Ricardo Sánchez 🇪🇸🇮🇪

No se puede saberlo todo acerca de todo. Intentandolo sin descanso desde 1987

[Ricardo Sánchez 🇪🇸🇮🇪 @richnologies]

@nicolasfulghum REE pide cambios urgentes a la CNMC ante el riesgo de un nuevo apagón que encarecerán la factura - https://t.co/p7pKVwiO72

[Lex Fridman @lexfridman]

The assassination of Charlie Kirk is truly disturbing. I'm in shock. There should be no place for political violence in this country. This is not what America is about. We are better than this. Rest in Peace Charlie 🙏

[Ricardo Sánchez 🇪🇸🇮🇪 @richnologies]

Sí, correcto: lo que significa es que cada vez somos más iguales… igual de pobres.
La clase media no mide calidad de vida, solo un rango estadístico de ingresos.
Y si con 14.700€ al año en Madrid eres clase media, la conclusión es sencilla: en España la clase media es pobre.
Y si España es un país de clase media… entonces es un país de pobres.

[Ricardo Sánchez 🇪🇸🇮🇪 @richnologies]

Se os ha ido la mano con el ChatGPT: conviene revisar lo que escribe, porque a veces suelta expresiones dignas de Cervantes.

Lo de los 0,68 € “de nada” supone en realidad una subida de más del 6%, que es una barbaridad tratándose de una empresa en régimen de monopolio. Y compararlo con el 21% de otra empresa que compite en un mercado abierto es mezclar peras con manzanas.

[Ricardo Sánchez 🇪🇸🇮🇪 @richnologies]

El Gobierno anuncia refuerzos tras el apagón, pero sin asumir fallos. La planificación de red es competencia estatal. ¿Desde cuándo conocían las carencias? ¿Por qué no se actuó antes? Culpar a REE y generadoras es echar balones y no exime de responsabilidad política. Mucho ruido y ni una misera disculpa. Lo de siempre.

[Jared Isaacman @rookisaacman]

Caroline--everything you are attempting to misrepresent from my youth pales in comparison to the far more irresponsible decision I made later in life... getting into politics.

This isn’t investigative journalism nor is this new information. The details you are referencing were publicly disclosed in March and again in April as part of my Senate questionnaire (receipt link below).

I am a moderate and donated to both parties for different reasons...my largest contribution was to President Trump--because I support many of his policies. I definitely did not like the direction this country was going over the last 4 years. But what exactly is the fight about? The President made his decision. I respect him and I am moving on.

I have never spoken against the President. I have never voted against him. I didn’t enter politics to fight with you or Steve Bannon—or to enrich SpaceX or myself and I did divest all my aerospace equity even at Shift4. I got involved because I have lived the American dream--I owe this country and I wanted to serve the President by advancing American leadership in space.

I understand this is politics and that you may feel compelled to settle scores...but your latest posts are false and libelous... I am not your enemy but you should delete them.

I support the President and I will continue doing everything I can to advocate the competitiveness of our nation 🇺🇸

https://t.co/7JM5hOtgRr

[Ricardo Sánchez 🇪🇸🇮🇪 @richnologies]

@kikollan Trabajazo, como siempre. Estaría bien compararlos con los del artículo de 2022, porque da la sensación de que ha habido poco avance. La inflación, con suerte 🍀

[Ricardo Sánchez 🇪🇸🇮🇪 @richnologies]

@KonstantinKisin @elonmusk 💯

[Ricardo Sánchez 🇪🇸🇮🇪 @richnologies]

You said, “And if home prices crash, we take on the risk!” But that risk is already priced in — that’s why you’re lending $55K against hundreds of thousands in home equity.

You said, “We take none of the BTC upside at all, and it’s all yours right from the get-go to self-custody or do whatever you want with it.” Then why market this as a Bitcoin product?

Anyway, good luck with your product. Definitely not for me.

[Ricardo Sánchez 🇪🇸🇮🇪 @richnologies]

For a $500K house with $120K mortgage ($380K equity), they give you $55K in BTC. But if your home stays flat, you owe $120K in 10 yrs. At 4% growth, it’s $186K—77.5% of the $240K gain. Horizon earns 8–13% annually while you take all the BTC risk. They use past Bitcoin gains to sell a high-interest, backloaded loan as “smart investing.” It’s just expensive debt in disguise. They sell “no monthly payments” as a feature—but it’s a red flag: interest compounds silently, making this one of the most expensive loans you’ll ever take. All dressed in Bitcoin hype.

[raul cimas @picuetin]

El hacker

[Patrick Collison @patrickc]

A good read on this topic: https://t.co/TCf4perWaP.

[Stripe Developers @StripeDev]

Reliable Webhooks with Hookdeck https://t.co/pwdxCIs0s5

[Ricardo Sánchez 🇪🇸🇮🇪 @richnologies]

@marcvidal Si si, para ver que no hacer 🤦‍♂️

[Ricardo Sánchez 🇪🇸🇮🇪 @richnologies]

@Jongonzlz El problema de estos gráficos es que nunca incluyen el peso de la burocracia. Cuando del dinero se pierde en procedimientos vacíos y funcionarios poco productivos.

[Gautam Kedia @thegautam]

TL;DR: We built a transformer-based payments foundation model. It works.

For years, Stripe has been using machine learning models trained on discrete features (BIN, zip, payment method, etc.) to improve our products for users. And these feature-by-feature efforts have worked well: +15% conversion, -30% fraud.

But these models have limitations. We have to select (and therefore constrain) the features considered by the model. And each model requires task-specific training: for authorization, for fraud, for disputes, and so on.
Given the learning power of generalized transformer architectures, we wondered whether an LLM-style approach could work here. It wasn’t obvious that it would—payments is like language in some ways (structural patterns similar to syntax and semantics, temporally sequential) and extremely unlike language in others (fewer distinct ‘tokens’, contextual sparsity, fewer organizing principles akin to grammatical rules).

So we built a payments foundation model—a self-supervised network that learns dense, general-purpose vectors for every transaction, much like a language model embeds words. Trained on tens of billions of transactions, it distills each charge’s key signals into a single, versatile embedding.

You can think of the result as a vast distribution of payments in a high-dimensional vector space. The location of each embedding captures rich data, including how different elements relate to each other. Payments that share similarities naturally cluster together: transactions from the same card issuer are positioned closer together, those from the same bank even closer, and those sharing the same email address are nearly identical.

These rich embeddings make it significantly easier to spot nuanced, adversarial patterns of transactions; and to build more accurate classifiers based on both the features of an individual payment and its relationship to other payments in the sequence.

Take card-testing. Over the past couple of years traditional ML approaches (engineering new features, labeling emerging attack patterns, rapidly retraining our models) have reduced card testing for users on Stripe by 80%. But the most sophisticated card testers hide novel attack patterns in the volumes of the largest companies, so they’re hard to spot with these methods.

We built a classifier that ingests sequences of embeddings from the foundation model, and predicts if the traffic slice is under an attack. It leverages transformer architecture to detect subtle patterns across transaction sequences. And it does this all in real time so we can block attacks before they hit businesses.

This approach improved our detection rate for card-testing attacks on large users from 59% to 97% overnight.

This has an instant impact for our large users. But the real power of the foundation model is that these same embeddings can be applied across other tasks, like disputes or authorizations.

Perhaps even more fundamentally, it suggests that payments have semantic meaning. Just like words in a sentence, transactions possess complex sequential dependencies and latent feature interactions that simply can’t be captured by manual feature engineering.

Turns out attention was all payments needed!

[Ricardo Sánchez 🇪🇸🇮🇪 @richnologies]

@viOnemedia De Salamanca

[Ricardo Sánchez 🇪🇸🇮🇪 @richnologies]

@elmundoes Ahora abrir un simple formulario en Microsoft lo llaman “montar un sistema”.
Sin verificación de identidad, la ola de troleo que se viene va a ser histórica.

Las “conclusiones” saldrán, como siempre, de donde les dé la gana. Pérdida de tiempo monumental.