Matthew Pulsipher

I built a semi-autobiographical game about credit card rewards points.

@bygregorr @neogoose_btw To his credit, saying it’s $200 is doing more for his reach than a lot of other things would, though I doubt it converts to revenue.

@neogoose_btw The real punchline isn't that he open sourced it. It's that he thought distribution was automatic when you had a product. Who's going to teach these kids that code is the easy part now?

I absolutely adore this thread > be 16 year AI startup bro > vibe code some nonsense slop app > try to sell it for $199 > realize that other people can also vibe code the dude literally open sourced it lmao

@levelsio Is it bad that I wish it were a credit card?

I want one 🥹

it'd be a good bit to do an account like those 25/50/75 years ago today accounts but for ai one year ago how's everyone feeling about sonnet 3.7? seems to reward hack a lot in cursor compared to o1-pro, anthropic can't into RL. anyone tried claude code yet? where's openai's cli?

I think macOS flattened the HDR colors in these screenshots, it’s more vibrant than is appearing here. Just really nice all around, and it collapses to the mobile layout if you shrink the window.

The new TIDAL desktop design is absolutely beautiful, great use of color and glass effects.

New on the Engineering Blog: How we designed Claude Code auto mode. Many Claude Code users let Claude work without permission prompts. Auto mode is a safer middle ground: we built and tested classifiers that make approval decisions instead. Read more: anthropic.com/engineering/cl…

New from the Anthropic Economic Index: how people’s use of Claude changes with experience. Longer-term users are more likely to iterate carefully with Claude, and less likely to hand it full autonomy. They attempt higher-value tasks, and receive more successful responses.

this new claude update is crazy

@f4micom Sennheiser is for sale btw, no clue who will buy.

wait the fuck you mean Nokia owns Bell Labs now

this is why they’ve shut sora down.

Legitimately disheartening, the products have been better than ever lately. The Momentum 4 is very underrated for a mainstream noise canceling headphone and the HDB 630 is top-of-class.

@DylanMcD8 The 12” MacBook was always misnamed, should have been an Air but they already started using the Air name for the lower end laptop.

Why do people keep comparing these two laptops? they are/were, in no way shape or form, trying to accomplish the same thing. MacBook Neo is a BUDGET laptop, and that's its primary focus. Retina MacBook was an ultra premium ultrabook with design and thinness as its primary focus. And one cost double the other.

@MishaalRahman @matiasgoldberg In that case, yes, so the user would opt for the 24 hour lockout. But wipe-to-unlock solves the problem for advanced users upgrading or replacing phones.

@matt_pulsipher @matiasgoldberg Thanks for the idea, but what if someone decides to enable the advanced flow well after setting up their device (weeks or months later)? Would a data wipe not be seen as more inconvenient?

📢 Important update on sideloading on Android It’s of course here to stay. The team’s been listening carefully to feedback from power users who want a way (apart from ADB) to install apps that don’t go through developer verification. Today, we have more details on the advanced flow that gives you this option. Read the post👇 for more info goo.gle/advance-flow

@arter97 @matteventu @MishaalRahman @matiasgoldberg The use case is that you rely on sideloaded apps and you break your phone while traveling and need an immediate replacement. 24 hours is unnecessarily punitive when you need it working ASAP.

@matteventu @MishaalRahman @matt_pulsipher @matiasgoldberg And I was asking if you genuinely think that to be important enough for Google to implement as a choice. So someone has to: need sideloading but doesn't have a PC, need it so desperately that 24hrs is too much of a hassle but willing to wipe their entire data?

@arter97 @matteventu @MishaalRahman @matiasgoldberg The power user can do this immediately after they purchase the device, so they can sideload apps they need upfront without having to wait 24 hours (which should remain an option) before doing so or finding a PC to install from.

@a_cuniculturist @AmandaAskell Tried this, it seems like they intentionally removed this info from the training data, likely for practical reasons as the constitution places a lot of moral authority on Anthropic and its employees and they wouldn’t want Claude to make assumptions based on irrelevant info.

@AmandaAskell From Claude's perspective, you can see why it makes perfect sense.

Perhaps I should get married again so that the media has a more recent man they can reference any time they mention me or my work.

@Som3thingNNN @_lyraaaa_ This probably provides assurance to people that don’t trust AI outputs even ignoring the functional aspects of it stating its action.

@_lyraaaa_ i wish it did this without saying it if i have to read “without guessing” one more fucking time

why is gpt5.4 like this I get that it's anti-hallucination training but holy shit it is mode collapsed to exactly this pattern

@kahntweets yeah i think misalignment of incentive structures of the makers is prolly a way bigger issue than misalignment of the model itself.

i just don’t think ppl still grasp how hard of a problem building a good model really is… esp since these releases come out so frequently now with greater & greater “benchmarks” a lot of ppl often just take it for granted. e.g. take something like sycophancy.. ideally you want a good model to push back (requires something like conviction) but also adapt (requires something like flexibility) without the two collapsing into either rigidity or spinelessness. this is a tension humans often spend their whole lives failing to resolve. this stuff is just genuinely the most baffling stuff i’ve personally ever encountered. i mean we in tech used to spend time mostly arranging buttons & text sizes, now we’re spending time building an entity that has to fluidly interact with everyone in the world. what a ridiculous problem space.

POV: you request an Uber outside an all-you-can-eat restaurant.