Josh Pello

my first pet's name was Bayes. doxx me. idc.

[Croatian Football  @CroatiaFooty]

81': Luka Modrić comes on, and the Spanish crowd gives him a standing ovation and chants:

"LUKA LUKA" 🇭🇷🐐

[Josh Pello @jgagathon]

Let’s start from minus-cognition.
Let’s see what awaits on the other side of mystery.

[Josh Pello @jgagathon]

Wondering why Blaga's transcendental censorship is so (onto)logical-annoying, ngl.

[Josh Pello @jgagathon]

Turns out, our minds are built to only find joy in the chase of what remains hidden~

[Josh Pello @jgagathon]

@talkingtrasshh

[Obsidian @obsdmd]

Obsidian 1.14.3 (early access) is now available!

- New Group menu in Bases lets you control columns in Kanban. Also works with tables, lists, cards.
- Added macOS Quick Look, so you can instantly preview Markdown files even when Obsidian is not open.
- New document icons for .md, .base, .canvas files.
- Quick Switcher is now five times faster.

And lots more!

[Obsidian @obsdmd]

Obsidian 1.14.3 (early access) is now available!

- New Group menu in Bases lets you control columns in Kanban. Also works with tables, lists, cards.
- Added macOS Quick Look, so you can instantly preview Markdown files even when Obsidian is not open.
- New document icons for .md, .base, .canvas files.
- Quick Switcher is now five times faster.

And lots more!

[Rob Haisfield @RobertHaisfield]

@andy_matuschak @instantricecook If you’re still online at all you’d love this @elzr https://t.co/Fs2qY5JN2M

[Andy Matuschak @andy_matuschak]

✍️ Playing with programmable highlighters, e.g.
* green highlight on a citation -> finds and prints the cited paper
* orange highlight on a claim -> prints a skeptical fact-check

Extends naturally to "if I write a comment like 'Is there more recent work?' and circle it in purple, send the comment with surrounding context to an agent"

[Pope Leo XIV @Pontifex]

A world in which machines, albeit sophisticated and faster than us, are the ones to decide to bomb defenceless people, would be truly frightening. The conscience of a single person can be worth the entire world.

[Pope Leo XIV @Pontifex]

God favors smallness, a sign of his discreet love, by which he leaves us free to accept or reject him. His love makes its presence felt even among the weeds, acts in a hidden and invisible way like the smallest of all seeds, and leavens the dough without making a sound. #GospelOfToday (Mt 13:24–43)

[Kat ⊷ the Poet Engineer @poetengineer__]

i made a reading interface for spinoza and its commentaries throughout centuries, inspire by talmud:
- scroll to adjust each era's thickness
- hover to discover cross-references between commentators

src code for subscribers ↓

[Josh Pello @jgagathon]

Nothing but evil

[kepano @kepano]

the question is what secrets do you want to give away, and what new secrets will you create?
https://t.co/sOZVoLi2fK

[kepano @kepano]

the question is what secrets do you want to give away, and what new secrets will you create?
https://t.co/sOZVoLi2fK

[Josh Pello @jgagathon]

🇻🇦🇻🇦🇻🇦

[sui @birdabo]

PewDiePie, one of the biggest YouTubers in the world just dropped an important video.

he claims the algorithm is destroying your brain.

a guy once beloved by the algo that made him a millionaire is now going against it.

“the key is intent. if you go around your life not making your own choices, then who the heck are you?”

his secret fix?
> add friction to dopamine apps
> unfollow everyone
> kill reels (important)
> self host everything.
> block at DNS level.

this is a wake up call. no more autopilot scrolling. you either choose what goes in your brain or the algorithm chooses for you.

[sui @birdabo]

PewDiePie, one of the biggest YouTubers in the world just dropped an important video.

he claims the algorithm is destroying your brain.

a guy once beloved by the algo that made him a millionaire is now going against it.

“the key is intent. if you go around your life not making your own choices, then who the heck are you?”

his secret fix?
> add friction to dopamine apps
> unfollow everyone
> kill reels (important)
> self host everything.
> block at DNS level.

this is a wake up call. no more autopilot scrolling. you either choose what goes in your brain or the algorithm chooses for you.

[𝗿𝗮𝗺𝗮𝗸𝗿𝘂𝘀𝗵𝗻𝗮— 𝗲/𝗮𝗰𝗰 @techwith_ram]

𝗚𝗮𝗺𝗲 𝗧𝗵𝗲𝗼𝗿𝘆 𝗯𝘆 𝗚𝗶𝗮𝗰𝗼𝗺𝗼 𝗕𝗼𝗻𝗮𝗻𝗻𝗼

Probably one of the best book on Game Theory. Access the PDF here: https://t.co/QW3Jlk29xj

𝗠𝗮𝗶𝗻 𝗧𝗼𝗽𝗶𝗰𝘀 𝗳𝗿𝗼𝗺 𝘁𝗵𝗲 𝗣𝗮𝗽𝗲𝗿:

- Introduction to Non-Cooperative Game Theory
- Strategic-Form Games with Ordinal Payoffs
- Dominance Relations (Strict and Weak) and Iterated Deletion Procedures
- Second-Price Auctions and Pivotal (Clarke) Mechanism
- Nash Equilibrium in Finite and Infinite Strategy Sets
- Dynamic Games with Perfect Information and Backward Induction
- Extensive-Form Games with Imperfect Information
Subgame-Perfect Equilibrium
- Games with Chance Moves
.......

[Francis Fukuyama @FukuyamaFrancis]

North Korea Was Right About Nuclear Weapons https://t.co/LKfeFgpEE4

[Rohan Paul @rohanpaul_ai]

DeepSeek's innovation level is really at another level.

Its new paper just uncovered a new U-shaped scaling law.

Shows that N-grams still matter. Instead of dropping them in favor of neural networks, they hybridize the 2. This clears up the dimensionality problem and removes a big source of inefficiency in modern LLMs.

Uncovers a U-shaped scaling law that optimizes the trade-off between neural computation (MoE) and static memory (Engram).

Right now, even “smart” LLMs waste a bunch of their early layers re-building common phrases and names from scratch, because they do not have a simple built-in “lookup table” feature.

Mixture-of-Experts already saves compute by only running a few expert blocks per token, but it still forces the model to spend compute to recall static stuff like named entities and formula-style text.

Engram is basically a giant memory table that gets queried using the last few tokens, so when the model sees a familiar short pattern it can fetch a stored vector quickly instead of rebuilding it through many layers.

They implement that query using hashed 2-gram and 3-gram patterns, which means the model always does the same small amount of lookup work per token even if the table is huge.

The big benefit is that if early layers stop burning time on “static reconstruction,” the rest of the network has more depth left for real reasoning, and that is why reasoning scores go up even though this sounds like “just memory.”

The long-context benefit is also solid, because offloading local phrase glue to memory frees attention to focus on far-away relationships, and Multi-Query Needle-in-a-Haystack goes from 84.2 to 97.0 in their matched comparison.

The system-level big deal is cost and scaling, because they show you can offload a 100B memory table to CPU memory and the throughput drop stays under 3%, so you can add a lot more “stored stuff” without needing to fit it all on GPU memory.

[Obsidian @obsdmd]

Obsidian Sync now has a headless client, so you can sync vaults to a server without using the desktop app.

Try the open beta: