Silva, J.P. (@ertwro) header
Silva, J.P. (@ertwro) profile picture

Libertarian. posets and harmonic analysis. I can't change the past but can improve the present. Trust in my efforts. o(n) or GTFO, No triangles allowed.

Bogotá, D.C., Colombia
Joined October 2009
Photo by Silva, J.P. (@ertwro)Photo by Silva, J.P. (@ertwro)Photo by Silva, J.P. (@ertwro)Photo by Silva, J.P. (@ertwro)Photo by Silva, J.P. (@ertwro)Photo by Silva, J.P. (@ertwro)Photo by Silva, J.P. (@ertwro)Photo by Silva, J.P. (@ertwro)Photo by Silva, J.P. (@ertwro)
AI broke the Pax Romana of math academia. This is MMA meets traditional martial arts. Your karate-style incremental advances mean little when a bigger, meaner lab with more tokens in the latest model or a lucky amateur can jiujitsu your ass. Mathematicians know who works on their problems, how far they are, and discuss which to attack next at conferences. Now an amateur can one-shot the central problem of your PhD out of the blue. Before, merit lay in sharing partial results. Now someone can take your partial results, clean-room engineer them, and leave you citing the person who scooped you. Hiding or publishing your work is a double-edged sword. The shift is toward Tartaglia-style challenges. Not from nostalgia, but to keep professionals relevant. There, mathematicians’ training will demolish amateurs (prompt gamblers), and the scarce resource of attention will be won. Results will bypass the fiendish amount of LaTeX and Lean slop. Great mathematicians will keep doing great work. Collaboration will continue, but a new route has opened for young people. Why? If you are not from a big institution and solve a big problem you can show to the right people, most editors and reviewers will not dare touch you. They will resent the extra work they neither care about nor can judge. It is time to boldly go where no one has gone before. We will know more math than ever thanks to and despite it.
An internal version of Astra, @OpenAI’s next major model family, solved 10 major open problems in mathematics, quantum complexity, and theoretical computer science. We believe it will be a major step for scientific reasoning. openai.com/index/ten-advance…
1
1
1
604
When cellos are financed with the same incentives as houses.
If you think Ravel’s Boléro needs a full orchestra, listen to this version: just one cello, yet it’s every bit as captivating.
1
58
🤖 Made with AI
1
34
🤖 Made with AI
34
An idea: any paper where all creative mathematical contributions are due to an AI model should be positioned as "an exposition of a proof by Model X". That'd be intellectually honest. It would also add value, rather than DDoS-ing arXiv with daily conjectures. Example below.
I think this kind of rush to put one’s name on duplicative work largely generated by AI is pretty corrosive to the field. Important that we find a way to incentivize work with higher marginal value.
4
14
1
83
11,737
Me to my stumbling first-year-old when learning to walk. “Enough of partial results. Make a breakthrough.”
Dinitz-Garg-Goemans conjecture is false. This graph theory problem was open for ~30 years. The graph below has fractional flow cost 58. Any unsplittable flow (with capacity violation <=15) has cost at least 60. Chat with GPT 5.6 Pro where this was found: chatgpt.com/share/6a60b2eb-0…
1
132
Silva, J.P. retweeted
USD1 use cases, continued: → First @Morpho vault, live → First DeFi presence on @Monad → Curated by @SteakhouseFi → Built by @reservoir_xyz. That's a lot of firsts. 🦅☝️
Introducing the first USD1 vault on @Morpho Powered by Reservoir, @worldlibertyfi, @SteakhouseFi and @monad We are thrilled to announce the first Morpho vault for USD1 with collateralization facilitated by Reservoir. Curated by Steakhouse Live on Monad Starting today, users can bridge USD1 to Monad via @LayerZero_Core and deposit into the vault. app.morpho.org/monad/vault/0… Another use case now live for LibertyVoir 🦅
68
61
36
352
155,941
Terribly sad a man as good as Charlie Kirk was murdered. Hope if there’s a heaven he is there and my condolences to his family.
1
289
Can’t believe I’m paying $100/month just to scream and complain at Claude code about how it creates fake complexity, wastes my time, and builds catacombs of death code everywhere that misunderstand the basic functionality am describing.
272
Lo que hizo Javier Milei presidente de Argentina ha sido una absoluta cagada. Lo apoyó desde 2019. Aún si decenas de miles de ludopatas no hubieran perdido su dinero, el gobierno solo debe ser árbitro. No debe escoger ganadores o perdedores. Sin importar si le apoyan o no.
250
It’s so funny bitcoin maximalists angry at Trumpcoin. Your bitcoin only narrative died in 2013, not with ethereum, with dogecoin. Like in movies that don’t use cellphones because it destroys the whole plot, that was solving the bizantine generals problem and having a memecoin.
172
“Chief complaint” si bien pueden muchas necesidades o malestares, estos deben ser consignados en la revisión por sistemas, es deber del médico seleccionar el principal, expandirlo en la anamnesis, indagar si los otros están o no relacionados a este y resolver el principal. Como? nitter.gq/polvodel93/status/1786…
1
107
Puede verse después. Una uña del pie sospechosa para hongo. Ha empeorado? Estamos seguros que es un hongo? Primero resolvamos esa pregunta. Es la “Chief responsibility”. Algunos podrán pasar los 20 minutos sin permitir avanzar por su condición o verbosidad pero:
1
84
1) compasivo pero firme (el respeto es mutuo) 2) registrar lo que el paciente diga y clasificarlo: No hay 4 motivos de consulta, solo hay uno 3) consignar el estado del paciente (“Neuroma acústico, remisión a neurocirugía”. No es aceptable ¿Cómo está el paciente?) Eso es todo.
82
In my job I had no assigned tasks. Everyday was “build your own adventure” due to communication issues between HR (hired me) and my boss (wanted someone else). Theoretically I could do absolutely nothing and still get paid. I just quit today that kafkian nightmare.
1
107
Boletas compradas para mi esposa y yo con meses de anticipación para #kingofthering entre @Nikoarrieta y @la_liendra_ , para que al final lo tiren a uno donde caiga, el primero que llegue. Casi dos horas de fila. Nos vamos y nunca en la vida volveré a nada al @RoyalCenter
1
3
831