May 4, 2025

Daily Log Digest – Week 17, 2025

2025-04-26

Coffee Varieties beyond Arabica and Robusta

That’s why Davis is so interested in rediscovering lost coffee species. It’s not that any single species will provide a magical solution to the many challenges facing global coffee production—not even a mythical heat-and-drought-resistant coffee plant like stenophylla. But certain rare species could be used both to partly replace crops in areas that are becoming too hot for them and to crossbreed new, sturdier strains. To make this happen, though, scientists will need genetic resources from which to choose the best traits—and not only those that make them unusually resilient, but also those that impart a great taste, too. That last part is crucial. No species, however resilient, can become the coffee of the future if people don’t find it delicious.

Manufacturing Viability in US

Why It’s Impossible for Most Small Businesses to Manufacture in the US | WIRED #tarrifs #us #china #manufacturing

This is a great article with a lot of research and interviews on the ground with vendors and suppliers that makes the case that tariffs are unlikely to stimulate manufacturing in the US because there are so many other factors at play here.

Cost is undeniably an important reason why businesses choose to source from China. But experts say it’s incorrect to assume that lower prices mean lower quality, and the reason manufacturing in China is cheaper than other regions doesn’t always have to do with how much workers are paid. In fact, lower wages have become a less important aspect of China’s manufacturing strength as the country has moved up the value chain, says Eli Friedman, an associate professor studying China’s labor force at Cornell University.

“You definitely can’t say because wages in Chinese factories are only 25 percent of what American counterparts are working for, that the quality is going to be 25 percent of the American product,” Friedman says. “That’s much too simplistic a way to think about this.”

Cultural norms like working long hours and intentionally spending decades in the same industry often means that workers in China have become more skilled and specialized in certain areas. China is also a world leader in the production of industrial tools, which means factories can easily adjust machinery to fit the ever-changing needs of their customers. As a result, Chinese factories are often more responsive to customization demands from clients and more capable of precisely orchestrating their design intentions.

Kim Vaccarella, the founder and CEO of a handbag company called Bogg, makes products out of EVA, a rubber-like petroleum byproduct also used for flipflops and yoga mats. Vaccarella says it’s possible to make EVA products in Vietnam, but when she researched sourcing from there, she found that a lot of the factories were Chinese-owned and employed Chinese engineers. “China has mastered EVA. They’ve been doing shoes in EVA for 20-plus years, so it was really our first choice,” Vaccarella says.

If Bogg tried to move its manufacturing to the US, Vaccarella says she believes she would also need to hire Chinese talent to help ensure the production lines were set up correctly. But she worries that would be difficult, especially given the Trump administration’s current policies to reduce immigration. “With everything going on with our borders, is it going to be hard to get the visas for the Chinese counterparts to come in and be able to help us build this business?” she asks.

I am Martin Parr

Watched this movie finally and totally loved it!

One notable thing about the movie was that it did not discuss camera gear at any point in the film. There is a discussion of Parr moving from black-and-white film to color film. You can also see him holding a Canon DSLR in many of the frames. But at no point in the film did the film talk about the cameras that Parr uses. I thought that was very refreshing.

2025-04-27

The End of Deep Reading

Ultra-Processed Minds: The End of Deep Reading and What It Costs Us #reading #distractions

I recall that experience today somewhat wistfully, in an unrecognisable age where so little of what we read asks anything of us. The idea of lingering with a text that doesn’t yield immediate meaning feels increasingly alien today. In a world shaped by feeds and fragments, where comprehension is measured in clicks and content must gratify instantly or be discarded, a book like Faulkner’s feels almost impossible.

I read to my 6 year old and 4 year old daughters every night and I worry what kind of reading life they will have. I know it will be fundamentally different to mine but what I fear they’ll miss is not literature’s beauty but its resistance. The way it trains the mind to slow down, to reflect, to tolerate ambiguity. To reside in the discomfort of the interstitial and be ok with it. The way it sharpens perception by refusing to simplify. The world they are entering is freighted with superficiality and false certainty with its Blinkist-ification and summaries of summaries and algorithmic banalities. As Baudrillard put it, copies of copies of which we have now lost the original.

2025-05-01

Adolescence TV show

The problem with Adolescence | Dazed #tv #misogyny #gender #manosphere

Finally ended up watching this show. Now catching with all the internet think pieces on it!

It’s true that those of us on social media – that’s 98 per cent of us Gen Zers – are entirely at the mercy of algorithms. Major platforms like TikTok remain maddeningly opaque about exactly how their algorithms work, but it’s not difficult to spot patterns in what they choose to recommend. Speaking to the BBC in 2024, Andrew Kaung, a former analyst at TikTok, algorithms are designed to fuel engagement by showing you content which you’re inclined to spend longer watching. Often, this sort of thumb-stopping content is extremist in nature; independent research and reporting has consistently found that social media algorithms amplify misogynistic and inflammatory content.

But while it is impossible for parents to ensure their children are forever shielded from the likes of Andrew Tate, there’s no reason why this problem can’t be tackled at the root. If parents and teachers are powerless to adequately monitor the kind of content children are consuming online, why can’t social media companies do something? Why can’t misogynistic content be removed from platforms before it has a chance to poison impressionable young minds? It’s a question worth asking, but one Adolescence fails to pose. Instead the show meekly throws its hands up in defeat. “You can’t keep an eye on them all the time, love. We just can’t,” Eddie says.

Typing Practise

Studio: How To Type Fast #typing #practice

Found in this video: Learning to Type FAST in 5 Days - 150+ wpm Guide from MKBHD Team - YouTube

Typing Practice 2

If you’re from the video welcome! The following is the exact instructions I gave our team to follow for 1 work week which is why you’ll see instructions to record scores for each day. Enjoy!

For the next 5 work days dedicate 20 minutes a day to doing these practice exercises.

• 10 minutes before lunch, 10 minutes after lunch.
• You can break these 10 minutes up, but please make sure you’re doing at least 20 minutes a day.
• During all of these tests you should be trying your hardest to never look at your keyboard. Try to focus on not looking during regular work day typing as well.
• Focus on accuracy over speed at all times, even when you’re taking tests at the end of each day.
• Use the same keyboard for all of this practice.

NOTES:

• Try to do all of these activities at least once unless they seem too hard.
• I’d suggest starting with an easy activity first and then moving onto the harder ones.
• At the end of each day, go to Monkeytype and take 5x 15 second tests. Record your best score each day.

First

Keybr - This test needs to be finished before starting any other activities.

• Read all the instructions before you start.
• Create an account to keep track of progress.
• Unlock all letters as green before moving on
• If it takes you the full 5 days to complete this that’s fine

After Keybr is completed and all letters are green, here’s a list of different ways to practice during your time slots. I’ve grouped them in different tiers of difficulty.

Easy:

1. Go to Monkeytype and choose the “words” category on top.

2. You’ll have 50 words to type with no time limit

3. Focus on finishing each test with no errors

4. Before you start another test, select “Practice Words”, select “Words” for missed words and “On” for slow words and click “Start” and take the new focused test.

5. Open “Zen” mode in Monkeytype

6. Here you wont have words, but you just type what comes to mind.

7. Type in here what you need to accomplish for the day or just some things on your mind.
8. We practice a lot of typing while reading words on a screen, but in the real world you’ll be typing something in your head.

9. Press “Shift - Enter” to finish Zen mode.

10. Go to Monkeytype and select “Custom” then “Change”

11. In the custom box put your full name.

12. Our names are something we type all the time

13. Set it to 30 seconds and practice typing your name.

14. Continue Practice Mode in Keybr

Medium:

15. Go to Monkeytype and select “Quote”

16. This will add capitalization and punctuation to your tests

17. Play TypeRacer

18. Racing game where you race against other people with similar typing speeds

19. This also includes punctuation

20. Play Z Type

21. Asteroids based typing game.

22. No punctuation or capitalization

23. Go to Monkeytype and select times over 1-2 minutes

24. Allow yourself to type for longer amounts of time vs a short test.

25. Focus on accuracy

Hard:

26. Go to Monkeytype and select either “time” or “words”

27. Go to Settings and “Funbox”

28. Select “Read Ahead Easy”
29. This will remove the next word as you type forcing you to read ahead while typing.

30. If this is too easy, move to “Read Ahead” or “Read Ahead Hard”

31. Go to TypeLit.io

32. This website lets you choose a book and type the chapters.

33. Lots of punctuation and general formatting you must follow

34. Type a page at a time

35. Go to Monkeytype, in funbox settings click “Wikipedia”

36. This will give you prompts that are based on Wikipedia articles

37. Lots of punctuation again, but in the familiar Monkeytype layout

• Try to do all of these activities at least once unless they seem too hard.
• I’d suggest starting with an easy activity first and then moving onto the harder ones.
• At the end of each day, go to Monkeytype and take 5 15 second tests. Record your scores if you want to keep track!

Case for Living Online

Tyler Cowen: The Case for Living Online - by Tyler Cowen #online #culture

Why do I spend so much of my time with email, group chats, and also writing for larger audiences such as Free Press readers? I ask myself that earnestly, and I have arrived at a pretty good answer. I believe that by spending time online I will meet and befriend a collection of individuals around the world, who are pretty much exactly the people I want to be in touch with. And then I will be in touch with them regularly.

I call them “the perfect people for me.”

I recognize that many of these communications are online, and thus they are “thinner” than many more local, face-to-face relationships. Yet I do end up meeting most of these people, and with great pleasure. That, in turn, enhances the quality of the online communications. And frankly, if forced to choose, I would rather have thinner relationships with “the perfect people for me” than regular bear hugs and beer guzzlings with “people who are in the 87th percentile for me.”

The internet, in other words, has invented a new means of human connection, characterized by “the perfect people for me.” For me, it’s people who are into analytical thinking and tech and AI and music and economics, and much more. For others? It can be Survivor obsessives or vegans or knitters or Survivor obsessives who are vegan and love to knit. The point is that there is a niche for all 8 billion of us. And now we know where to find each other.

And it turns out we value that very, very highly. So highly that we are willing to obsess over our little devices known as smartphones.

The renaissance of personal software

The 70% problem: Hard truths about AI-assisted coding #ai #programming

I believe we're going to see a renaissance of personal software development. As the market gets flooded with AI-generated MVPs, the products that will stand out are those built by developers who:

• Take pride in their craft

• Care about the little details

• Focus on the full user experience

• Build for the edge cases

• Create truly self-serve experiences

The irony? AI tools might actually enable this renaissance. By handling the routine coding tasks, they free up developers to focus on what matters most - creating software that truly serves and delights users.

How to live like an Epicurean — 9 key habits

How to live like an Epicurean — 9 key habits #epicurean #philosophy

If you’re looking for a way to live a more fulfilling or simpler life, living like an Epicurean might be the answer. Epicureanism is a philosophy that emphasises the importance of pursuing pleasure and avoiding pain to live a fulfilling life. However, this idea of pleasure was not limited to physical pleasure alone but also included intellectual pleasures such as knowledge and wisdom. It teaches that the greatest good is pleasure and the absence of pain.

1. Focus on Inner Happiness: Seek joy from within rather than external possessions.
2. Practice Self-Control: Master your emotions to achieve tranquility.
3. Prioritize Meaningful Relationships: Cultivate genuine friendships for support and happiness.
4. Embrace the Present Moment: Live fully in the now and enjoy simple pleasures.
5. Seek Healthy Pleasures: Find joy in virtuous living rather than excess.
6. Cultivate a Mindfulness Practice: Engage in meditation or reflection for greater self-awareness.
7. Accept What You Cannot Control: Let go of the need to control everything around you.
8. Challenge Adversity: View challenges as opportunities for growth and resilience.
9. Live Moderately: Avoid excess and focus on balanced choices that promote well-being.

Avoiding Skill Atrophy in the Age of AI

Avoiding Skill Atrophy in the Age of AI - by Addy Osmani #ai #software #programming

Here are the key points from the section "Using AI as a collaborator, not a crutch":

• Mindful Engagement - Treat AI as a collaborator rather than an infallible tool. Engage actively with its outputs.
• Practice AI Hygiene - Always verify AI suggestions. Test and understand AI-generated code instead of accepting it blindly.
• No AI for Fundamentals - Reserve time for manual coding to strengthen foundational skills. Consider implementing "No-AI Days."
• Attempt Problems Independently - Try solving issues on your own before consulting AI. This enhances problem-solving skills.
• Augment Code Review with AI - Use AI to assist in code reviews but maintain human oversight to catch potential errors.
• Active Learning - Follow up on AI solutions to deepen understanding. Engage with AI by asking questions about its outputs.
• Track AI Assistance - Keep a record of frequent AI queries to identify knowledge gaps and areas for improvement.
• Pair Programming with AI - Collaborate with AI in a pair programming style to maintain control over the coding process.
• Long-term Skill Preservation - Ensure that reliance on AI does not lead to skill atrophy. Maintain curiosity and understanding of coding fundamentals.

These practices aim to leverage AI's advantages while preserving essential coding skills and critical thinking abilities.

"AI-first" is the new Return To Office

"AI-first" is the new Return To Office - Anil Dash

Big tech CEOs and VCs really love performing for each other. We know they hang out in group chats like high schoolers, preening and sending each other texts, each trying to make sure they're all wearing the latest fashions, whether it's a gold chain or a MAGA hat or just repeating a phrase that they heard from another founder. A key way of showing that they're part of this cohort is to make sure they're having a tantrum and acting out against their workers fairly regularly.

The return to office fad was a big part of this effort, often largely motivated by reacting to the show of worker power in the racial justice activism efforts of 2020. Similarly, being AI-first shows that a company is participating in the AI trend in the "right" way, by imposing it on workers, rather than trusting workers to judge what tools are useful for them to do their jobs.

There's an orthodoxy in tech tycoon circles that's increasingly referred to, ironically, as "tech optimism". I say "ironically", because there's nothing optimistic about it. The culture is one of deep insecurity, reacting defensively, or even lashing out aggressively, when faced with any critical conversation about new technology. That tendency is paired with a desperate and facile cheerleading of startups, ignoring the often equally interesting technologies stories that come from academia, or from mature industries, or from noncommercial and open source communities that don't get tons of media coverage, but quietly push forward innovating without the fame and fortune. By contrast, those of us who actually are optimistic about technology (usually because we either create it, or are in communities with those who do) are just happily moving forward, not worrying when people point out the bugs that we all ought to be fixing together.

Reimagining Democracy

Reimagining Democracy - Schneier on Security #democracy #politics

Bruce Schneier writes with a lot of clarity. The whole article is worth reading.

Indeed, the very idea of representative government was a hack to get around technological limitations. Voting is easier now. Does it still make sense for all of us living in the same place to organize every few years and choose one of us to go to a single big room far away and make laws in our name? Representative districts are organized around geography because that was the only way that made sense two hundred-plus years ago. But we do not need to do it that way anymore. We could organize representation by age: one representative for the thirty-year-olds, another for the forty-year-olds, and so on. We could organize representation randomly: by birthday, perhaps. We can organize in any way we want. American citizens currently elect people to federal posts for terms ranging from two to six years. Would ten years be better for some posts? Would ten days be better for others? There are lots of possibilities. Maybe we can make more use of direct democracy by way of plebiscites. Certainly we do not want all of us, individually, to vote on every amendment to every bill, but what is the optimal balance between votes made in our name and ballot initiatives that we all vote on?

Manas Saloi on his favorite thinkers

This tweet by Manas Saloi intrigued me a little bit so I went on a bit of a rabbit hole to locate all the resources he is referencing:

• The Internet of Beefs
• How many businesses-in-a-box are there?
• College as an incubator of Girardian terror | Dan Wang
• Game of Thrones: A Girardian Reading | Dan Wang
• Election Day 2020: René Girard, Part 2 – Alex Danco's Newsletter
• The Founding Murder and the Final Boss – Alex Danco's Newsletter
• Secrets about People: A Short and Dangerous Introduction to René Girard – Alex Danco's Newsletter
• The Mike Speiser Incubation Playbook - kwokchain
• Making Uncommon Knowledge Common - kwokchain
• 11. Eugene Wei - Amusing Each Other to Death - Jackson Dahl

Welcome to the Era of Experience

“Welcome to the Era of Experience” by David Silver and Richard Sutton

In key domains such as mathematics, coding, and science, the knowledge extracted from human data is rapidly approaching a limit. The majority of high-quality data sources - those that can actually improve a strong agent’s performance - have either already been, or soon will be consumed. The pace of progress driven solely by supervised learning from human data is demonstrably slowing, signaling the need for a new approach. Furthermore, valuable new insights, such as new theorems, technologies or scientific breakthroughs, lie beyond the current boundaries of human understanding and cannot be captured by existing human data.

To progress significantly further, a new source of data is required. This data must be generated in a way that continually improves as the agent becomes stronger; any static procedure for synthetically generating data will quickly become outstripped. This can be achieved by allowing agents to learn continually from their own experience, i.e., data that is generated by the agent interacting with its environment. AI is at the cusp of a new period in which experience will become the dominant medium of improvement and ultimately dwarf the scale of human data used in today’s systems.

An experiential agent can continue to learn throughout a lifetime. In the era of human data, language-based AI has largely focused on short interaction episodes: e.g., a user asks a question and (perhaps after a few thinking steps or tool-use actions) the agent responds. Typically, little or no information carries over from one episode to the next, precluding any adaptation over time. Furthermore, the agent aims exclusively for outcomes within the current episode, such as directly answering a user's question. In contrast, humans (and other animals) exist in an ongoing stream of actions and observations that continues for many years. Information is carried across the entire stream, and their behaviour adapts from past experiences to self-correct and improve. Furthermore, goals may be specified in terms of actions and observations that stretch far into the future of the stream. For example, humans may select actions to achieve long-term goals like improving their health, learning a language, or achieving a scientific breakthrough.

The era of human data offered an appealing solution. Massive corpuses of human data contain examples of natural language for a huge diversity of tasks. Agents trained on this data achieved a wide range of competencies compared to the more narrow successes of the era of simulation. Consequently, the methodology of experiential RL was largely discarded in favour of more general-purpose agents, resulting in a widespread transition to human-centric AI. However, something was lost in this transition: an agent's ability to self-discover its own knowledge.

2025-05-02

10 books that are dating red flags

10 books that are dating red flags | Dazed #dating #books

I first thought the title was a bit unserious, but the actual article turned out to be very insightful and funny.

On A Clockwork Orange

It’s a thought-provoking read and probably one of my favourite books, but if the person you’re dating thinks Alex is some kind of aspirational antihero, it’s safe to say you should probably run a mile.

On American Psycho

So if a man you’re dating loves American Psycho, just try to make sure this is due to its trenchant critique of consumer-capitalism, and not because he thinks Patrick Bateman is a based alpha giga-chad.

TIL, there is such a thing as "dude-bro" books:

I’ve always been perplexed by the idea that there are large numbers of obnoxious literary bros out there, bragging about having read Infinite Jest and terrorising the people around them with Jack Kerouac quotes. I have met a handful of men like that in my life, but they don’t exist as a meaningful constituency – most men simply don’t read fiction, if they read at all. If I met someone who loved Pynchon, DeLillo, Bolaño or any other author from the “dude bro” canon, I’d be more inclined to think of them as interesting than as pretentious.

On All About Love by Bell Hooks

I actually don’t think All About Love is inherently a red flag book. But it can be alarming to many when certain people (men) have this book in their possession. When I think about All About Love, I think of that picture a guy took of himself on a beach reading it, and everyone commented that he was only on page one and was already taking pictures of himself 😭. All About Love has become associated with a kind of performance for men. It often sits on the corner of their desks collecting dust, but it’s there so that any potential romantic partner they bring home will be impressed by their supposed desire to engage with hooks’ work and better themselves. Beyond that, I know people have a lot of problems with All About Love, especially because hooks writes that love and abuse cannot coexist. When I first read the book at 17, that particular line triggered one of the worst mental breakdowns I’ve had to date. Now that I’m 25, I understand that what people write in books isn’t always fact and that they can be wrong.

On Crime and Punishment

If you’re seeing someone who is reading a book by Fyodor Dostoevsky, good news! They have a brain cell. The bad news, however, is that that brain cell is deficient in serotonin; this person likely takes themselves quite seriously, has a morose outlook on life, and struggles with chronic depression (at least, if they didn’t when they started it, they will have developed it by the time they’ve finished it).

What Goes Around Comes Around... And Around...

What Goes Around Comes Around... And Around... | ACM SIGMOD Record #databases #sql #relational

The PDF can also be found here: whatgoesaround-sigmodrec2024.pdf

This is a great survey of all the interesting things that have happened in databases w.r.t to data modeling and query languages, which concludes that ultimately every converges to Relational Modeling and SQL.

In this paper, we analyze the last 20 years of data model and query language activity in databases. We structure our commentary into the following areas: 1. MapReduce Systems 2. Key-value Stores 3. Document Databases 4. Column Family / Wide-Column 5. Text Search Engines 6. Array Databases 7. Vector Databases 8. Graph Databases

We contend that most systems that deviated from SQL or the RM have not dominated the DBMS landscape and often only serve niche markets. Many systems that started out rejecting the RM with much fanfare (think NoSQL) now expose a SQL-like interface for RM databases. Such systems are now on a path to convergence with RDBMSs. Meanwhile, SQL incorporated the best query language ideas to expand its support for modern applications and remain relevant.

Although there has not been much change in RM fundamentals, there were dramatic changes in RM system implementations. The second part of this paper discusses advancements in DBMS architectures that address modern applications and hardware: 9. Columnar Systems 10. Cloud Databases 11. Data Lakes / Lakehouses 12. NewSQL Systems 13. Hardware Accelerators 14. Blockchain Databases

Some of these are profound changes to DBMS implementations, while others are merely trends based on faulty premises.

Karpathy vibe-coding a production grade web-app

Vibe coding MenuGen | karpathy #llm #coding #software #programming #vibe-coding

TLDR. Vibe coding menugen was exhilarating and fun escapade as a local demo, but a bit of a painful slog as a deployed, real app. Building a modern app is a bit like assembling IKEA future. There are all these services, docs, API keys, configurations, dev/prod deployments, team and security features, rate limits, pricing tiers... Meanwhile the LLMs have slightly outdated knowledge of everything, they make subtle but critical design mistakes when you watch them closely, and sometimes they hallucinate or gaslight you about solutions. But the most interesting part to me was that I didn't even spend all that much work in the code editor itself. I spent most of it in the browser, moving between tabs and settings and configuring and gluing a monster. All of this work and state is not even accessible or manipulatable by an LLM - how are we supposed to be automating society by 2027 like this?