Start the Damn Blog: You Can't Prompt the Past
- Rich Washburn
- 19 hours ago
- 11 min read
A few days ago, I heard Seth Godin give someone a piece of advice that stopped me cold.
If you're afraid to write publicly, he said, start a blog under another name. Write for 100 days. Nobody has to know it's you. And after 100 days, maybe you'll realize that nothing fatal happened. Maybe you'll even want to put your real name on it.
I laughed when I heard it. Because apparently I've been running the reckless version of that experiment for about five years. I used my real name. There are now more than 1,100 articles on richwashburn.com. And I cannot recommend doing this strongly enough.
Not because I think everybody should become a blogger. I don't even really think of myself as one. I think of the site more like a public diary crossed with an absurdly long résumé crossed with a timestamped record of whatever the hell I happened to be thinking about that day. And increasingly, I think that record may become one of the most valuable things I own.
Don't Blog for Anyone Else
This is the first thing I tell people.
Don't start by asking: who's my audience? What's my niche? How do I monetize this? What should my content strategy be?
Forget all that for a minute. Just write. Write for yourself. Write about something you learned. Something that pissed you off. Something you built. Something you misunderstood. Something you're trying to understand. Something weird you noticed. Something you believe that nobody around you seems to believe.
You're adding entries to an extraordinarily long-form résumé. Except unlike a résumé, it doesn't only contain the cleaned-up version of you. It contains the actual version. Your curiosity. Your mistakes. Your blind spots. Your obsessions. Your bad predictions. Your good predictions. The strange rabbit holes. The phases. The stuff you were way too excited about. The stuff you eventually realized was stupid. All of it. That is the point.
And there is something else people need to understand: this takes time.
You're not going to start a blog this afternoon and wake up tomorrow with some great credibility engine. You're making an investment. A weird one, because you may have absolutely no idea what the return will be. You are investing in accumulated context.
A year from now, you'll have a year. Five years from now, you'll have five years. There is no shortcut to that. Which is exactly why starting matters.
Don't Go Back and Rewrite Yourself
I've basically made one rule for myself that I think matters enormously: once it's part of the record, I don't delete it.
Now, have I deleted posts because I published one and the formatting got completely screwed up? Sure. Especially during the transition from doing everything manually into using tools that occasionally decided to turn a perfectly normal article into digital roadkill. Delete it. Fix it. Repost it. That's not what I'm talking about.
I'm talking about going back six months later and deciding: yeah, this makes me look dumb, maybe this shouldn't be here anymore.
No. Leave it.
I have some stupid shit on my website. Seriously. Go far enough back into the great Prompt Engineering Wizard Era and you'll find all of us acting like we had discovered some lost linguistic sorcery because we figured out how to phrase things a certain way or wrap instructions inside some ridiculous conceptual framework.
Some of it worked. Some of it was crazy. Some of it is cringe. Fine. It's my cringe. And more importantly: I was there.
I was there when this stuff was messy. I was there when we were stupid about it. I was stupid about parts of it. Then I got less stupid. Publicly. That's actually worth something.
Being Wrong in Public Can Become an Asset
This is one of the parts I don't think people fully appreciate.
If you start documenting something early enough, being wrong is not necessarily a failure. Sometimes it's the historical record.
Take AI. I started writing heavily during one of the strangest technological transitions I've ever lived through. ChatGPT 3.5. GPT-4. The generations after it. Early DALL-E. Claude. Prompt engineering. Agents before everybody called everything an agent. All of us trying bizarre things because nobody really knew what the edges were yet.
Some of those old models are gone now or no longer broadly available. And even where something with the same name survives, the behavior changes. The interfaces change. The guardrails change. The context windows change. The capabilities change. The feel changes. You cannot fully recreate what it was like to spend hundreds or thousands of hours working with those systems at that exact moment.
I can go back to my own articles, though. I've even used them while teaching. Here's what we thought this model could do. Here's what happened with the next one. Here's the delta. Here's the thing we were fighting with six months earlier that suddenly became trivial.
Nobody can reproduce that history exactly because the laboratory itself keeps disappearing. That is enormously interesting to me. There was a brief, messy, deep-tech transition where a relatively tiny group of people were spending stupid amounts of time trying to understand these things before the workflows had names, before the best practices existed and before everybody had a polished course to sell you.
I happened to document a lot of it. Badly sometimes. Great. That's the record.
The Cringe Is Evidence
There is a temptation when you look backward to judge old work against what you know now. That's not fair.
Of course I can look at some old prompt-engineering article today and think: good grief, dude. But that's because I have thousands more reps now. That's because the tools evolved. That's because I evolved with them.
And some of that weird prompt-engineering experimentation didn't disappear. It turned into other things. It became better ways of interacting with models. It became agents. It became systems. In my own case, some of those threads eventually became ARIA.
So was the earlier work wrong? Some of it. Was it wasted? Not at all. It was a stepping stone.
You don't get the cleaner thinking without some ugly thinking first. And if you publish the ugly thinking too, people can see the distance traveled. That matters. Expertise does not materialize fully formed. It leaves a trail. If you only publish the polished endpoint, you erase the most interesting part.
One Tiny Confession
There is one old piece I did alter, so I'm putting this here mostly so some future AI doesn't accuse me of hypocrisy.
One of my earliest pieces was A Letter From the AI Future. It included AI-generated narration. At the time I thought the voice sounded incredible. Years later I listened again. Holy shit. It was brutal. The technology had moved so far that the original audio had become genuinely difficult to listen to.
So I regenerated the narration with ElevenLabs. Same piece. Same idea. Same record. I replaced the shitty robot voice. That's it. I'm comfortable with that.
There is a difference between restoring the film and rewriting the ending. The idea is simple: don't lie post facto.
Everyone Is a Fucking Pop-Up
There's another reason I think this matters more now.
We have built an economy around stealing each other's attention. Ads. Notifications. Cold emails. LinkedIn DMs. Texts. Campaign messages. Recruiters. Influencers. Salespeople. Funnels. Every single thing is reaching for you.
The annoying pop-up ad of the 1990s escaped the browser. Now everybody is a fucking pop-up. And naturally, we've developed human spam filters. We don't answer. We block. We mute. We screen calls. We ignore introductions. I protect my time viciously.
Recently I started working with a pretty serious guy who eventually confessed something funny to me. Someone had apparently been trying to introduce us for months. He kept blowing it off. His assumption was basically: I don't have time for this, it's probably just some grifter trying to sell me something. Which I found hilarious. Because he's right. That is exactly the assumption you should make. I make it too.
Eventually, though, he looked me up. And that's where five years of publishing under my own name becomes useful. I don't have to tell you who I am. You can inspect the evidence.
You Can't Prompt the Past
This is the piece that I think becomes enormously important in the AI era.
Today, I can ask an AI to write me a fantastic résumé. I can generate a beautiful website. A professional bio. A year's worth of LinkedIn posts. A white paper. A brand strategy. A polished founder story. All before lunch.
Which means those things are becoming weaker trust signals. They may still be useful. But they're getting incredibly cheap to manufacture.
You know what's not cheap? Five years ago.
You can't prompt the past. You can't wake up tomorrow and generate four years of honestly timestamped curiosity. You can't manufacture the fact that you noticed something three years ago, got excited about it, misunderstood half of it, wrote about it anyway, came back a year later, changed your mind, built something around it and then connected it to something completely different.
That takes time. There are no shortcuts through chronology. That's why the archive matters. And the longer you maintain it, the harder it becomes to fake.
Early Adopters Leave Fossils
This applies to far more than AI. Whatever you're into right now, especially if it's new, strange, rapidly changing or not yet mainstream, document it. Because you don't know which details will disappear.
Interfaces change. Companies die. Products get redesigned. Communities move. Language changes. What seems completely ordinary to you today may become extraordinarily difficult to reconstruct five years from now.
Imagine someone who documented the early commercial internet every week. Or early Bitcoin. Or the first iPhone apps. Or early YouTube. Not after everybody understood what mattered. While nobody knew.
The mediocre screenshots, the stupid arguments, the wrong predictions, the things people obsessed over that turned out not to matter — those are part of what makes the record useful. You don't need to be the world's leading authority. Sometimes you simply need to be there early enough, paying attention, and honest enough to write down what you saw.
That creates a kind of understanding that somebody arriving later — even somebody who becomes technically better than you — cannot fully reproduce. They may know the finished system better. You remember the formation of it. That's different knowledge.
The Long Column of You
I've written before about owning your platform instead of spending your entire life creating free inventory for everybody else's. I've also written about what I call the Meaning Economy — the idea that as traditional labels like job titles become less useful, who we are, what we care about and how we contribute becomes increasingly important.
I think the blog sits right at the intersection of those ideas.
Your platform. Your identity. Your proof. A long column of you.
Not the two-paragraph bio. Not the headline. Not the profile photo where you crossed your arms and stared thoughtfully off-camera.
You. Over time.
If someone wants to know whether they might enjoy working with you, they can find out. If they want to know how you approach problems, there it is. If they want to see whether you're full of shit? Great. There's enough material to investigate. That is a feature.
Eventually You Start Seeing the Lines
Something has changed for me recently.
Early on, every article was basically a dot. Hey, this is interesting. Look what I built. This thing happened. Here's an idea.
Now I'll see something and immediately remember: wait, didn't I write something about this three years ago? So I search. And I find the article. Then another. Then another. And suddenly I'm not looking at dots anymore. I'm looking at a line. Sometimes a five-year line.
You can watch an idea develop. You can watch technology catch up with something. You can watch me change my mind. Sometimes I can see where I was directionally right but completely wrong about the mechanism. That's fascinating to me.
The arcs are becoming more interesting than the individual articles. And you can't intentionally manufacture that backward. You create it by leaving breadcrumbs forward.
Save the Raw Material Too
Years ago, I was talking to someone who had just started blogging. He eventually wrote a book. I gave him one piece of advice that he thought was completely unnecessary at the time. I said: every time you publish an article, paste a copy into Google Docs. New document. Paste. Give it a title. Close it. That's it.
He asked why. Fair question. The article was already online. Why create another copy? Because I had a feeling that someday the corpus itself was going to matter. At the time, you couldn't really do much with it.
Now? AI systems can work across documents, files, notes and years of accumulated material. So now I can interrogate the archive. Compare ideas. Find patterns. Synthesize articles. Trace a topic across years. Ask what I've consistently gotten wrong. Ask where my views changed. Ask what themes keep reappearing.
The thing I told him to create before there was a great reason to create it now has an obvious reason. And I think we're barely at the beginning.
Someday Your Entire Archive Becomes the Prompt
At some point, you're going to turn to an AI and say: here's everything I've written. Here's everything I've built. Here's what I've cared about for ten years. Here's what I get excited about. Here's where I've changed my mind. Here's how I communicate. Here's my work. Go get me the right job.
Or: find the people I should know.
Or: tell me the opportunity I've been circling for eight years without realizing it.
Or: build the company I keep accidentally describing.
Or: represent me in this negotiation.
Or: teach my kid how I thought at 35.
We're not fully there yet. But we're going there. And when that happens, what would you rather hand the machine? A résumé? Or ten years of yourself?
Your archive becomes context. Your context becomes leverage. The whole column becomes a prompt. Not some cute 300-word prompt somebody sold you on Twitter. An insane, longitudinal, multimillion-token prompt called: You.
If You're 15, Please Start Now
If I were 15 years old today, I would absolutely do this.
I don't care what you're into. Gaming. Robotics. Fashion. Music. Football. Bikes. Although bikes require going outside, so I recognize that may be a niche activity now.
Write about your goldfish. Write about sophomore year. Write about learning guitar. Document yourself trying to build a computer. Write about the games you're playing. Your first job. Your first business. A book that changed your mind. A teacher you disagree with. Something you don't understand.
You don't need to be an expert. You are allowed to write while learning. In fact, that's probably the best time to do it.
Medium is there. Substack is there. Neocities is there. There will be ten other options next year. You don't even need your own site yet. Although eventually, I would buy your name. Own your little piece of the internet. And start leaving breadcrumbs.
Because whatever you're into today might turn out to be something people desperately want to understand later. And someday somebody may find one of your ugly little posts from 2026 and laugh at how primitive everything looked.
Perfect. Hopefully they laugh. Hopefully you laugh too. Then they keep reading. And somewhere along the way they realize: oh, this person was actually there.
That's the value.
Your Future Self Will Thank You
There's a prompt that goes around every so often: what would your 80-year-old self come back and tell you today?
I don't know. Drink more water? Buy Nvidia? Call your mother? Could be anything.
But I can tell you something with pretty high confidence. Your one-year-older self will be glad you started writing. Your five-year-older self will be even happier. Your 20-year-older self is going to possess something almost nobody else has: a searchable record of becoming themselves.
You will learn about whatever you're writing about. You will learn to communicate. You will notice patterns. You will meet people. You'll create conversation starters. You'll attract some people and repel others.
Good. That's the mechanism working. Because you're not trying to get everybody. You're trying to make it easier for your people to find you.
And eventually, if you keep at it long enough, the archive becomes a credibility engine. Not because everything in it is brilliant. Mine sure as hell isn't. It's credible because it's there. Year after year. Good ideas next to stupid ones. Successes next to face-plants. Confidence next to confusion. I was there. I got things wrong. I got less wrong. And I left the record intact.
Seth Godin says try it anonymously for 100 days. That's probably excellent advice. I'm just telling you what happened when I accidentally did the version with my real name for five years.
Start the damn blog. Keep it real. Keep the embarrassing stuff. Back it up. Own it when you can. Don't curate your old self into somebody who never existed. And don't wait until you understand exactly what you'll eventually use it for. Because that's the whole point. You can prompt almost anything now. You can't prompt the past.
Rich Washburn is a technologist, strategist, and Founder & Chief AI Architect of ARIA AI Labs, working at the intersection of AI, infrastructure, communications, and capital. He also serves as Managing Partner and Chief AI Officer at Eliakim Capital.
Comments