mister automate

There's a lot of advice about whether you should follow your passion or be realistic and chase the money.

Neither is much help by itself.

You can love doing something and still have nobody willing to pay for it. The market doesn't care how passionate you are. It cares whether what you're doing is useful to somebody else.

You can also find something people will pay for and have absolutely no desire to spend the next ten years doing it.

That's a problem too.

Anything difficult enough to produce an unusual result is probably going to take longer than you expect. You'll get bored. You'll have stretches where it isn't working. You'll watch other people quit. If the only thing keeping you interested is the eventual payout, that's a long time to manufacture motivation.

So there are some realities you have to deal with. People have to value what you're doing. The economics have to work. You need enough ability to do it well. You probably need to stick with it for a long time.

But persistence creates a strange complication.

Sometimes the reasonable thing is to quit. The evidence isn't there. You've tried repeatedly. Other people don't see what you see.

And sometimes the person who refuses to quit is eventually right.

That's what makes reality difficult to read. Some walls are actually walls. Others are just the point where most people turn around.

You need enough respect for reality to recognize the first kind and enough conviction to keep pushing on the second.

I think that's a much better definition of being realistic.

Find something you care enough about to keep doing, make sure it creates real value for somebody else, and pay close attention to what happens when the two meet.

Then adjust.

Reality gets the final vote.

We're obsessed with closing loops. Say goodbye before someone dies. Get the last word in an argument. Finish the project instead of shelving it. Get an answer, any answer, instead of sitting with "I don't know."

Here's the uncomfortable part: a lot of the time, the other side of that loop doesn't even know it's open. If someone dies before you get to say goodbye, they don't experience an incomplete goodbye. There's no ledger on their end waiting to be balanced. The open loop was always yours. You weren't closing something for them. You were closing something in your own head.

An open loop is really just an unanswered question about your own footing — did I handle that, is this going to come back on me. Not-knowing feels like exposure. Closing the loop, even badly, turns an unknown into a known. That's why you re-send an email you already sent, or over-explain a decision that was already fine — nothing about the outcome changes, but the exposure does.

Here's the trap: your brain can't tell the difference between a loop that's resolved and one that's resolved well. It only tracks open versus closed. A trivial unanswered text costs the same anxious energy as a genuinely important unfinished project, because to your nervous system, both just register as "open." Your task list works the same way. Half of what's on it isn't urgent, it's just open — and open is uncomfortable enough that you'll work on the wrong thing just to make a loop disappear.

So the fix isn't "finish more things." It's closing loops on purpose, cheaply, when the work wouldn't have mattered anyway:

• Write the "no" down. A task you've quietly already decided not to do stays open forever if you never officially closed it. Cross it off. A decided "no" is an ending, same as a finished "yes" — and if it was actually important, it'll raise its head again on its own. You don't need to keep it open just in case; the ones that matter come back without your help.
• Separate the list from the priority. Write down everything, then star the few that actually matter. That's it. The starred items get your attention. Everything else can sit on the list unstarred and stop competing for the same anxiety.
• Notice the loops that were never yours to close. Some open loops are you chasing someone else's reaction or peace of mind. You don't control that ending. Naming it as not yours is sometimes the only close available.

Not every open loop deserves to be closed well. Some just need an ending, any ending, so your head can stop paying rent on something that was never going to pay you back.

A few hours of iteration later, here's what happened.

Made in honor of Escatto. Unaffiliated.

Audio may lag behind animations when using Bluetooth.

Warren Buffett has more capital than he can easily put to work.

That sounds like a great problem to have. But with that much capital, most opportunities aren't worth pursuing. Small investments barely matter to Berkshire Hathaway, and there are only so many places where billions of dollars can produce an attractive return.

So Buffett passes on almost everything.

AI is creating a strangely similar problem for the rest of us.

Until recently, most ideas came with a natural filter. You might have an idea for an app, but you couldn't code. You might want to analyze a pile of data, automate part of your job, make a video, or build some little tool for yourself.

Eventually you ran into something you couldn't do, couldn't afford, or didn't have time to learn.

A lot of those barriers are disappearing.

I notice it constantly now. I'll have an idea and realize I could probably have a working version by tonight.

The problem is that I have more of those ideas than I have nights.

When the cost of trying something drops far enough, saying yes gets very easy. You can start another project before you've decided whether the project deserves to exist. Do that enough and you can spend a lot of time proving that you could do things without producing much that compounds.

Buffett has dealt with his version of this problem for decades. One of the ways he describes his boundaries is the "circle of competence."

There are businesses he understands deeply enough to make a judgment about, and there are businesses he doesn't. If something falls outside the area where he believes he has the understanding required to make a good decision, he can pass.

He has also talked about having a "too hard" pile.

AI is making that pile smaller for everyone.

You can enter an unfamiliar field and get oriented in minutes. You can write code in a language you've never used, work through financial statements, research an industry, or prototype something without spending months acquiring the underlying skills first.

That's enormously useful. It also makes it easy to confuse access with judgment.

AI can get me surprisingly far into a field I know almost nothing about. What it can't give me instantly are the thousands of little corrections that come from having actually done the work. I don't recognize the edge cases immediately. I don't have years of mistakes sitting in the back of my head telling me when something looks technically correct but feels wrong.

In the areas where I do have those reps, AI is completely different. I can move much faster because I know what I'm looking at. I can reject bad output quickly. I can ask better questions.

That makes the circle of competence more important as AI gets better.

There are going to be more things we can do than things worth spending our time on.

Buffett has capital.

We have hours.

Spend accordingly.

We tell people not to worry like it's helpful advice. "Don't worry about it, worrying won't change anything." It sounds wise. It's usually wrong.

Worry is basically a brainstorming session of ways to reduce risk. When you worry about something, you're running scenarios, checking exposure, looking for a way in. Tell someone to stop worrying and you're telling them to stop looking.

The mistake is bailing out too early. Most people worry on the surface for a minute, hit "there's nothing I can do," and stop — so of course it feels useless. But that first "nothing I can do" is rarely the real answer. Dig a couple layers past it and there's almost always a lever: something you can prep for, ask about, hedge against, or just control the framing of. The worry wasn't pointless. You just quit the brainstorm before it produced anything.