The hardest thing about a one-person company is not self-discipline, but the fact that no one is waiting for you

✍️ This article is reposted from “Vista Newsletter” No.136, originally published in iamvista.substack.com (opens in a new tab). If you like this kind of content, welcome to free subscription to “Vista Newsletter” (opens in a new tab) to receive my observations and methods first-hand every week.

On August 12th, I did two things that were almost the same.

In the morning, I finished proofreading the manuscript of new book “Unmanned Company” and sent it to the editor-in-chief of the publishing house. At night, I sat in front of the computer debugging a personal dashboard until late at night.

Both are finishing touches, neither is the excitement of starting from scratch. But by the end of the day, it was clear to me that while I was energized by sending the manuscript, fixing the bugs was draining me.

I originally thought the difference was in difficulty, or in the old saying that the beginning is more exciting and the end is more tiring. However, both of these are finishing touches, and the workload of proofreading the manuscript is far greater than that of the dashboard. I wrote a sentence in my review that day. Looking back now, I think it was the most important discovery I made in August:

The dividing line between strength and weariness is not at the beginning or end, but actually lies in whether there is a recipient.

The reason is very simple, because the editor-in-chief is waiting for the manuscript. He will reply to the letter, arrange printing, and give me a printing date of early December. But the dashboard is different. Only I can see it from start to finish. It’s also the finishing touch, one thing is to hand over, and the other thing is to hand over to your future self.

I spread out the two weeks’ records for comparison.

I have the habit of reviewing the results every day, so this guess can be verified. I checked out all the records for the two weeks from August 4th to August 17th, and the results were more tidy than I thought.

On August 6th, I built founders.tw (opens in a new tab) from scratch: image station, candidate funnel, and roundtable discussion area with three levels of authority, all in one day. In the review report on the same day, I wrote that none of the five things I listed the day before that I wanted to push forward today had not been touched. And founders.tw is not on that list at all.

What’s the difference? founders.tw (opens in a new tab) There are two old comrades waiting for me, but not the other five.

The conclusion I wrote down that day was: If someone is waiting for you, always be on time, and if you only tell yourself, always be late.

On August 10th, the manuscript of “Unmanned Company” broke the ice. I went back and forth between hospitals that day, and my time was cut into pieces. Logically speaking, it was the day that should be the least likely to make any progress. But I read Chapter 7 from the prologue in one breath, and before that, this line had made zero progress for seven consecutive days. What’s even better is that I looked through the records and found that the two ice-breakers for the manuscript happened on the days when I was chopped up by the clinic.

Therefore, the real stuck point is not the lack of a whole block of time. I had always regarded this matter as a matter of time, and seriously thought about clearing out a full weekend. The truth is, I get the most done on the most fragmented day, because that day there is a clear target waiting to be harvested.

August 11th is even more exaggerated. I only pushed five commits that day, which was the lowest day in those two weeks. But I finished proofreading the entire manuscript that day. By comparison, on August 4th I pushed 112 commits, 84 of which were all-in on the briefing system, and none of which were new features.

Using output as an indicator, August 11th was the worst day in those two weeks. But if delivery is the metric, it was the best day.

You need to be more disciplined, this may be the wrong diagnosis

If I tell people about this situation, 80% of the time they will receive the sentence “You need to be more self-disciplined”, or you should try Pomodoro, time box or other methods that can improve work efficiency.

But what I want to say is this: In the past two weeks, I pushed more than 600 commits in 23 projects, opened two new websites, delivered a workshop of 20 people, and finished proofreading a book.

Self-discipline is not my problem, nor is time management.

The real problem is actually structural: working alone, there is no recipient by default.

In an organization, recipients are assigned to you automatically by the system. The supervisor will ask about the progress, colleagues will wait for the period you have, and the customer will ask for the delivery date. You don’t have to create stress yourself, the environment will do it for you. You think that is self-discipline, but actually it is being waited for.

After leaving the organization, this ration ceases. Most of the discussions about One Person Company teach you how to do things more efficiently, but few people remind you: you have to fill in the recipients yourself.

You may be curious, what happens to the thread that cannot be repaired? My backlog for the past two weeks looks like this: August’s positioning convergence has been postponed from July 19th to now, the 29th day. Two thousand four hundred and ninety-eight additions to the old list have been listed in the to-do list for twelve consecutive days, and have not been touched for a single day.

Both of these things are not difficult. They are both things that I consider important, and neither of them is waiting for anyone.

Three ways to assign recipients for yourself

If this is the diagnosis, the solution is not to force yourself to work harder, but to find ways to make things grow out of the recipient. I have used three methods in the past two weeks, and the effects are very different. I will rank them according to the effects.

The first type: narrow the delivery target to a person who can be named

Readers, market, target customer groups, these words are too far away, so far away that no one will reply.

The recipient of a book manuscript is the editor-in-chief, not the person who buys the book. The recipients of founders.tw are two old comrades, not future members. The difference is huge: abstract objects don’t rush you, but concrete people do. And when you hand something to a person with a face, the quality standard will automatically rise because you can imagine his expression when he opens the file.

It’s easy to write it down: After this thing is done, who do I want to send it to? If you can’t send it out, you probably won’t be able to finish it.

The second way: Let the external calendar bite you

I took stock of the most stable positive patterns in the past two weeks, and found something a little funny: As long as you have an appointment with a real person that day, you will almost certainly get concrete results that day.

I met two friends on August 8th, the day researcher.tw (opens in a new tab) was officially launched. There were two external classes on August 13th, and the entire tool chain was directly merged and launched on that day. On August 14th, I had coffee with a radio host to talk about his new program for next year. That night I compiled a complete manuscript of the personal stories I had put aside for a long time. I met with my friend Lawrence on August 15th and solved an outstanding domain issue on the spot.

In four encounters, something was pushed out four times. This is no coincidence. Making an appointment to meet is like adding a delivery date to something, and it’s the kind of delivery date that you would be embarrassed to skip.

Now I will deliberately use this trick: If I can’t push a certain line, I will go and have coffee with someone related to this line. Not for discussion, but so that I have something to talk about before that day.

The third type: If you can’t hand it over to people, just hand it over to the machine.

There are some things that no one really waits for, like debugging your own tools. There is another option at this time: change the recipient to a system that will return your mail.

At the end of July, I was giving a class in Kaohsiung. I found that several pictures in the presentation were blown up, so I had to change them on the spot. After I came back, of the 112 commits I pushed on August 4th, 84 were doing the same thing: solidifying that pain into a gate that blocked deployment. Deployment will be blocked if the text overflows the layout, an error will be reported if the exported file is missing content, and a content format verification will be run before submission. On the same day, the verification level actually blocked an English title that exceeded the upper limit of word count.

None of these mechanics are new, and none will be visible to the reader. But they did one thing for me: instead of remembering to check it in the future, it will be rejected now.

On August 13th, I handed over the entire dashboard finishing line that was costing me to my AI team to run. The recipient will change from his future self to a process that will report back, accept, and tell me where I failed. Same thing, completely different feelings.

The recipient does not have to be a human being, but it must not be your future self. In the future, I will be the most unreliable recipient in the world. He never replies to letters and never urges me to write.

The recipient will also let you see the paragraph you missed.

On August 16, I held an “AI-empowered academic research and writing practical workshop (opens in a new tab)”, with a total of twenty students coming from all over the north, central and south. I made a 319-page briefing for this class and revised it several times.

After the course started, I discovered that most students had only used the web version and app version of Claude, and almost no one had installed Claude Code in PowerShell or Terminal. So a lot of time was spent on site assisting with the installation, which cut down on practice time.

After class that day, my first reaction was to revise the presentation, trying to write down the installation process more clearly. After correcting it, I realized: I made a mistake again.

The three hundred and nineteen-page briefing serves students who I imagine are already prepared. As for the twenty people who actually walked into the classroom, what they needed was someone to take care of the period between registration and class. That section is empty in my process. All the effort in preparing lessons is spent on teaching materials, and no one is responsible for the pre-class environmental inspection because it does not belong to any deliverable.

Modifying the presentation is to supplement the teaching materials, not the process. Well, it’s actually two things.

So, my current approach is to send out the environmental checklist a week before class and hope students will respond. To put it simply, that is the recipient mechanism. It will tell me how many people have not installed it yet, instead of waiting for me to find out when I stand in the classroom.

Even subtraction requires a recipient

There’s one more thing that surprised me.

On August 13th, I spent a day making a weekly news radar gadget, which was launched that day. On August 14th, I took it completely off the shelf: deleted the deployment, stopped the automation, and made the source code private, leaving only a backup copy. On the same day, I also removed the full version of a homemade kit from sale and put it on a waiting list.

Deleting something is technically not difficult and it only takes ten minutes to do it. The difficulty is that both of these things require an external explanation, and the explanation has to be faced with people. Someone has to tell you what to do with people who have bought it, and why you should take away the things you said you were going to make.

So my solution for the day was to collect first and then think about it: stop what needs to be stopped first, and make up for the explanatory notes later. If you do it the other way around, and wait until the instructions are clear before you dare to take action, the thing will always hang there and consume you.

By the way, this is why a one-person company with many people will have more and more loading. The reason is simple, because you feel excited when opening new lines, but closing old lines requires facing people; obviously, we are naturally good at the former.

Three things you can do today

After reading this article, if I could only take away one action, I would suggest this:

First, open your to-do list and write a name after each item. Who will this be handed over to? Those items that you can’t name are the items that you will put off until next year. You know it now, you don’t have to wait until the end of the year to find out.

Second, pick out the one you care about the most and find a real recipient for it. Meet a friend for coffee, or send a letter to someone and say I’ll give you a copy on Friday. You don’t have to wait for something to be perfect, you want someone who is waiting for you.

Third, if you really can’t find anyone, just write the acceptance conditions as a machine that will return the items. It could be a checklist, an automated script, or even just a weekly reminder to keep you updated on your progress. The point is not how powerful the automation is, but that it will return your items.

It took me a long time to accept one thing: loneliness is not the most expensive cost when working alone. You will notice loneliness, and you will know that you are sad. The lack of recipients won’t make you feel, it will just quietly make your list grow longer and make you think you are not disciplined enough.

As for the list of 2,498 transactions, I will deal with it this week. This time I included it in the newsletter, so you are the recipients.

A few recent developments

**The free online class “Disassembly and Innovation” is now online. ** I have broken down the innovative cases I have written over the past year into six units and thirty-seven lessons. It is completely free, with no paywall and no need to leave an email. If you want to see it, go here: inno.tw Dismantling Innovation (opens in a new tab).

**August 30th, AI content production system workshop. ** The physical venue in Taipei on Sunday morning is limited to 16 people. The teaching is the same production line that I have used to feed six websites at the same time in the past two weeks. The registration window will close in the next few days. If you are interested, please come early: AI Content Production System Workshop.

**On September 12th, the second phase of the AI-empowered academic research and writing practical workshop. ** The first phase was just completed on August 16th. The second phase is also a three-hour physical event, limited to 20 people, and I will personally lead it. In this issue, I will supplement the pre-class environmental inspection process: AI Empowering Academic Research and Writing Practical Workshop (opens in a new tab).

In addition, I will go to the radio station to record a radio show this Wednesday and talk about Mark. Xue Fo’s “AI changes your customers (opens in a new tab)”, I will tell you after the broadcast time is confirmed.

---

📮 This article is reposted from “Vista Newsletter” No.136. The original text can be found at iamvista.substack.com/p/vista-no136 (opens in a new tab). If you like content like this, welcome to subscribe for free to the Vista Newsletter (opens in a new tab) and receive a helpful letter every week.