Arturo

Fully human, often wrong, always learning ☸️ I sometimes make photographs and music. Software engineer.

[Arturo @arturovm]

Hi friends 👋🏽 I just updated shades-of-purple-emacs, and it now includes rainbow-delimiters support, better line numbers and gutter colors, and the right current line highlight background color. 🎉 (Plus some random idris-mode faces).

https://t.co/vCpDcAXww5

#emacs

[Arturo @arturovm]

@brion369john @arithmoquine This poem gives me Neuromancer vibes. Very nice. Did you write it?

[andi (twocents.com) @Nexuist]

i get the sense the people most bullish on vibe coding are people who have spent their careers talking to SWEs over 15-30min calls and haven’t quite grasped how many hundreds of daily decisions are hidden from them out of respect for time

[Arturo @arturovm]

"Vibe coders" are just "idea guys" with access to ML tools, right? Just checking I understand the meme.

[Santiago @svpino]

Quick reminder: I'm charging $1,000/hour to fix your vibe-coded mess.

[Andrew Ng @AndrewYNg]

The buzz over DeepSeek this week crystallized, for many people, a few important trends that have been happening in plain sight: (i) China is catching up to the U.S. in generative AI, with implications for the AI supply chain. (ii) Open weight models are commoditizing the foundation-model layer, which creates opportunities for application builders. (iii) Scaling up isn’t the only path to AI progress. Despite the massive focus on and hype around processing power, algorithmic innovations are rapidly pushing down training costs.

About a week ago, DeepSeek, a company based in China, released DeepSeek-R1, a remarkable model whose performance on benchmarks is comparable to OpenAI’s o1. Further, it was released as an open weight model with a permissive MIT license. At Davos last week, I got a lot of questions about it from non-technical business leaders. And on Monday, the stock market saw a “DeepSeek selloff”: The share prices of Nvidia and a number of other U.S. tech companies plunged. (As of the time of writing, some have recovered somewhat.)

Here’s what I think DeepSeek has caused many people to realize:

China is catching up to the U.S. in generative AI. When ChatGPT was launched in November 2022, the U.S. was significantly ahead of China in generative AI. Impressions change slowly, and so even recently I heard friends in both the U.S. and China say they thought China was behind. But in reality, this gap has rapidly eroded over the past two years. With models from China such as Qwen (which my teams have used for months), Kimi, InternVL, and DeepSeek, China had clearly been closing the gap, and in areas such as video generation there were already moments where China seemed to be in the lead.

I’m thrilled that DeepSeek-R1 was released as an open weight model, with a technical report that shares many details. In contrast, a number of U.S. companies have pushed for regulation to stifle open source by hyping up hypothetical AI dangers such as human extinction. It is now clear that open source/open weight models are a key part of the AI supply chain: Many companies will use them. If the U.S. continues to stymie open source, China will come to dominate this part of the supply chain and many businesses will end up using models that reflect China’s values much more than America’s.

Open weight models are commoditizing the foundation-model layer. As I wrote previously, LLM token prices have been falling rapidly, and open weights have contributed to this trend and given developers more choice. OpenAI’s o1 costs $60 per million output tokens; DeepSeek R1 costs $2.19. This nearly 30x difference brought the trend of falling prices to the attention of many people.

The business of training foundation models and selling API access is tough. Many companies in this area are still looking for a path to recouping the massive cost of model training. Sequoia’s article “AI’s $600B Question” lays out the challenge well (but, to be clear, I think the foundation model companies are doing great work, and I hope they succeed). In contrast, building applications on top of foundation models presents many great business opportunities. Now that others have spent billions training such models, you can access these models for mere dollars to build customer service chatbots, email summarizers, AI doctors, legal document assistants, and much more.

Scaling up isn’t the only path to AI progress. There’s been a lot of hype around scaling up models as a way to drive progress. To be fair, I was an early proponent of scaling up models. A number of companies raised billions of dollars by generating buzz around the narrative that, with more capital, they could (i) scale up and (ii) predictably drive improvements. Consequently, there has been a huge focus on scaling up, as opposed to a more nuanced view that gives due attention to the many different ways we can make progress. Driven in part by the U.S. AI chip embargo, the DeepSeek team had to innovate on many optimizations to run on less-capable H800 GPUs rather than H100s, leading ultimately to a model trained (omitting research costs) for under $6M of compute.

It remains to be seen if this will actually reduce demand for compute. Sometimes making each unit of a good cheaper can result in more dollars in total going to buy that good. I think the demand for intelligence and compute has practically no ceiling over the long term, so I remain bullish that humanity will use more intelligence even as it gets cheaper.

I saw many different interpretations of DeepSeek’s progress here in X, as if it was a Rorschach test that allowed many people to project their own meaning onto it. I think DeepSeek-R1 has geopolitical implications that are yet to be worked out. And it’s also great for AI application builders. My team has already been brainstorming ideas that are newly possible only because we have easy access to an open advanced reasoning model. This continues to be a great time to build!

[Original text: https://t.co/yiOHeGJgLZ ]

[Marcel van Oost @oost_marcel]

𝗡𝗲𝘄𝘀 🚨: @nubank (eSIM) mobile phone plans have officially been launched. Previously available through a waitlist since its launch in October, the service called 𝗡𝘂𝗖𝗲𝗹 is now fully operational.

Gradually, the NuCel tab will appear in the Nu app, and customers in Brazil will receive notifications about the new offer, allowing them to purchase the product directly and transfer their numbers seamlessly—all in a fully digital manner.

Initially, the service will be provided to customers who have already expressed interest in acquiring NuCel. Gradually, it will be extended to other Nu customers in Brazil.

At this stage, NuCel will operate only on devices with eSIM, supporting Android and iOS operating systems.

“The interest from our customers has been overwhelmingly positive. We are taking another step in our mission to transform the mobile telecommunications sector by placing the customer at the center of decisions, eliminating complexities, ensuring transparency, and offering the best service and support,” says Paulo Menescal Barbosa, General Manager of NuCel.

Source/more info: https://t.co/iCfvwJHZgt

[Andrew D. Huberman, Ph.D. @hubermanlab]

If you wake up after 4-5 hours of sleep & find it hard to go back to sleep, it’s likely you offset your primary sleep drive (due to adenosine buildup etc). The next 2-3hrs of would-be sleep is when learning associated brain changes occur. 3 things help in this scenario. (Thread)

[The Cultural Tutor @culturaltutor]

Urban design isn't magic — there are specific reasons why we like some places more than others.

So here are 10 ways to make a street more (or less) interesting...

[Arturo @arturovm]

So many views, so much becoming.

[Arturo @arturovm]

@andremooij And I hope it's a peaceful hill with tall grass swaying in the wind and a few birds calling.

[Jonathan Blow @Jonathan_Blow]

I keep hearing on podcasts all these people raving about how AI is revolutionizing programming and I just don't see it. At all. (I think this may happen eventually, but what is being done right now is not really on the path to it).

Originally I thought, okay, these are just people who don't understand programming saying this stuff. It'll fizzle out once people try it and realize it doesn't work that well. But that was almost 2 years ago and it is still going, and the claims keep getting bigger.

[Arturo @arturovm]

Any wars between programming language fanatics are, ultimately, about whether one thinks that the shortcomings of a program written in each language are the fault of the language designers or of the users.

I.e. whether one believes user error indicates a flawed design or not.

[Daniel Pink @DanielPink]

Think huge rewards always lead to better performance?

Research shows that’s not always the case.

If-then rewards (“If you do this, then you get that”) are great for simple tasks w/ short time horizons. But they’re less great for creative tasks that require the long view.

Time to rethink motivation in the workplace. 🧠✨

[Arturo @arturovm]

@MandyLeao Tsss sáquense las rebajadas

[Math Cafe @Riazi_Cafe_en]

Competitive Programmer's Handbook

PDF: https://t.co/OAhdksFeQw

Intended for students who want to learn algorithms and possibly participate in the International Olympiad in Informatics (IOI) or in the International Collegiate Programming Contest (ICPC).

[Arturo @arturovm]

@rickasaurus Nice, what's your handle over there?

[Arturo @arturovm]

@cestlemieux I suspect this is some UX "trick" to account for the fact that if users have a complex password, it's easier to mistype, so it can lead to inconsistent results.

[Present & Correct @presentcorrect]

Important November Dates

1 Vegan Day
2 Book Lovers Day
3 Sandwich Day
3 Cliché Day
5 Bonfire Night
6 Nachos Day
11 Origami Day
13 Kindness Day
14 Pickle Day
16 Party With A Bear
16 Fast Food Day
19 Toilet Day
26 Cake Day
30 Stay Home Day