While I haven't yet gotten around to using Val Town myself, I've had my eye on it thanks to Tom MacWright, the CTO of Val Town's small team and whose own blog I've enjoyed reading for a while. After reading this small post on Val Town's own excellent blog, I'm much more interested in trying it out.

The post discusses the idea of prompting Claude and other agents into what they're calling "Slow Mode", in which human learning and decision making is prioritized over speed and productivity. This really gets to the heart of the existential questions that I, along with many other programmers, am having:

There’s an inherent learning-versus-productivity tradeoff when using AI. I am no stranger to sacrificing my own learning or solid understanding of an implementation to prioritize shipping. When you have a lot on your plate, that can be the right side of the tradeoff. But there’s also a short-term gains versus long-term maintenance tradeoff, and I’m betting Slow Mode will help with that...By embracing Slow Mode at Val Town, we’re telling vibe coders we believe in your ability to learn fundamentals about code that might’ve previously seemed unreachable.

I'm trying to find the right balance here, too. I don't even recognize code that I wrote all on my own a few months ago. However, I oftentimes still retain some mental models of that process, the structural patterns and higher-level designs driving the long-forgotten individual lines of code. However, so much of that mental model was built up in the friction of slowly writing out the code myself, debugging and adjusting as I went along. Those hours can be frustrating and painful, but if the goal is to retain more of that information, that might be the point.

The Superpowers plugin is the second-most installed plugin on the official Anthropic marketplace. It's brainstorming workflow has become my go-to starting point for my coding sessions with Claude. After some back and forth discussion, it generates a high level design doc and then, upon user approval of that, a super detailed plan for sequentially implementing that design. It feels like I'm holding a better grasp of what's getting built, but I have no doubt it's not sticking in my brain a fraction of the amount it would have had I done everything myself. The overarching question, "do I really need to fully understand this?", seems pretty context-dependent (emphasis mine):

The polar opposite of Slow Mode is YOLO Mode, where your agent loops again and again without stopping to ask for permission. It fulfills vibe coding’s original definition. In Townie we do have YOLO Mode (a.k.a. “Allow all”) because, well, to each her own. I enjoy YOLO’ing sometimes, typically when the stakes are low and time is short.

A lot of the pure YOLO Mode work has been on this website, but that's very intentional. The stakes are indeed low, and while certain aspects of the site present fun coding side quests (e.g. the recently added listening page), I'm much more interested in using this as a place for my words. As I said from the very start:

I really want this refreshed site to focus on being designed for writing, insomuch that it meets two needs:

1. The look and feel of it should prioritizing reading, ideally a quiet and cozy corner of the internet.
2. The code of the website should be designed and arranged to make it very easy to push out new posts/chunks of text.

Do I have other fun ideas for things that could live here, too? Sure. But by focusing on the basics, it gets me over the hump of just having something live and actively getting updated.

In the meantime, I'm looking forward to experimenting with slow mode. Shipping stuff is satisfying in its own right, but finding the sweet spot between pure output and a more deeply internalized understanding is a search that continues.