djma retweetledi
djma
3.7K posts

djma retweetledi

Today I’m stepping into the CEO role at Messari. After conversations with Eric and the board, we agreed this is the right step for the company’s next chapter.
This transition also includes a difficult decision: we’ve parted ways with many teammates who helped build Messari into what it is today. I’m incredibly grateful for their work and the impact they’ve had on the company. They’re an exceptionally talented group, and I’m eager to help connect them with teams that are hiring.
Looking ahead, we’re doubling down on Messari as an AI-first company serving institutions through research and AI products.
The industry and the world are changing quickly, but our mission remains the same: helping customers navigate crypto with confidence.
English

Nice middle class Asian family but that ain't winning.
Winning is the patriarch running borderline illegal business in a primary industry in Indonesia, Philippines or Cambodia or misappropriating state assets from a China SOE before moving to Singapore under the Global Investor Programme and hard pivoting into real estate bc it’s the one thing your overseas enemies can never seize.
If patriarch is Singaporean to start with, wealth is from a questionable “trading” / “consulting” business involving members or alleged members of the Malaysian or Thai royal family. A guy who knows people in East Timor might be involved somewhere. Money is invested into cash flow businesses or real estate because it’s the easiest way to launder the money.
Oldest son is groomed to take over the real estate empire / cash flow business but gets sent overseas to university for “exposure” and either turns out gay and decides to pursue art / music or becomes a lawyer / accountant that wants nothing to do with the skeletons in the family closet. Either way they are a disgrace.
Younger son is left to his own devices. Either ends up in Ivy League if naturally high IQ and then goes to New York / California or fucks it all up and is sent to study business management in Australia to save the family face.
The former shows up every 3 CNYs, the months before the patriarchs death and the will reading.
The latter discovers drugs and hookers and will be recalled by the patriarch to permanently occupy a seat as “director” in the family office and various portcos where he can do literally no damage to the family’s reputation or wealth.
The daughter(s) grow up spoiled. The patriarch likely optimized for a hot wife so depending on the dice roll, the daughters either end up hot as fuck like mummy or somewhat resembling a bloated toad like daddy.
The hot ones will at some point attempt an influencer career, possibly venturing into money-losing business ventures like skincare, spas or bars which daddy will write a blank cheque for. Depending on repression level, she might bring home an Indian man at some point just to give dad a heart attack. The 4D chess move here is for the patriarch to throw a minor fit before accepting her choice - they will break up within 6 months in that case. If the patriarch objects, she will marry him and the family will have to cope by constantly exclaiming how cute Chindian babies are.
The ugly ones will go overseas to study something completely worthless and promptly identify as a they/them, likely also finding their way into body positivity movements. Likely dead to the family until she comes crawling back in her 40s to look after the aging parents after realizing she spent 15 years hanging out with retards with nothing to show for it. Christianity might be discovered along the way.
The success in this family is that none of this shit fucking matters because daddy's gains have compounded so hard that he can sit in his GCB with four Ferraris, five mistresses and an entourage of hanger-ons knowing that none of the generation's fuckups matter (except maybe deep down, the Chindian babies if any) because he never had expectations of greatness to begin with.
He will die knowing he can never be overshadowed and that the family's legacy is his and his alone.
And that is what winning as a family looks like.
Wholesome@wholesome_X_
What winning as a family looks like…
English

@0xkyle__ @WazzCrypto First make 10m then do it 10 times. Works every time. Would you like me to start making 10m? Just say the word!
English

Hey Claude, port this gameboy code to switch. Make no mistakes.
<adds $100m in revenue>
Pokémon@Pokemon
════════════ Pokémon FireRed and Pokémon LeafGreen confirmed for Nintendo Switch! ═══════════ These download-exclusive titles will be available after the #PokemonDay Presents presentation which begins Friday, February 27, 2026, at 6AM PST. #PokemonFRLG
English

@raychi_god Who would subject themselves to such experiment when good sleep is clearly #1 on the don’t die protocol
English

@davidma How useful did you find LLMs in solving the challenge (apart from implementation work, obviously useful there).
English

What worked for me:
- LastLayer is the only one with shape 48,1
- There are 48 inp slots and 48 out slots. That's 48*48 ways to construct a Block (and 48! ways to construct 48 blocks), and 48! ways to order the blocks once they are constructed.
- Computed cosine similarity between X and residual for the 48*48 ways to construct a Block. The idea is that if X is basically unrelated to inp/out, then the cosine similarity will -> zero as the dimensions -> inf. If X and inp/outp have a relationship, then there's a chance that the cosine similarity is not 0 and the chance doesn't -> 0 as dim -> inf.
- I looked at the top 48 pairs with outlier cosine similarities and to my surprise they are all disjointed which gave me a high confidence I've found the right pairings. If they weren't totally disjointed, I would have used that to reduce the search space.
- For the block ordering, which I "solved" before the Block pairing, compute how large the residual is for every block (residual norm) applied to X. The idea is that earlier blocks will contribute more to modifying X to match y_pred. But in practice it seems the reverse is true. From that permutation, brute force pairwise swaps until there's no more improvement.
What partially worked:
- Define an affinity metric between inp and out: Single-Block MSE (compute the MSE between y_pred and the y after one block and final layer).
- Hungarian algo to pair up the inp and out.
What didn't work:
- Bunch of profiling: looking at feature distribution, features correlation matrix, weights std/norm, zeros, residual norm / X norm, visualizing the layers
- Trying to incorporate the idea that the gradient should be small when y_pred ~= y_true
- Greedy construction with various orderings.
- Didn't try Sinkhorn as suggested by AI.
djma@davidma
o shit, I solved Jane Street's droppedaneuralnet puzzle from the latest @dwarkesh_sp sha1 of solution: b58600d51f2524415ddf919ea4b356a7bf4dfaec
English















