Havoc Pennington

@halfline @bill_nottingham @blandford @atnalind I don't ... but that doesn't mean a whole lot 😀

@bill_nottingham @havocp @blandford @atnalind any of you remember layoffs in 2004 ?

red hat didn't have large scale layoffs in 2004 did they? I don't remember that happening. I remember people were forced to move from raleigh to boston. I remember our financials got restated, our CFO resigned, and our stock price tanked. But i don't remember layoffs?

@devonzuegel appraisers find houses of comparable location and “quality level”, grab price per square foot of those, multiply subject property sf times the comparable price per sf. Oversimplified slightly but not a lot. Bank won’t approve a loan value over that number.

@devonzuegel one possibility is the market is affected by mortgage appraisal methodology which uses sf. Tax appraisal even more so, tax appraisers may not even look at the house just the recorded size and location. singlefamily.fanniemae.com/media/12371/di…

Why do home buyers look at price per sqft but not renters?

@sinamobasheri I think I would ask the people working on it today! I have seen a fair amount of porting C to Rust, in casual observation. But I don't know what people's overall feelings are.

Ben Shapiro is discovering what a murder mystery is in real time and acting like he understands what writing one means and I am truly and honestly crying. This is my Christmas gift. #GlassOnion

@mcmansionhell @amyhoy kinda reminds me of every year everyone picking the same baby names because they are so unique

@mcmansionhell @amyhoy On homebuilding facebook groups people usually say they are avoiding color because they fear it is “trendy” and they want “timeless” … meanwhile seemingly 95% are doing this “timeless” thing … at the same time 🤷‍♂️

And this is what we have been deprived of: zpr.io/EeP6qFyVHbnU

The history of peer reviews, and how it is destroying science:

The 2023 open source maintainer survey closes this Friday, Dec 9th! Have you shared your voice yet? There’s still time: bit.ly/3iG9mTL

On Mastodon, I now follow 25% the accounts I follow here. Found movetodon.org an interactive tool to follow folks you followed here, much nicer and simpler to use.

9 to 5 (Elon edition)

i cant believe the "click here to ruin your fucking life" button didnt have more takers

Many many people have pointed out of course that this means the AI models have our cultural frames, stereotypes, and cognitive biases. Except they do not have the ability to step out of those through reasoning or reality-check. Humans also vary in their ability to do that...

The traditional Data-from-Star-Trek sci-fi take on AI might be backward. We've managed to convincingly model some social functions of communication on a computer, before we've managed to model semantic/referential thinking.

Linguistic anthropology has a framework for describing what these models do, perhaps. They seem based on "non-referential indexicality" (see #In_linguistic_anthropology" target="_blank" rel="nofollow noopener">en.wikipedia.org/wiki/Indexical…). They take language as a closed system that's about co-occurring features (indexes). Performance of genre or role.