P.S. You Should Know... | Issue #359

Out loud

my story 🚀

I dressed them alike so they’d be easier to find if I lost one of them at the airport

i’ve been thinking 💭

My best thinking. I was contemplating a minor business problem with a colleague. He sat quietly while a stream of words came out of my mouth. Eventually he commented that we were both thinking, but he was doing so in his head while I was doing so out loud. I think he meant that my stream of words wasn’t going to yield much because he was preoccupied. But often enough the noises coming from my mouth help my brain work better.

You may be familiar with the show House M.D. It’s a medical series about Dr. Gregory House, a diagnostic genius with a Sherlock Holmes-like approach to solving complex medical mysteries, often breaking norms and clashing with peers to diagnose patients correctly. [Credit: ChatGPT]

There’s a line from one of the episodes that has always stuck with me:

Dr. Gregory House : How come you guys have never tried to yank me off this case? I'm having hallucinations. Blackouts.

Dr. Eric Foreman : But you're always insane and you're always right.

Dr. Gregory House : I'm *almost* always *eventually* right.

My track record of being right is not comparable to House’s. Yet, I think the nature of my process could be. The point is that an iterative approach to problem solving (“eventually”) is the one that works well for me. One of the most powerful ways for me to iterate is by moving my lips and listening with my ears. When an idea is in my head, it often “sounds” better than it is. When it escapes my mouth and enters my ears, I evaluate it more critically. When you hear me say something and then immediately question or qualify the thing I said, that’s what is happening.

A few more comments on this topic:

1. When there is no other person present, I don’t do this. I’m not sure why! Upon reflection, maybe this is the reason rubberducking involves an actual rubber duck.

2. When the other person does participate in the dialogue, that is often even better! It turns up the heat, creating more opportunities for new ideas and sharper explanations.

3. This phenomenon can work for me in chatroom-style text interactions. If I’m feeling comfortable, I will tend to post one sentence at a time. That gives me time to course-correct with interjections like “wait, that’s not right.”

The last comment makes me wonder about how LLMs might be able to eventually support this iterative process for improving ideas and solving problems. Today’s common one-response-per-prompt interaction model doesn’t quite fit.

As always, I am hungry for your thoughts, tips and comments on the subject. You can reply directly to this email or leave a public comment below.

fun facts 🙌

I just love this video of Argentina’s new president. It’s a solid couple minutes of entertainment watching Javier Milei systematically cut down his government to libertarian-appropriate size. ~ learn more

The quietest places in the world’s loudest cities. “Urban noise pollution is not going away. That’s why we’ve put together this guide to the ‘noisiest’ cities across the US, UK, and mainland Europe, and the places within them where it’s possible to escape the hustle and bustle of city life.” ~ learn more

Someone stole $25 million with a deepfake multi-person video call. Warning: slow down whenever money is at stake! “A multinational company lost HK$200 million (US$25.6 million) in a scam after employees at its Hong Kong branch were fooled by deepfake technology, with one incident involving a digitally recreated version of its chief financial officer ordering money transfers in a video conference call, police said.” ~ learn more

tech, startups, internet ⚡

The only Vision Pro review I’ve read. From Ben Thompson of Stratechery. My takeaways are that it’s an absolutely incredible personal entertainment device, yet hostile to both guest users and productivity work that requires more than one app. ~ learn more

The coming dark patterns of Botomation in AI. Rob May predicts that bots will become the default collectors of information, which could suck for us because they never tire and have no shame. Will we be inundated with non-stop requests to benefit the AI models more than ourselves? ~ learn more

AI design patterns. This post offers some lightweight education. “As we’ve been researching the AI landscape & how to build applications, a few design patterns are emerging for AI products. These design patterns are simple mental models. They help us understand how builders are engineering AI applications today & which components may be important in the future.” ~ learn more

better doing 🎯

Deciding whether an investment is worthwhile. Investment is broadly defined in this article, not just financial. Point #1: don’t use Expected Value. The framework Jason Cohen provides in this post looks valuable to me. ~ learn more

Microsoft’s New Future of Work Report 2023. They use their proprietary data and published research to assess state of the art work practices. “A summary of recent research from Microsoft and around the world that can help us create a new and better future of work with AI.” ~ learn more

under the microscope 🔬

A promising new oral insulin pill. It works with micro-encapsulation of insulin with a coating that breaks down specifically in the liver and specifically when blood sugar levels are high. Amazing! “The medication that can be taken orally has already been tested on baboons, in which it was found to lower the blood sugar levels without causing hypoglycemia. The new insulin is ready to be tested on humans in 2025. If all goes well, diabetics are facing an easier life without injections.” ~ learn more

DNA as hardware rather than software. The team created a nanoscale electromotor, an innovative piece of equipment engineered from DNA. This motor features a turbine design constructed entirely out of DNA strands, arranged to form an axle and three blades, powered by the hydrodynamic flow within a nanopore. ~ learn more

big ideas 📚

Decarbonization 2024. Nat Bullard is a self-described expert on clean energy and climate finance. He puts together this annual presentation on “the state of decarbonization told with climate, capital markets, technology, and sector data.” It’s very detailed and full of interesting data. This 2024 version is 200 slides long. ~ learn more

How to decarbonize 85% of all industry using today's technology. I have not fact-checked this but it did make it through some sort of peer-review. “In a new study published in the journal Joule, the researchers went through a range of different industrial sectors looking at the available options for decarbonization, their emissions reduction potential, and their technology readiness level (TRL) – a measure of how close a given technology is to being ready for widespread mass adoption. … Here's a brief summary of the areas they assessed, the technologies that are ready to roll, and the places where there are still gaps.” ~ learn more

on the blockchain ⛓

Vitalik on the promise and challenges of crypto + AI. It’s not all sunshine and rainbows, that’s for sure. ”A particular challenge is: in cryptography, open source is the only way to make something truly secure, but in AI, a model (or even its training data) being open greatly increases its vulnerability to adversarial machine learning attacks. This post will go through a classification of different ways that crypto + AI could intersect, and the prospects and challenges of each category.” ~ learn more

thanks for reading!