One idea worked

One idea worked. The interesting part is the pile that didn’t.

For digital packrats like myself who have hoarded every line of code since high school, AI is an amazing tool for reflecting on the past. I dumped every project I could find into Claude and asked it to do a book report. That gave me a look at how I actually spent my life. This isn’t the résumé version. This is the packrat version: half-truths, dead ideas with good in them, moments of insight, and dreadful decisions. This is the graveyard that I had a different agent organize and assess so I can interact with it.

The reason I’m doing this is that I had a good career with a reasonable exit, and I am still restless. I am using AI to study the deeper DNA of how my mind thinks rather than reacting to the vibes of the current world. I have the benefit and the curse of optionality. Being mindful and intentional is how I keep from walking into another burden of command when I meant to walk toward joy.

The corpus is linked here as a eulogy. I want to share the human-level insights from reviewing my own life.

First, my graveyard is the story of a man who kept coming to the plate to take a swing. These projects tell a tale of many lessons learned and many fruitless paths. For a long time I treated this as the pathology of a flighty bird with full-on ADHD. I didn’t stick to a project more than a year, always jumping to the next shiny object, either bigger or better. Now that I can witness the whole pile, I can relax that regret. It was persistence with a bad reputation, and the same loop got me through big tech and kept me swinging until luck happened. The parable that luck favors the prepared rings true for me if you replace preparation with sheer volume. The pathology wasn’t resolved on victory, and I still became restless.

Second, the graveyard is full of stagnated seeds that never got planted in public so I could reap public adoration. Seeing them laid out bare and rotting, the good and the bad called honestly, is a relief more than a compliment. The finish line I actually crossed, complete with a full victory lap, taught me what success actually costs. Once you know that price, the pile sorts itself, which is also why hierarchies are so oppressive to creative diversion: more wood behind fewer arrows. Most of those projects were not hidden home runs. Maybe a presentation and a tool with a lifelong burden. They were correctly dead for me, and only a few could have ended better. I can stop mourning them as if every abandoned idea was a stolen future, or a TED talk on how brilliant I am with my creative mind of splendor.

Third, my great victory did not point forward, and neither did the recent huge failure. Both dumped me into a rare state: enough options to build a fortress of solitude, and a restless creative human still living with myself. My career was a random walk up a ladder. Harder and harder things. A joy that came from the next rung telling me what to do. That joy is gone, so I harvested the past because I need a direction that isn’t just another burden of command or a death march to grind my mind against. The pile is where I looked for the passions that survived contact with reality in this new future with the great AI machine.

Fourth, having the machine analyze and assess the body of my work has been cathartic, and strangely emotional. It is, in essence, a feeling of being witnessed by something greater. It isn’t the same as jamming with another technical peer. It is like being judged in the purest and most mathematical way by a mathematical machine god. It made me feel seen in a way that the legion of failures made me feel small. The process gave me closure over the emotions I had for this moment where the entire field is changing, and it gave me pause to reflect on what is next.

Instead of inventing a new war for the sake of it: where is the beauty of mind still relevant, given the easy button that spins up a project?

The easy button is why the fortress is tempting, and it is also why the world is filling up with work nobody will claim. We are already throwing slop grenades into companies, and the same old rules and costs still apply whether a human or a model fucked up. More projects with fewer owners feels really bad given how expensive victory is. So the question from the graveyard is not “what else can I invent.” It is “what kind of mind is still worth bearing the burden of command.”

The lifetime signal in the corpus is not that I am restless. I am an absolute beast of a control freak. I invent and think at the layer where the language itself is mutable: domain-specific languages where good things are cheap and bad things are impossible. I hate being indoctrinated into a culture of tolerating footguns. I want absolute perfection.

Seeking perfection used to be a liability of never shipping for marginal gains. Now it is table stakes. Labor was expensive, formalisms were niche, and the market paid people to poke code like monkeys. The glorious easy button flipped the economics and shifted the field. Somebody has to educate the machine via new abstractions, new layers, new operating boundaries, and then stand behind them. That is a job for control freaks.

The future looks brighter if we let those people out to play. An engineer can actually go learn TLA+ and use the machine to work at that level, instead of generating another thousand lines of plausible residue. The model will write, and it will not care nor complain about it. Caring and taste were always the scarce part, where we argued. The deep passion that echoes from the litany of graves shouts at me that this is the swing I will keep taking.

This process has given me a technical direction that does bring me joy, so I can be intentional with my next steps rather than reactive to the vibes of the day. The next step is not another product just for the sake of making products because that’s what is visible. It will be a place where a language can make the bad impossible, with people to stand behind the boundary when models throw slop or decide to do other things. It may be a new domain, my fortress, an idea garden, or a lab. I will not decide based on the next rung of some ladder. I know what victory costs.

One idea worked well enough, and the rest taught me which kind of mind I actually am. The pile is the map. And I’m going to follow the constraint, not the noise.