cscheid.net
Principal Engineer at Posit (fka RStudio), building quarto.org
Learning Brazilian Choro and the 7-string guitar
1,605 posts
3,330 followers
460 following
Regular Contributor
Active Commenter
comment in response to
post
People ask me about it!
comment in response to
post
The number I gave wasn't for "white alone", which is also not the number reported in the post.
Just chill, why the fuck don't you.
comment in response to
post
Science! We do it because it's easy, and also because it makes us rich!
comment in response to
post
Have you read The Sharing Knife series? it's a romance/fantasy crossover series from Lois McMaster Bujold.
Romance is 100% not my jam, but I enjoyed the first book so much that I read the whole series. Really good.
comment in response to
post
"86% are white" How many people in the US are white? oh yeah, 86% you dumb fuck.
comment in response to
post
> And maybe, the question itself could mean that there is no clear visibility of what's being worked on.
A large project like ours has more coordination points than we can make public, and "adding visibility" here means _more_ people asking about things we can't provide them.
(ask me how I know)
comment in response to
post
So, honest request from someone who finds open-source work meaningful: can you help me change this "any updates?" culture?
Thank you.
I would appreciate it more than you can imagine. 9/9
comment in response to
post
But "any updates?" just makes everyone feel bad. When we try to minimize our exposure to it by shortly answering "when there are updates, we will post them here", we are told that this is an unkind response. (I'm not going to put the person on blast here, but I promise you it happens) 8/
comment in response to
post
Instead, you could (for example) add _new_ information to the post about how you want to use it, or other workarounds. This might give us ideas for different fixes. Like I said, we read every discussion post.
Or, you could upvote the request, so we can prioritize the fix when we have the time 7/
comment in response to
post
But we genuinely are working on other things.
So now imagine getting something like ~5 new "any updates?" question every day. There's nothing I can do from my side that will make the asker happy, and I promise you I will feel a little worse about it as well.
So it's a net negative for everyone 6/
comment in response to
post
But "any updates?" is not something we can act on.
All we can do is say, "sorry, no new updates". I can dress this up however I want, but I know that no matter how I answer it to you, it will be a disappointment. It's true that it's disappointing, and it sucks for you! I'm sorry, we know it. 5/
comment in response to
post
Now, when you see a bug or feature request that has had no updates for a while and you ask "any updates?", the way _we_ see it is a new notification on email (or whatever). In quarto-cli, I get about 20-40 of these a day. It takes about 90 minutes to get through all of them. And I do! 4/
comment in response to
post
That is a humbling, honoring feeling - you all trust us at least in part with some of your work, and I can't think of many more positive gestures than that.
So, when there's a feature or bug request that is out there for us to fix, I _promise you_ we know about it. We live that, 8 hours a day. 3/
comment in response to
post
The population that uses any sufficiently-successful software project scales much faster than the people who can meaningfully make fixes to it. (we're currently _three_ full-time engineers at quarto-cli, to give a concrete example)
I promise you we feel, very viscerally, how you outnumber us 2/
comment in response to
post
if big tech can control access to automation, we the lowly spreadsheet farmers are well and truly fucked
if stuff like claude code can be fully built in open source with models like ollama, then we at least stand a chance.
comment in response to
post
100%! I do wish that the discourse around automation would include, um I don't know, who controls the means of production^Wautomation.
To me, the fundamental q today is whether it's possible to have a moat around LLM-style automation.
yes -> dystopia
no -> much volatility, but hope for future.
comment in response to
post
The one that sends me every time is the “official law firm of Rutgers Athletics”
comment in response to
post
I’ve wondered about setting up honeypots so that you automatically catch people blatantly distorting their resumes in inconsistent ways and use that to filter bad-faith applications. But maybe that doesn’t even work, maybe it really is just too little friction!
comment in response to
post
I wonder about how to model this so that the math comes out to be that way. It feels hard!
(I’ve been on the hiring side of this recently, and it’s a complete mess. You really do get buried with what I can only guess are bullshit spruced up resumes with ai)
comment in response to
post
O Texas do Brasil mesmo 🤦♂️
comment in response to
post
It works, ha
comment in response to
post
bsky mangled that link, just as a heads up
christophertkenny.com/posts/2025-0...
comment in response to
post
Well thank you I just found my next book to read
comment in response to
post
I _am_ driving!
comment in response to
post
I've reread neuromancer recently and to me it was more like "wow, this dude was _hurting_ at the time". But I agree that if you look at the Neuromancer -> Pattern Recognition -> Peripheral arc, it's pretty clear what's up.
comment in response to
post
Also I guess people didn't read Peripheral?
comment in response to
post
I perennially have podcasts with five minutes of ads left unlistened. Dismissing all of them without dismissing the ones I still want to hear is an exercise is navigating their glitches
Yay for no incentives to fix your shit. Must be nice!
comment in response to
post
Spotify UI has had rendering bugs for at least a year - you hit “dismiss” on a podcast banner and it registers on the servers but doesn’t rerender the UI Then, if you dismiss the one down the list, it actually dismisses the *following* one, I assume because it uses UI indices rather than content ids
comment in response to
post
That’s a nice though
comment in response to
post
I think Penrose identifies as a pretty straightforward platonist in his books, so maybe even stronger!
comment in response to
post
Ok that’s more or less what I thought. Real estate play to keep the century 21 property around enough for the next thing to come through, call it a museum to dodge taxes in the meantime…?
comment in response to
post
All the ads near WTC feature AI generated imagery; the $35m price tag; the $60 admission price… this all screams “vc+grifty artist meet cute”
comment in response to
post
Check out the camera-ready version of our ICML position paper ("Position: Evaluating Generative AI Systems Is a Social Science Measurement Challenge") to learn more!!! arxiv.org/abs/2502.00561
(6/6)
comment in response to
post
It’s the kind of thing which I’d have a hard time believing anyone else, had it not happened to me.
comment in response to
post
It’s worse.
Developers don’t know about them either, so cli apps aren’t generally prepared to deal with them these days (nor are they built to be agnostic to their presence)
The default user experience sucks, which further pushes the incentives away from named pipes.
Sad stuff
comment in response to
post
Ok, that’s actually really good.