April 27, 2025

Daily Log Digest – Week 16, 2025

2025-04-20

Review Code and LLMs

Alex Bird: "I have found that reviewing code is harder than w…" - Mastodon Canada #llm #code #reviews

I have found that reviewing code is harder than writing code. Increasingly, I am trying to write code that is easier to review -- that when someone looks at it, it is easier for them to tell if it does what it ought.

One of my objections to widespread LLM use for code generation is that we replace the easier task of code writing with the harder task of code reviewing.

(I include ensemble/mob programming here as a type of code review -- continuous code review)

2025-04-21

AI Phobia

AI Phobia Is Just Fear That ‘Easier’ Equals ‘Cheating’ #ai #phobia #skepticism

Why do these businesses care so much? I suspect they aren’t really worried about AI—they're clinging to an old belief that if work isn't visibly difficult to produce, it must be less valuable. When we dig beneath the surface of "no-AI" policies and detection tools, we find an age-old assumption that worth must be measured in struggle. This mindset shows up again and again, from “hustle culture” and the “rise and grind’ mindset that defined the 2010s to recent return-to-office mandates that prioritize presence over performance. In a culture that values butts in seats and availability on Slack, it becomes easy to mistake friction for effort and effort for worth.

Oddly enough, the very thing we’re resisting—the ease of AI—might be what sets us free. AI isn’t the first tool to challenge how we think about work, but it may be the most direct. By shifting the locus of effort, AI forces us to confront our dysfunctional relationship with work. It holds up a mirror to our culture’s deeply rooted belief that struggle equals value—and in that reflection lies a rare opportunity: to reimagine work in terms of outcomes, not optics; human flourishing, not performance theater.

Anne Helen Petersen once memorably described this as "LARPing your job"—performing a theatrical version of productivity. Workers engage in elaborate displays of "being at work": staying visible on Slack, responding to emails at all hours, and maintaining a digital presence that signals industriousness. The tools have evolved—from software that monitors keyboard activity to AI that analyzes facial expressions in video calls—but the underlying philosophy remains pure Taylorism.

The irony is that these measurements often have little correlation with value creation. Knowledge work rarely follows linear patterns. Our most valuable contributions often come from reflection, seemingly "unproductive" conversations, exploration of dead ends, and invisible mental processing.

A Survey of Reinforcement Learning

The State of Reinforcement Learning for LLM Reasoning #llm #rlhf #rl #reinforcement

This piece is by Sebastian Raschka of Build a Large Language Model fame.

Slowly making my way through this. Found it via a tweet and this is a lot of signal for the relatively short length of the article. Really enjoying it!

2025-04-23

Banking and Crypto

Crypto Might Get Some Banks - Bloomberg #crypto #stablecoin

But narrow banking pops up elsewhere. One important modern form is stablecoins. A stablecoin is a crypto form of banking: You deposit dollars with a stablecoin issuer, it gives you back tokens entitling you to get your dollars back, and meanwhile it does whatever it wants with the dollars. In the unregulated early days of crypto, “whatever it wants” could be quite spicy indeed, but these days stablecoins are a big business and there is something of a norm of parking the deposits in very safe short-term dollar-denominated assets, ideally Treasury bills or reverse repos or a BlackRock money market fund. If you launched a new stablecoin today and said “we will take your dollars and use them to make loans to emerging crypto entrepreneurs,” you’d have a hard time competing with the big incumbent stablecoins that say “we will take your dollars and use them to buy Treasury bills.” (Especially if, like most stablecoins, you didn’t pay interest.)

The Future is Augmentation

Why LLM-Powered Programming is More Mech Suit Than Artificial Human #llm #coding #software #programming

There is a view in many circles that LLMs will replace programmers. I am hesitant to say that this will never happen, becuase a lot of things with LLMs have surprised me recently, and I expect more surprises to come. For now, however, I don’t see LLMs effectively replacing programmers; but they are transforming how we work. Like Ripley in her Power Loader, we’re learning to operate powerful new tools that extend our capabilities far beyond what we could achieve alone.

This transformation will change what we value in developers. Raw coding ability becomes less important; architectural thinking, pattern recognition, and technical judgment become more crucial. The ability to effectively direct and collaborate with AI tools emerges as a vital skill in itself.

The developers who thrive in this new environment won’t be those who fear or resist AI tools, but those who master them—who understand both their extraordinary potential and their very real limitations. They’ll recognise that the goal isn’t to remove humans from the equation but to enhance what humans can accomplish.

In my view, that’s something to embrace, not fear. The mech suit awaits, and with it comes the potential to build software at scales and speeds previously unimaginable—but only for those skilled enough to operate the machines in ways that don’t harm themselves or those around them.

2025-04-24

Be Easy To Work With

How to teach yourself about AI

How to teach yourself about AI - by Mike White #ai #learning #books

One of the most important abilities you need for a successful career is knowing how to update your skills. As I tell my kids, this isn’t just important advice for scientists, who work in a profession that is supposed to generate new knowledge and new technology, and thus is always changing. The New York Times recently covered the “Gen X career meltdown”, describing the challenges of media professionals who entered journalism, advertising, film, etc. in the 1990’s and early 2000’s. The media business looks nothing like it did 20-30 years ago.

…

In that spirit, I want to encourage everyone learn more about AI, including scientists whose training and work may be mostly non-computational. Especially if you’re still early in your career, you have time to build your skills and bring deep learning into your work. To be clear, if you’re serious about doing computational biology, you’ll need to do more than just read some books; you’ll need to spend some time training with real computational biologists. But it is possible to, in the words of Harvard computational biologist Sean Eddy, “ go where a question takes you, not where your training left you.”

…

For the antedsciplinary scientists (and non-scientists) out there, here are my recommendations for books to learn AI. They range from gentle and popular introductions to fantastic textbooks that rigorously cover the math. (Eddy’s piece, by the way, helped tip the scales in favor of my decision to come to Washington University in St. Louis, where Eddy was at the time. I wanted to be around people who thought like that.)

What Happens When Everything Becomes a Meme

What Happens When Everything Becomes a Meme? #meme #commons

In economics, the tragedy of the commons is what happens when a shared resource - like farmland, fisheries, or clean air - is exploited so extensively that it eventually collapses. Today, we're experiencing a modern variation of this tragedy, not just in physical resources, but in our essential societal infrastructure:

• The social commons: trust, relationships, community.
• The cognitive commons: curiosity, education, critical thought.
• The economic commons: stable markets, shared prosperity, institutional trust.
• The informational commons: language, reality, basic consensus.

Unlike traditional commons, which collapse through actual physical depletion, these invisible resources are slowly dismantled through systemic incentives that reward isolation, compliance, instability, and division.

on the social commons

And honestly, a society built on transactional interactions and shallow connections is inherently fragile. People who can't trust each other in daily life don't suddenly trust each other at the ballot box. People who can't commit to friendships or partners might have trouble committing to democratic institutions or civic engagement. I am being sweeping in my assertions, but think of the foundation - a society without stable relationships can't sustain stable democracies.

Without genuine community ties, citizens disengage. Civic participation declines, and as Guy Debord warned us - politics devolves into spectacle rather than substance.

on the cognitive commons

Anne Helen Petersen has a nice essay on that here, writing “the logic we’ve internalized is pernicious and persistent: if you’re spending time doing something, and there’s a potential to make money off that thing, leaving that money on the table is fiscally irresponsible.” The obsessive, monetized pursuit of hobbies isn't mere escapism but a reaction to general pressures: education-driven burnout, economic precarity, and performative living. It's a way for people to assert identity and agency within structural limitations. Optimization, efficiency, monetization. Repeat!

Without curiosity or critical thinking, we become vulnerable to manipulation, susceptible to polarizing narratives, and ultimately lose the ability to make independent judgments, which is really important for democratic citizenship.

on the economic commons

We burned our economic commons, not because it makes sense, but because our political leaders have confused economic policy with personal vendettas. The market for chaos is booming, and trust is evaporating.

on the informational commons

The informational commons - language, reality, and basic consensus - is collapsing because we've monetized division. Social media platforms aren't built for clarity or understanding; they're optimized for engagement, outrage, and polarization. Algorithms don't reward nuance; they reward certainty, controversy, and emotional triggers.

What replaces consensus reality? Loyalty realities. Tribal realities. Personalized realities!! We no longer debate ideas or solutions - we debate whose facts count, whose feelings matter, whose truth wins. Truth itself becomes a loyalty test, not a shared ground. And without a shared informational commons, cooperation becomes impossible. We don’t solve problems, we fight over who gets to define them! Language is weaponized and reality is fractured.

and finally

Each of these commons has been chipped away, monetized, and exploited. Social trust turned into transactional loneliness. Curiosity replaced by compliance and cognitive outsourcing. Stable economic governance overtaken by chaotic spectacle. Shared reality splintered into competing tribes and personalized truths.

The societal infrastructure isn't gone forever. Unlike a depleted fishery or farmland, these intangible resources can regenerate if we choose connection over transaction, critical thought over compliance, substance over spectacle, and shared reality over isolated tribes, etc. But, you know, for now… bitcoin?

Millenial Hobby Energy

What is Millennial Hobby Energy? - by Anne Helen Petersen #hobby #hustle #culture

Many bourgeois or upwardly aspirational millennials have a hobby story similar to mine. Maybe they had a period in their teens or early 20s where they did something just because they liked it. But most people came to understand activities as “achievements” early on: if you’re doing something that’s not directly related to grades, then it should be extremely legible as a line on your college resume.

If you did what we called an extra-curricular, it was less because it was fun, or because you wanted to, but because “it looked good,” or communicated something “interesting” or “well-rounded” about your personality. And when you instrumentalize leisure in this way, you lose touch with your understanding of what leisure even is. Did you like basketball? Did you like playing the piano? Or did you do it because it — or something like it — was what you did?

Within this framework, there was very little room for activities that resisted narrativization in a college essay. Hobbies that didn’t produce something, or help someone, establish you as superlative, or in some way highlight your entrepreneurial spirit weren’t really hobbies at all. They were fucking around: invisible, if not altogether shameful. Listening to music = not a hobby. Scrapbooking = not a hobby. Zine making could be a hobby, but only if you distributed it to every high school in your metro area and formed a movement around it, etc. etc. (On this subject, I always refer people to Malcolm Harris’s Kids These Days, which outlines how millennials came to understand themselves as a product to be ameliorated).

2025-04-25

Be a disappointment

The Imperfectionist: Be a disappointment #burkeman #disappointment

From Oliver Burkeman's latest:

The older I get, the more convinced I become that it’s a critical life-skill – at least if you’re roughly the sort of person I am – to get better at disappointing other people.

I don’t just mean you should go easier on yourself when you catch yourself feeling bad for falling short of others’ expectations (although you should do that, too). I mean that it’s worth deliberately and consciously practicing disappointing others, letting the associated feelings sink into your bones, and generally spending time hanging out in the space of ‘being a disappointment’.

You’re especially likely to benefit if you belong to the category of people psychologists call “insecure overachievers”. (At book festivals and other events, it’s always fun to see people’s eyes widen in recognition when I use that phrase.) That is to say you’re the sort who works hard, gets stuff done, and impresses others with your achievements – but that to some degree, for whatever combination of reasons to do with upbringing, culture or DNA, you do it all because you feel that otherwise you won’t quite have earned your right to exist on the planet.

Accomplishments that ought to be a source of delight – good grades, promotions, professional success – can feel ironically oppressive to insecure overachievers, because once you’ve met a standard like that, it becomes the new minimum standard you’ve got to meet, next time, in order to carry on feeling adequate.

Karpathy on AI assisted non-"vibe-coding"

How Y2K Shaped Modern Misogyny

How Y2K pop culture shaped modern misogyny | Dazed #feminism #y2k #pop-culture #culture

There were certain words that kept coming up over and over and over again during my research, and ‘empowering’ was one of them. Almost inevitably, whenever it came up, it was being used in a defensive sense, after someone had been critiqued for something. The first Wonderbra ad with Eva Herzigova in 1994 was on billboards everywhere; it was very old-school bombshell, like the death knell for third-wave feminism. But the defence of it was that it was ‘empowering’. She made a lot of money, so maybe it was empowering [laughs]. Then there was a movie poster [for 2007 film Hostel 2] where a woman was being tortured and confined. It was quite dark, and when there were complaints, one of the producers claimed that they were ‘empowering’, because in the end, she fights back.

Marketers love nothing more than a good buzzword, right? When they find a word that they can imbue with a certain kind of progressive meaning, that is always a word that you should be suspicious of. Even the word ‘feminism’ is something that is so loaded at this point. But feminism does have a very clear meaning: women should have equal rights to men, and have equal protection under the law.

Sophie Gilbert: A lot of people think of this book as cultural criticism – that’s what I thought it would be, when I was pitching it – but to me, now, it is much more of a history book. And the point of history is not to rehash the stories that people already know; it is to look for the hidden stories that weren’t told at the time.

History is so often presented through a male frame. These are the stories that weren’t told or weren’t put together at the time, because people don’t really care about women that much, and also people don’t take seriously cultural products that are deemed to be ‘trash’, whether reality TV or gossip magazines or different kinds of media that women enjoy. But at the same time, they have such a massive influence. What I really wanted to do was to look at these, critically and historically, and see what they told me.