Miles McBain

I don’t think so. The vast majority of people in DS I have come across take uncertainty and its reporting seriously.

It was the boss that wasn’t fluent in uncertainty. And this I have seen a few times. Statisticians / data scientists climbing the corporate ladder high into management seems rare?

Have a look at the source of R CMD check and feel better about how you’re managing complexity.

Once again appreciating the sober takes gang. 🙏

@pivot-to-ai.com.web.brid.gy there’s a daily here for you.

ChatGPT’s signature writing style is everywhere now, and I hate it. It reminds me of when we tried mixing all the beverages at the soda fountain in middle school. We didn’t actually create the perfect drink, we just made a cloying monstrosity that lost everything good about its constituent parts.

Make a joke map.

#Fake #30DayMapChallenge #RStats

YouTube video by Posit PBC Practical {renv} (Shannon Pileggi, The PCCTC) | posit::conf(2025)

The recording of my #rstats posit::conf talk ✨ practical {renv}✨ is available! 🎉

I created this talk for frustrated {renv} users 🫠 and potential {renv} users🤩

youtu.be/l01u7Ue9pIQ?...

Even if {renv} isn't on your radar, you may still find this talk interesting if you are curious about🤔...

I started watching this with a pit in my stomach due to how hard it is to utter a single statement about anything in this space that isn’t loaded with caveats.

But that was gone by the end! Love the ecosystem map idea. Nice one.

But if you do try to do more, be careful you don’t make the package more annoying to test / work with. That’s not in your user’s interest either!

Increment a major version number (per semver), and mention the breaking change in your NEWS.

If you have reverse dependencies you’ll have to let them know and give them some time to update.

Anything beyond that is a kindness.

(spending 9 billion dollars to make a robot say "i have always loved you") hard times create strong men

It’s a bad look. It wipes anything user had sitting in the env from the rprofile for one. Suggests total dependence on interactively running scripts manually, which in turn suggests lack of training in reproducible data sci / eng practices.

that people use AI technologies in their work and explain that it saves them precious time that they can devote to other more urgent tasks tells me less about the wondrousness of the technologies and more about the overextension and immiseration of workers.

Very good!

But if you were moving into a new home and had to assemble a bunch of flatpack furniture back to back you’d be wishing for the screwdriver after the first or second one.

Now imagine you have to assemble flatpacks 5 days a week for the rest of your career.

It’s electric drill time!

I think that’s pretty good. One thing I would add is that the value of specialisation becomes more obvious at scale.

Like if you have a choice of one tool for your home you choose swiss army knife as it will do a passable job at all the odd jobs...

Cc @thewaroncars.bsky.social Randall enters the chat

I don’t even know if there’s really a contest for greatest DS lang…

“The second best language for any task” - this is how Python market Python. They know it’s mashed potatoes.

Python is not a great language for data science. Part 1: The experience It may be a good language for data science, but it’s not a great one.

You know it's gonna be interesting when it starts with, "Yes, I’m ready to touch the hot stove." 😂

I've seen folks missing the base R in the footnote 👀

What say you, #databs #rstats #pydata? /Aragorn voice/

Is there civilized discussion to be had? 🤔

blog.genesmindsmachines.com/p/python-is-...

And look at this function. A deprecated argument. A vestige of refactorings past. Unmistakably human authored.

Also inhale deeply through your nose. It’s got a faint odour to it no?

There are similarities to the untrained eye. But trust me your users WILL know the difference.

Sir we are purveyors of quality hand-coded software. Typed out in the traditional way by fingers on mechanical keys. Some backlit by LEDs.

Look here, look at this code. It looks clean, just like shoddy AI work, but let me draw your attention to this unexpected line break here…

Make n high enough and you’ll get the famed ‘stack overflow’. I think its about 10000. That’s why you don’t see it in the wild.

However there is s technique to avoid the overflow, see Tailcall().

So what are we doing custom build, layouts, bindings? All of the above?

You will use these like twice a year and twice a year you will feel like a data wrangling jedi.

Going out on a limb here but the most exquisitely designed deeply domain specific functions in the #rstats #tidyverse are probably the tidyr::separate_ family.

separate_wider_regex() is my main. The patterns arg! Beautiful. The debug mode! So sympathetic to the user’s whole process.

Ice Creameries live and die on foot traffic so I see what you did there.

Child of former icecream family here. 🫡

Unmatched as a tool for identifying great people you should get to know.

George I need you to do to these companies what you could never do to your books and finish them.

No! It’s like it doesn’t exist.

It was actually jarring when a a vendor mentioned it today. Like suddenly remembering a bad dream.