I lost the voice.
Not the words. Not the ability to string sentences together. The voice. The one that published three posts a week from coffee shops and airports and yoga studios without asking anyone's permission, without knowing how it would turn out, without caring whether the paragraph was structurally sound before it went out the door.
That voice is gone from me. I killed it with competence.
I got good enough at writing to know when a sentence was weak, and that knowledge destroyed the thing that made the writing work. The recklessness. The willingness to publish something before it was ready, because the thought was alive and that mattered more than the execution.
Now I write correctly. Which means I barely write at all.
--
Let me also say the other thing, the part I have been avoiding.
The world accelerated while I was being careful.
In 2011 I published an ebook called Augmented Humanity. The argument was that some people were developing superhuman mental tools and others were being left behind. I believed it then. I still believe it. I just stopped acting on it.
AI went from a curiosity to infrastructure in roughly two years. ChatGPT reached 100 million users in two months -- faster than any consumer product in internet history. It hit 700 million by early 2025. By 2026 it was approaching a billion weekly users. This is not a trend you observe from a comfortable distance. This is a rewrite of the operating environment happening in real time.
Ray Kurzweil published "The Singularity Is Nearer" in 2024 and reaffirmed his prediction that AI reaches human-level intelligence by 2029. He then said 2029 was starting to look pessimistic. Andrej Karpathy, the former Tesla AI director, wrote in 2025 that the programmer profession is being "dramatically refactored" and that anyone whose AI knowledge is thirty days out of date already holds obsolete views.
Thirty days.
The people who picked it up early are running faster now. The people who waited are explaining why they are still thinking about it. And the people who refused are writing essays about authenticity on platforms that are feeding their refusal into a training set without asking.
Everyone is getting used whether they said yes or not. The question is whether you are using back.
I was not using back. I was preserving a voice that could not survive the preservation. I was the person I warned about in 2011, being left behind while insisting I had a reason.
--
So here is what the actual workflow looks like now.
Voice memo. Thirty seconds. I am on the river, I am in the kayak, I am somewhere with a good view and no interest in sitting at a desk. I say the thing. The agent gets the audio and turns it into a draft.
The agent does not have my opinions. It does not know what it felt like to live out of a bag in 2010, or what the blog meant when it was growing, or what it cost to stop. It cannot write from experience it does not have.
What it can do is catch the thought before I edit it to death.
I send thirty seconds of talking. The agent returns a draft. Then I come back and apply the only thing I actually have that the agent does not: judgment. Is this true? Is this what I meant? Does this have a nerve in it, or is it just tidy?
I cut what is tidy. I keep what has a nerve.
The agent does the lifting. I do the deciding.
This is not dishonest. The silence was dishonest. Three years of not publishing because the voice felt wrong was dishonest. The blog is a public record of thinking, and I stopped keeping the record because I was precious about how it sounded.
That was the failure. Not this.
--
The capture method is voice memos and that is not an accident.
The 2010 blog was location independent because that was the point. The laptop came with. The bag was small. The income, such as it was, did not require a specific room or a specific continent.
That is still the operating model. Kayak in the morning. Talk about what happened or what I am thinking. Send it to the agent. Keep moving. Post exists by evening.
That is a publishing operation that fits inside a backpack. That is what the blog was supposed to be. That is what it can be again, if I stop treating the friction as proof of virtue.
High volume matters. Not because more is always better, but because the archive is the asset. One post is a note. A hundred posts is a position. A thousand posts is an institution with a point of view that compounds over time.
The voice memo pipeline is how you get to a thousand posts without the blog becoming a full-time job that destroys the freedom it was supposed to produce.
--
The FBTS archive is all here, restored from PDFs and Wayback snapshots, over a hundred posts from a blog I ran from 2009 to 2012. I will always call it FBTS. Someone else owns the domain and I would rather not think about that.
The agents have read it. They know what the posts sounded like. They know the voice was direct, confessional, unafraid of announcing things before they happened.
They use it as a baseline. Not to impersonate a younger version of me, but to push against the drift. When a sentence sounds like a LinkedIn newsletter, the agent knows that is wrong because it has a better reference. When the ending wanders off instead of landing, the agent finds the sharper version.
I still have to want to say the thing. The agent just stops me from killing it in revision.
--
If this makes you uncomfortable, that is worth examining.
The discomfort is usually one of two things. Either you think AI-assisted writing is cheating, in which case I would like to know what you think is being cheated and who holds the trophy. Or you are still waiting to figure out how you feel about AI before you start using it, which is a decision you are making slower than the acceleration warrants.
The acceleration does not wait for the feeling.
Every month you spend deciding whether AI is philosophically acceptable is a month of output you do not have. It is a month of archive that does not exist. It is a month of thinking that happened but did not get recorded, which means it did not compound, which means it is gone.
I did that for three years. I would not recommend it.
The blog is publishing again.
That is not a small thing. That is the whole thing.
Pick up the tool.