto 2 (q2) implementation, but this is likely an early signal of where publishing tooling is heading – towards even faster and more reliable execution https://lnkd.in/p/e9ebVZez
- I thought I would never stop recommending the Hard Fork podcast. And I won’t. Only, the original hosts are moving to create their own new podcast called Machine Gods, to be launched on 19th October. At the same time, Hard Fork continues at NYT with new hosts, and it’s also good. But I heartily encourage you to follow the new show of Kevin Roos and Casey Newton – I find it being the single most useful source in attempts to digest the ongoing tsunami of AI related news. I’m linking Casey’s final reflection on Hard Fork and the transition to independent Machine Gods podcast https://www.platformer.news/hard-fork-machine-gods/?ref=platformer-newsletter
June 2026
- On Tuesday 25th August 2026 at 14:00 [NOTE unusual day/time!], we are going to have a special 4R/um meeting – we are joining forces with Rasmus Bååth, the organizer of CopenhagenR useR Group via Meetup. We are going to have 3 ~20min presentations, 2 from guest speakers and 1 from an internal DST speaker. More info will follow after summer
- REMINDER: On 25th September 2026 we are going to have a very special guest at 4R/um – Thomas Lin Pedersen, one of the most influential R developers who works at Posit, the author of {ggforce}, {patchwork}, {ragg}, {ggraph}, {ggfx}, the redefined {gganimate}, maintainer and lead developer of {ggplot2}, and many many more great packages. He will present his latest major pivot – {ggsql}. Please mark in your calendars, come, and bring your non-R colleague who works with SQL – https://ggsql.org
- Hadley Wickham announced getting back to his Substack blog. And of course you can see that the dominating topic is the ongoing AI revolution.
- {tidypredict} is a package of the established tidymodels family that follows the idea of SQL revival and enablement. It allows fitting statistical models directly on the database – https://tidypredict.tidymodels.org/
May 2026
- Lean R 4.6.0 installation is live and flying – try it, you would really feel the speed boost, clearly visible even from the startup time, packages panel is usable, autocomplete is not lagging =)
- On 25th September 2026 we are going to have a very special guest at 4R/um – Thomas Lin Pedersen, one of the most influential R developers who works at Posit, the author of {ggforce}, {patchwork}, {ragg}, {ggraph}, {ggfx}, the redefined {gganimate}, maintainer and lead developer of {ggplot2}, and many many more great packages. He will present his latest major pivot – {ggsql}. Please mark in your calendars, come, and bring your non-R colleague who works with SQL – https://ggsql.org
- Another internal news. On 8, 9, and 12 June at 9:00–12:00 I will be teaching Dataviz with R course. [Note: 3 days instead of 5 claimed in the initial dstalle email] Apparently, I overestimated the demand for the course, and we are not at risk of not fitting in the room =) this means that you may safely plan to attend one, two, or all three of the workshop days. The three days are planned as follows: 1) basic dataviz principles & recap of tidyverse; 2) intro to ggplot2 and a deeper dive into ggplot2 specific useful plot types; 3) maps in R + interactivity. Please feel free to join any of the days but do write to me (ikx@dst.dk) at least a day in advance so that I can properly plan the workshops. More info on the course and all the R scripts are here – https://github.com/ikashnitsky/dataviz-art-skill
- A balanced, thoughtful, well-argumented essay on the tradeoffs of choosing R/py vs SAS in pharma, a highly regulated industry with deep roots in SAS usage. The difference of official statistics from phara is only in the level of consolidated collaborative effort in building and maintaining open-source ecosystem – https://www.appsilon.com/post/r-python-sas-modern-sce-comparison + one example of SAS v R in production
- Typst is maturing into a powerful document authoring engine with a first-class support in Quarto – https://www.linkedin.com/posts/veerlevanleemput_lots-been-going-on-around-typst-lately-new-share-7457388222441062400-tRVe
- Andrey Karpathy’s fresh take on creating and maintaining an AI-driven knowledge base as a wiki of simple .md files caught the industry by surprise with it’s simplicity and power. Here is an R package {pensar} that implements the approach in our most cozy environment – https://cran.r-project.org/package=pensar
- You may have noticed that every newsletter features something to do with duckdb, the surprising disrupting technology in bigdata heavy-load computing. (Just have a look at this illuminating blog post). One field where the benefits of blazing fast processing are most notable is spatial analysis. And here comes the stable 1.0.0 release of the framework package {duckspatial} – https://www.linkedin.com/posts/adrian-cidre_duckspatial-v100-is-now-on-cran-ugcPost-7444462268743966720-_jRL
- The concept of skills.md for AI agents becomes a dominant field of tuning agentic AI systems to work productively and predictably. Look at the collection of skills tailors specifically for R – https://www.linkedin.com/posts/david-underwood-416a4513_a-few-claude-skills-for-r-users-r-works-share-7455232673876430850-2-nn + for {tidymodels} https://skills.tidymodels.org/
- Surely you’ve heard multiple times the claims that “AI will soon take our jobs”, probably often enough to dismiss it as fearmongering. The key word here is “our”. And the key question that likely worries everyone is: Will it soon grab my job? What’s really happening in job displacement today? Here’s he single most useful AI job revolution explainer https://sequoiacap.com/article/services-the-new-software
- Addy Osmani, head of Google Cloud AI team, wrote a beautiful essay on the distinction between letting AI helpers strengthen one’s working ability vs allowing it to displace yourself. He introduces a useful concept of “cognitive surrender” a provides a set of actionable tips how to make sure that AI helps you develop/grow – https://addyosmani.com/blog/cognitive-surrender
- Another great essay on the related topic – From technical debt to cognitive debt – https://margaretstorey.com/blog/2026/02/09/cognitive-debt
- AI coding agents/helpers have already redefined the programming world. Here is a recent real-world example that is nice to have at hand the next time someone would recycle the “stochastic parrots” argument staright back from 2023 – https://www.linkedin.com/posts/kirill-balakhonov_to-me-this-is-the-most-epic-coding-ai-story-ugcPost-7460692232183025665-fRYh
- This is a fun one. Imagine you traveled in 1930 and can ask the people of that time whatever. You have a chance with Talkie, a research project that trained an LLM only based on data before 1930 – https://talkie-lm.com/introducing-talkie (listen to the researcher who develops the project in last week’s Hard Fork episode)
March 2026
- With the recent {dplyr} major update, if_else() and case_when() are up to 30x faster — a dedicated post explains the memory and speed engineering behind the dplyr 1.2.0 gains – https://tidyverse.org/blog/2026/02/dplyr-case-when-faster
- RStudio rolled out its own AI coding agent. Unfortunately, it’s a paid plan with 14 days trial. But this may be transformative for those not switching to Positron/VSCode https://posit.co/blog/introducing-ai-in-rstudio
- Tad – an extremely capable small app to view CSV, parquet, duckdb, and SQLite. Open source, cross platform, fast, clean. I didn’t flinch at my request to open a 1 million rows 500MB csv file. But that’s not all: it can blazingly fast create pivot tables to explore your data – https://github.com/antonycourtney/tad
- Quarto v1.9 – a fresh Quarto release lands with notable Typst improvements. Typst is a new generation tool to compile PDFs that unlike LaTeX manages in seconds rather than minutes – https://quarto.org/docs/blog
- {webRoid} and {webRios} – two packages that allow you to run R right on you mobile phone. Yes, I know this is likely not for everyone =) but this is great option to have, especially since Android is drifting toward becoming a fully fleshed desktop OS. Did you know you can connect to a DST virtual machine via Android Microsoft app then plug the phone to you docking station and just work? – https://www.linkedin.com/posts/steven-mortimer_rstats-webroid-webrios-activity-7419714410864199680-9n2X
- {ragnar} 0.3.0 — the tidyverse’s RAG (retrieval-augmented generation) package keeps evolving: faster document ingestion, new embedding providers, improved retrieval, and new integrations for using ragnar stores directly from tool calls in LLM agents – https://tidyverse.org/blog/2026/01/ragnar-0-3-0
- {orbital} – a new development in the world of data science led by Posit – translate model creation from R or python to bare SQL and train the models right at the database. This one is just to observe and follow https://www.linkedin.com/posts/hadleywickham_orbital-is-a-such-tool-technology-build-share-7438300413973651456-RFpz
- Posit evaluation of small AI coding models – https://www.linkedin.com/posts/simonpcouch_i-had-wondered-how-todays-gpt-54-releases-share-7439764257778409472-A9-F
- A great discussion of how AI coding redefined the unspoken agreement between the actors in the Opensouce world – https://turso.tech/blog/what-happens-with-oss-in-the-age-of-ai