@Dev_Maven1 @fwrenzo1 You've got it backwards. Health insurance companies are pushing routine checkups because they don't want to cover something much more expensive that was detected late.
English
Ryan DeRose
221 posts
@joshpuckett Determinism is vastly overrated and not the point.
English
It’s kinda hilarious how much energy we’re all spending trying to make a magical indeterminate system actually determinate.
English
@GWHayduke97 They would rather simply maintain that level of stress than cut back in any way on luxury lifestyle choices.
English
"$408k isn't a lot of money. I know this is true because we spend $100 a day on food"
Sam ⚔️ Financial Samurai@financialsamura
So you think earning $408,000 a year is a lot of money? After working 50+ hours a week, being stressed out of your mind, and barely seeing your kids, it might not be the dream income you imagined. Look how quickly the money goes. $1/month left over. What would you cut?
English
@dioscuri This kind of thinking very often stops me from doing temporarily convenient things that would compound negatively if repeated for weeks, months, years.
English
Universalisability usually comes up in ethics (“what if everyone did this?”), but it works pragmatically within a single life too. Before you text at the wheel or ride without a helmet, ask if you could rationally endorse such a policy for all similar moments. Almost always no!
English
@wormsofbarlo That's pretty much what I assumed when I first heard about this!
English
@MarcosHernanz About a year ago I was describing it to a coworker: "It's like the model is terrified of letting the script fail. To the point where if it doesn't find an input file it will just do nothing and declare success." I'd rather it to fail immediately so the stack trace tells me.
English
LLMs love to overengineer, and this hasn't gotten any better in the last few months.
Models are optimizing for correctness, not for code quality.
Panta@thepanta82
If you follow Sol's plans, you'll end up creating like 5x the amount of code you actually need. That mf loves to overengineer.
English
@mrdoob What, he’s not allowed to run his team how he wants? What if the team member doesn’t want to use Jai, Jon shouldn’t have a say in that either?
It might make his team uncompetitive, or it might not. What’s your problem with it? His personality/politics I guess.
English
@ArguablyAnna Thinking less of people is over-dramatic. I don't have a dog in this fight but hypothesize that it comes down to breaking of immersion (willing suspension of disbelief).
English
I literally have ZERO problem with loading screens
I want these games to be BIG and FULL and if that means I need to walk through a door and waiting a few seconds I literally couldn’t care less
If you give a shit about loading screens in these massive RPGs I think less of you
Synth Potato🥔@SynthPotato
The Elder Scrolls 6 and Fallout 5 are confirmed to be running on the Creation Engine This means they will very likely both still require loading screens when entering Interiors. I think this is one of the biggest weaknesses of Bethesda games, it really does feel outdated…
English
@benmooreau Instead of just buying a safer car you want me to lobby the government to make everyone sell the safe cars and buy small ones?
English
Never put your child in a Camry, Corolla or civic.
Why would you put somebody you love in a car that makes you 2x likely to die in an accident?
Big cars only. Full size SUV.
BowTiedBroke@BowTiedBroke
Need some car advice from parents. Third daughter just hit 16 & looking for a car for her. Got my 1st daughter a used Camry when she turned 16 years ago. Was a great car. Second daughter - used Kia Forte. Decent car. Now number 3. For parents of teenagers that have hit the driving years..what are some cars that you’ve been thrilled with that were reasonable in price but also had good safety features? Leaning to a Camry or Corolla but wanted to hear some other experiences with vehicles and teen drivers.
English
@RyanJDeRose @qo01q @fentanylbrownie They're not upper fringe. They're a variety of Low Functioning Normie.
English
People are having a field day with the “no anime” thing, but I do have to come down on the side of making an effort to ensure your kids aren’t weird. The state of things is dire. They’re wearing cat ears and tails to school. It’s not how it used to be.
English
@alz_zyd_ If we're gonna keep doing remakes, it may as well be the classics.
English
we should remake the classics, age after age, each in each age's style with all its flaws, and with the most advanced technology available
alli@sonofalli
just heard a teenager say “the odyssey is based on a book?” open the schools
English
@qo01q @trobuling @fentanylbrownie Yeah, I was far from normal but at least had the good sense to avoid the theater kids.
English
@trobuling @RyanJDeRose @fentanylbrownie band kids are much more normal than theater kids. they're just dorky.
English
@justalexoki Philosophy majors are protected by god and are allowed to live a life of whimsy,
leisure and learning because in the end everything works out for them
Even the Husserl PHD guy it worked out for him in the end
English
every philosophy major just has exactly 129 IQ. they're all locked into the too stupid to do anything worthwhile but just smart enough to be life ruiningly anxious about it box
Hitchslap@Hitchslap1
The average philosophy major has an IQ of around 129. They’re among the smartest graduates on the planet. Few know this.
English
@edzitron @AllVentured Anthropic and Kimi have already seen demand outstrip supply.
English
@AllVentured They’re also making dramatic assumptions about how much actual demand there is!
English
The Jevon’s paradox cope on here is hilarious. There is $10T of paper asset value teetering on the OpenAI ponzi.
Nobody has a moat. Chinese models will be built on Chinese hardware at a fraction of the cost. AI becomes a public good. Nobody ever turns a profit.
English
@TheCryptoCPA If landlords didn't make at least some profit, they wouldn't exist. Therefore, on average a landlord charges more on rent than they pay in property upgrades.
English
@justanotherlaw @ben_sturgeon If most of the models today are over-fitted to agentic coding tasks (a reasonable hypothesis), then fewer parameters would presumably be available to store those obscure facts. Could this put a downward bias on the results? You could test on open models from a year ago.
English
A recent viral paper claims to reverse-engineer the parameter counts of frontier models: GPT-5.5 = 9.7T, Opus 4.7 = 4.0T, o1 = 3.5T, etc.
@ben_sturgeon and I investigated and found serious issues in the paper; fixing them gives GPT-5.5 as ~1.5T (90% CI: 256B-8.3T).
English
@slimer48484 @Michael_Druggan @justanotherlaw @ben_sturgeon But open weight models would presumably follow the same underlying physics, and on logarithmic scales the gap to frontier is not so far-fetched.
English
@Michael_Druggan @justanotherlaw @ben_sturgeon They did leave one out validation for every open weights model in their list. But frontier models are generally bigger than any open weights model, so the extrapolation potentially breaks down
English
@BarneyFlames @SilverVVulpes The internet created a vast array of subcultures on sub-generational timescales.
English
@SilverVVulpes From a hereditarian perspective it seems unlikely that the parents are extremely simple people while the child is complex af. So it obviously seems like flimsy whining
English
this comic is always funny because I've never seen anyone react to it with anything beyond denying the premise
I'm not sure what drives, beyond the free association of the comic with a 'whiner' and a'whiner' complaint must not be allowed to point at real'unfairness' in the world
adam ✧ ❥ ~@adam__isom
in this comic, the person on the left is just terminally online, that's all I don't think organic individuality makes people incompatible, just the morass of unlived and bizarre ideas they've memetically ingested
English
@jasondoesstuff Yes to all, except I expect a doubling of em dashes year-on-year as frontier AI labs scramble to saturate "agentic-em-dashes-bench"
English
1 year from now:
- No one uses skills
- No one needs to install apps via Terminal, etc
- API/MCP/etc are just an "allow" button away
- It remembers "no em dashes" ever
- Compaction is unnoticeable
- Model token size doesn't matter
You tell AI what to do, it just knows.
English