The Springtime Blues, Vibe Coding, Social-Media Shields
Genius Loci readers:
In today’s AI-packed missive, some notes on AI’s great, big book heist, vibe coding, a note on my website redesign, the 2025 travel writing book awards, books as social media shields, and a free Field Note journal when you pre-order Robert Macfarlane’s new book.
Heyyy from Orlando, where we’re experiencing the type of weather you’d expect in Paradise. The pollen and itchy eyes and sneezing and nasal sprays are minor complaints. The weather compels you to drive with the windows rolled down and the music turned up. Even music sounds differently in the spring — it enters you more deeply and heightens our love for the world. The general urge is to be out in the world in the perfect spot: half in the shade and half in the sun.
And while I should be outside enjoying the world, I find myself feeling a bit down. It happens to me every spring. The feeling is one of anxiety. It is the feeling of being not where I should (or want to) be, physically or creatively. I feel as if I’ve been wasting time, which is a feeling that has followed me for much of my adult life. From my desk I look through the windows and I daydream about what I could/should be doing instead of slouching over my laptop, grading assignments that were probably written by AI. I wish I were elsewhere, anywhere out in the world…
You might be thinking: just go outside, you schmuck! I do go outside quite a bit, but to be outside is not necessarily to be out in the world. Modern knowledge work treats being outside as a reward for the work you’ve been doing inside. Being outside then, like sabbaticals, is meant to refresh and “renew” you for the grueling hours of inside work you have ahead. This vision separates life from living. By being “out in the world” I mean a state of being receptive, open, and willing to be shaped by experience. To be “out in the world” means not to just move through the world, but letting the world move through you. And every spring, for whatever reason, I’m reminded of what is missing in my life.
Does anyone else feel the Springtime Blues?
Libgen and LLMs
The other day, The Atlantic in an article reported that Meta and OpenAI used the ebook piracy site Libgen to train its AI models. Per the article, Meta got the go-ahead from “MZ” - Mark Zuckerberg himself. It is unclear which pirated books the AI companies used to train their LLMs, but I’m willing to bet that the companies scraped as much data as they could. If you’ve published books, you can use this tool published by The Atlantic to see if your books have been pirated and potentially included in the LLM data sets scraped by Meta and OpenAI.
There seems to be some moral and legal confusion about the issue. To be sure, piracy is theft and a violation of copyright law. Piracy and pirates should be prosecuted. Yet, while stealing books to read them — as it appears these large language models (LLMs) have done — is theft and should be prosecuted, the act of learning from books is not illegal. Meta and OpenAI’s lawyers argue their use of this material doesn’t violate any copyright laws because it doesn’t reproduce the work but uses it in a “transformative” way (a key term used in the defense of copyright claims). But some of our collective unease with the situation revolves around the fact that these LLMs are profiting materially from this data, which itself was acquired illegally if The Atlantic’s reporting is true. Though, with questions of scale aside, it is unclear how this is different than, say, an author reading books, learning from books, and then writing new books. In any case, let’s hope OpenAI’s 2024 deal with News Corp. will set a precedent with the licensing of information so authors can at least get paid for the work these corporations stole and exploit.
Some authors are now questioning the standard inclusion of ebook rights in publishing contracts, arguing that ebooks don’t turn large profits for the authors and lead to “instant piracy” (and thus easily gobbled up by LLMs). I think this is a good move, especially to help authors preserve more of their rights. Others, like the Authors Guild, are advising authors to include a “NO AI TRAINING” statement on a book’s copyright page. Cute, but probably ineffective considering how several AI web crawlers don’t even respect robots.txt files preventing AI bots from scraping websites. It is not like those FBI piracy warnings on VHS tapes and DVDs actually stopped anyone from making and distributing copies.
Vibe Coding
On the other hand, there seems to be some optimism about the coding potential of LLMs.
“Vibe coding” is a new term that refers to prompting LLMs to code using natural language. In theory, you can write or speak instructions to a LLM and it will output what code or software you’re describing. LLMs retain a “chat history” and remember your conversation, thus allowing you to refine the codebase. Other, more sophisticated LLM integrations “live” in your terminal and can evaluate your entire codebase without the standard chat prompt functionality we’ve become accustomed to with AI agents.
The cyberpunk future is here, and it isn’t ruled by the socially-awkward hacker soldering transistors and capacitors to some device or typing into a command-line shell on a CRT screen. It is ruled by people like you an me who need custom tools. At the end of the March 7, 2025 episode of the Hard Fork podcast, listeners called into the show to discuss their bespoke, “vibecoded” software. Even Craig Mod, who knows a thing or two about software development, vibe coded a custom “social media network” for his community, with rate limiting user posts and a disappearing timeline.
You need to have a basic understanding of how software works, and the ability to articulate what you want, in order to implement and exploit the possibilities. In the case of APIs like Claude Code, you’ll need to know your way around the operating system’s terminal (and a lot more). But if you don’t, you can probably ask a LLM to walk you through the process.
I tried it out. While redesigning my website, I encountered some responsive-design issues. I pasted my CSS into a LLM and asked it to evaluate my code. After some back and forth, it solved my problem. I also asked it to organize my CSS file, which it did wonderfully. The process helped me learn a bit more about CSS, debugging, and the backend systems I use. While I cannot pretend to be fluent in any coding language, I’m not a complete beginner. But I still think there is a barrier to entry; those with little-to-no coding knowledge will have some difficulty getting a project off the ground.
Is there an ethical, intellectual property issue with using AI to code? It is unclear. AI models have been trained using code documentation and a large database of public source material. In the case of web development using CSS and html, source code is visible to anyone using their browser’s code inspector and dev tools. Whether learning from this information constitutes “stealing” is a matter of debate, as anyone had the ability to “learn” from it in the pre-AI age.
Website Redesign
If you’re reading this in an inbox, you won’t be able to see my website’s new design. My old site used a design borrowed from astrophysicist Brian Koberlein (he kindly sent me his source code and okayed my use of it). While I love his design, the site was built with his design logic (which is to say, a design logic that didn’t come naturally to me). And there was a bit of cruft in the code - there were references to features that had been discarded/disabled/abandoned.
I wanted to design my own site from the ground up, based on my own design and logic. I want to know where everything is and what every line of code does. My site isn’t fancy, and I’m still ironing out some kinks (I will ask a LLM to help me, surely), but I’m happy with what I’ve been able to create in such a short time. I know my site now more intimately, like an old book I’ve read many, many times.
I’ve also added a commenting feature to my website using Bluesky. For those who don’t know, Bluesky is a Twitter/X clone that has been gaining momentum recently. It is not owned by Meta, unlike Threads, which is a plus. It has a different vibe from X and Threads, much like Twitter did in its early years.
Here’s how commenting on my website now works:
- I’ll share posts on my website (like this newsletter) to my Bluesky timeline.
- To comment on my website, simply reply to my Bluesky post.
- Your comments will appear both on Bluesky and on my website.
Comments don’t “live” on your site. They are just visible on your site via a Bluesky API. Comments are easy to moderate. Just delete the comment and the post disappears. Have a troll? Just block him on Bluesky. Or, to disable comments on a page entirely, just don’t share the webpage on Bluesky.
Edward Stanford Travel Writing Awards
The Edward Stanford Travel Writing Awards announced its 2025 winners last month. Noo Saro-Wiwa’s book Black Ghosts: A Journey Into The Lives Of Africans In China won the non-fiction category, Travel Book of the Year. Elif Shafak’s book There are Rivers in the Sky won the fiction category, Fiction with a Sense of Place.
Books as Social Media Shields
I haven’t read much travel writing since stepping away from the Travel Writing World podcast. I needed a break from the genre. I couldn’t continue reading books whose genre interested me more than its subject matter. I’ve been making a deliberate point to avoid travel books, to instead find books 1) related to my current research topic, 2) books that I find interesting for whatever reason but are unrelated to my research interests, and 3) books that serve the particular function of being social-media shields.
The social-media-shield book must be:
- cheap
- easy to carry around (i.e.: mass-market paperbacks)
- brainless / entertaining
- easy to dip into/out of
- devoid of emotional attachment and therefore easy to abandon in a community bookshelf when done
Carrying a pocketable book helps stave off the impulse to reach for a device. If you find yourself with some time to kill, you will reach for your book instead of your device. Keep the book on your nightstand and you will read it instead of your phone. It helps us cultivate focus and attention, which is the opposite of what our phones are doing to our brains. I understand social-media addiction is more of a problem for the younger generation who haven’t experienced the pre-digital world, but it is easy for the rest of us to fall into the trap.
I had two social-media shields last month. The first was Brandon Sanderson’s Mistborn. The second one was Convenience Store Woman by Sayaka Murata, which I found in a community bookshelf in my neighborhood. I have no comment on these books, with the exception that they fulfilled the criteria and served their purpose.
I also finished reading Tod Papageorge’s Core Curriculum, which I reviewed here. If you’re interested in photography criticism, the book may interest you. It covers the following photographers: Eugène Atget, Henri Cartier-Bresson, Brassaï, Walker Evans, Robert Frank, Garry Winogrand, Josef Koudelka, and Robert Adams.
Pre-order Robert Macfarlane’s New Book
Pre-orders for Robert Macfarlane’s new book Is a River Alive? are now open. Macfarlane and the artist who created his book cover, Stanley Donwood, have partnered with Field Notes again. If you pre-order the new book from a participating indie bookstore, you will get a signed copy of the book and a Field Notes notebook. For more information, check Macfarlane’s social media post. I believe this deal is for US-based customers only.
Okay. This email is around 2,000 words. Time to shut it down and to go for a walk outside…
Thanks for your support and farewell,
JB