🌊 Stream

A place for short, informal posts and scattered thoughts—my own personal feed.

Why I Still Publish My Blog in Three Languages in the Age of AI

A few days ago, I went to Tokyo Python: Around the World in 80 Dags, and AI translation came up in a conversation afterward. AI really can translate quite well now. I use it to help translate my own posts, too, but there's still a real gap. With many works, even a small detail that feels off can spoil the whole experience. AI translations often have exactly that little something that gives them away.

In the Japanese anime I love watching, the voice actors are part of what gives a show its soul. They're often the reason I watch something in the first place. I don't think AI can match their performances, now or in the future.

I should come clean, though. My time and energy are limited, so my translations currently only get basic proofreading. That's why I include an AI notice. If you read English or Japanese and find something that sounds odd, please email me or leave a comment to let me know. Criticism is welcome, too.

The talk also mentioned that Apache Airflow has built-in support for multiple languages and welcomes contributors in different languages. I recommend Lee Wei's article, “Airflow 多語系化 - Make a sunrise that I know I’ll never see.” He's the main contributor driving the Taiwanese Mandarin translation and a core Airflow maintainer.

As for why I translate my own blog, I have three reasons.

Making it easier to read

I want readers of Chinese, English, and Japanese to be able to open a post and start reading, without having to install a translation tool themselves.

Keeping an eye on translation quality

I don't know what readers would use to translate a post. It might be Google Translate or an AI tool, and even then, different AI models give different results. From using AI for translation myself, I know it can translate proper nouns and idioms too literally and miss the mark. I usually don't accept the first version it produces.

I also have my own preferences. In the Japanese versions, I use the more formal, polite desu/masu style for Life. For Review and Stream, I want the feel of a friend sharing something, so I use plain form.

Some works or linked resources may not have English or Japanese versions. Translating the posts gives me a chance to catch that and adjust the relevant version, perhaps by adding a note or removing…

Read More & Comment →

Assuming Good Intent

I was listening to 🇹🇼 Overbuild, Episode 98 recently when this idea came up: always assume good intent.

Denny from Overbuild said that at the large multinational company he used to work for, people in different roles often saw different “facts.” Both sides could have a reasonable point of view and still be at odds. Starting from the assumption that the other person means well, then trying to understand where they're coming from and why, makes it easier to have a productive conversation.

A few days ago, I saw 🇹🇼 From Me Daily's November 13, 2026 post about a client he'd gone out of his way to help who ended up buying the home he'd shown them through someone else. It bothered him at first, but these days he tries to consider that the client might not have had much choice. He added in a comment, “Over time, you realize you're no better off for not wishing them well.”

Some people might think that's naive. But assuming someone means well doesn't mean resigning myself to getting the short end of the stick. When something rubs me the wrong way, or when I've already lost out, I can try looking at it from another angle.

It's exhausting to stay upset about everything. I figure most people are pretty busy, and very few have the time to sit around thinking about how to hurt someone else.

Some things are hard to see in a generous light in the moment. I'm still practicing. Here's a story about running into a childhood friend in college and having some money stolen.

I'd just moved when I ran into an elementary school classmate downstairs. A lot of us lived close to one another as kids, so we'd sometimes drop by each other's homes after school. I went to his place a lot because he always seemed to have the latest toys and games. I used to play with his younger brother and sister, too.

I'd moved around a lot since middle school. Taipei isn't that big, but I was still happy to run into an old friend in a different neighborhood. We picked up right where we'd left off, caught each other up on the past few years, and went out to eat together pretty often for a while.

One day he came upstairs to hang out at my place. After he left, I realized the…

Read More & Comment →

The Freedom to Leave Things Unused

My room has a 34-inch curved monitor and a perfectly serviceable desk and chair.

Sometimes I want to work from a café or somewhere else, but I feel as though I'm letting all that equipment go to waste.

Lately, though, I haven't been able to sit at my desk for very long. Yesterday, I moved to the sofa with my computer and found myself surprisingly focused. It made me feel like I'd been held hostage by my own stuff.

Buying something doesn't mean I have to use it all the time to get my money's worth. I can use it when it suits me. It's not as though I've abandoned it entirely; I'm just not using it constantly.

What matters is having the choice. I can buy a standing desk and barely ever adjust it, knowing I can when I want to. I can buy a big monitor and still work somewhere else, then come back and plug it in when I need several windows open at once. If I never use something, I can let it go when the time is right. But if there are situations where it's really useful, even if they don't come up often, I don't need to worry so much about how little I use it.

Maybe the same goes for other things. Doing something doesn't automatically earn me a gold star. There's freedom in not doing things, too. Maybe choosing what not to do matters even more right now. Life is finite; I don't have to do everything.

AI agent subscriptions give me the same feeling. When I haven't used up my five-hour allowance or this week's quota, it always feels a little wasteful. But using up credits just for the sake of it is its own kind of constraint.

Running out when I need them is annoying, but that's how it goes. My workload isn't going to line up perfectly with my token allowance.

One way to manage it might be to write prompts for things that aren't urgent, save them, and send them to the AI when I have quota left over. But that could just leave me with an ever-growing backlog. Maybe I should get things done as they come to mind and not worry so much about credits.

Or running out could be a chance to test my skills and understanding without AI. If I have no idea how…

Read More & Comment →

Picking Up a Pen Again

Around the end of August, I started writing by hand again for the first time in years.

What prompted it was a Substack post by leafwind sharing JR Lee Radio's video, “The Truth Behind Gen Z's Reversal in IQ: Why Is Educational Technology Hurting Our Children?”. In the original post, leafwind described its main argument as “Handwriting involves a complex environment with lots of signals, and learning is more effective in complex environments.”

When I first started, I kept thinking, “This is so inefficient.” Getting my first laptop in university had amazed me with how much faster typing was than handwriting. Suddenly, I could record so much more. From then on, I was a devoted believer in typing, moving from one note-taking app to the next.

But lately, I've realized that being able to settle down, finish the task in front of me, and deepen my understanding matters more than efficiency.

At first, sitting with a pen and a blank sheet of paper, my phone and computer put aside, I could finally get a clearer sense of what I actually knew. Without my notes app, Google Search, or AI, I discovered how little I could write on my own. It was like flipping through a textbook and thinking I understood everything, only to close it and find I could barely explain any of it.

Whether it's an AI response, Google search results, or my notes, having that information within reach makes it easy to mistake it for knowledge I already possess. But much of the time, I'm just hoarding information without really understanding it, let alone using it.

I'd had a vague sense of this problem for a while, but I hadn't wanted to face it. Stepping away from all that readily available information and admitting how little of it I actually understand is painful.

A while ago, I asked two very experienced software architects how they use AI. It turned out they just write prompts and ask Claude Code or Codex to do things. They don't use a bunch of clever tricks, but they can quickly steer the AI in the right direction.

Using the same tools, I often need several rounds of back-and-forth and multiple chats just to finish a first version. It leaves me feeling that my judgment and grasp of the subject still have a long way to go.

Chasing efficiency is dangerous. Speed matters, but…

Read More & Comment →

A Note of Thanks

On October 23, I got a notification from Boba: someone had bought me a bubble tea.

I'm not sure why I hadn't received any notifications before, because when I opened the dashboard, I discovered that several cups had already piled up!

I think the first two were from friends, but the others came from people I don't know at all.

Boba lets supporters mark their messages as private, so I won't share what you wrote. But I want you to know that I received your support and read your messages. Even though I don't know who you are, thank you from the bottom of my heart. It makes me happy to know you enjoy my website.

I hadn't realized how happy it would make me to have someone read my writing and the things I record here, then decide to support me financially. This is the first time my writing has brought me such tangible support.

Of course, comments, email replies, and all the other ways you respond make me really happy, too!

After publishing some recent Stream posts, I noticed people leaving comments almost right away. That's how I found out there really are people quietly following my website. It made me pretty happy!

I'm glad I set up the Stream section. It gives me somewhere to share short pieces more casually, though some keep growing as I write and end up becoming longer articles.

So far, the gaps between posts have been 2-2-4-7-3 days. It's surprisingly easy to let those gaps grow without noticing. I need to keep an eye on that!

As for the website's yearly running costs:

  1. Domain registration through Cloudflare: US$28.20
  2. Hyvor comments: €50

I've had the domain for three years, and I only added Hyvor comments this May. The site doesn't come close to paying for itself, but it's some of the best money I've ever spent.

One more time: thank you!

I'll keep documenting, improving, and creating!

Go To Comments →

How Much Advice Do I Need?

When I'm feeling lost about where my life is going, I research, read reviews, watch interviews, or reach out to people for coffee chats.

That might mean browsing social media like X and Threads, forums like Reddit, YouTube, or blogs, or getting an introduction through a friend.

But when someone offers advice, I've started asking myself: “What's their background, and does their advice fit my situation?”

We all come from different backgrounds, have different resources, and want different things. If someone's circumstances are very different from mine and I overlook that, their advice might only leave me more confused.

That's one reason I've started spending less time on social media. There's too much advice, it's all over the place, and it all gets flattened into the same feed. Anyone can jump in with an opinion, but not everyone's experience applies to me.

Someone who's lived in Japan for years and plans to stay, someone who's only here to experience life before returning to Taiwan, and someone living in Taiwan who knows Japan mainly through the internet can all stop by and weigh in. As I scroll past, though, all I see are posts and comments, without the context behind them.

Plenty of people share their success stories without giving enough of that context, too.

I've seen someone who went to Australia on a working holiday and later obtained a work visa still sharing their 2023 experience in 2026, even holding online seminars. But Australia's immigration policies have changed quite a bit in those years. Having gone to Australia myself, I don't think there's much in their approach that I can use.

I used to see what someone else had done without considering what they had going for them at the time. Then, when I did the same thing and it didn't go well, I'd start doubting myself, mistaking their success for evidence that I was falling short.

Of course, different circumstances don't mean there's nothing to learn from each other. And taking action doesn't guarantee a return, either. Most people build their skills and resources over time. I just need to remember that our situations and goals may differ, then work out which experiences I can learn from, which methods need adapting, and which conditions I can't replicate right now.

When I've asked people I really look up to for advice, they usually haven't rushed to give me an answer…

Read More & Comment →

Knowledge Moves at Different Speeds

I borrowed this title from a 📹 talk by Kuo-Wei Cheng, the founder of 🇹🇼 PanSci, in which he shares his journey in science communication.

What follows is my summary of this part of the talk, not a word-for-word quotation:

When choosing topics, his team considers not only the practical question of whether a subject will attract attention, but also what speed of knowledge it belongs to.

Fast knowledge is things like news. If you don't cover it today, other stories will bury it tomorrow.
Medium-speed knowledge is things like technology trends. They change quickly, but have a long history behind them. There may be news every day, but if you only chase the headlines, you'll see the waves and miss the currents beneath them. AI, semiconductors, space, energy, and biomedicine all fall into this category.
Slow knowledge doesn't rush to capture today's traffic or keep up with tomorrow's conversation. It takes time to unpack a question and look at its history, methods, controversies, and human nature.

For fast knowledge, PanSci makes three-minute science news videos. For medium-speed knowledge, it makes videos of 10–15 minutes, aiming to give newcomers a simple map of the subject. For slow knowledge, it makes long videos in its Slow Science series. This kind of knowledge is too hard to compress: you need to understand how scientists ask questions, how they collect samples, what mistakes they have made, and how they correct them. Even then, there may be no clear answer, only an understanding we can provisionally trust.

He also shared his approach to scientific literacy: “When things move fast, be able to judge; when things move slowly, be able to stay with them.” We're used to speeding everything up now: watching at double speed, reading quick summaries, mastering whatever subject in no time. But some things can't be mastered that quickly. When you think you've mastered something too fast, what you've usually acquired is a very complete misunderstanding.

In the age of AI, perhaps knowing how quickly or slowly to approach different kinds of knowledge is an essential skill.

Listening to Kuo-Wei talk about slow knowledge really drove home for me that understanding a large software project is a form of slow knowledge. Contributing to Apache Kafka as an open-source project has been all about dealing with its history, methods, controversies, and human nature. The project is already 15 years old, and many systems that process…

Read More & Comment →

Good Old Days in the share house

I still live in the share house I wrote about in Save Point Update.

Everyone I mentioned in that post has since moved out. The French guy went home, and the American spent a while in Kyoto before going home, too. The German guy later moved to a share house in Saitama.

A few days ago, the German guy came back with someone from Indonesia who used to live here.

His working holiday visa had run out, and he hadn't managed to switch to a work visa. Before leaving Japan, he wanted to see the place he'd called home for six months once more.

He still missed this place after moving to Saitama and had come back a few times. He said it really was dirtier and older here, but this was where he'd built a routine he loved: working out in the run-down gym, going out to box, taking walks in the afternoon, then coming back to the common room at night to watch the French guy and the other residents make a racket.

From what the German and Indonesian former residents told me, the house was at its liveliest around last winter. By the time I moved in that April, everyone was getting ready to go their separate ways. At least I caught the last of its glory days.

There was also a French woman doing a carpentry internship, a handsome older guy from Spain who worked remotely, and a British DJ who taught English.

The French woman's English wasn't very good, and she didn't speak Japanese, but she was always excited to tell everyone about her day, what she'd bought, or where she'd been. The Spanish guy once made me authentic croquetas. Before he left, he even gave me some of the small furniture he'd bought.

The British DJ is still in Tokyo, and I'm going to see him again this week. We once talked about small talk. He said that when he asked how someone was doing, he genuinely wanted to know; he wasn't just waiting to hear “All good.” Writing this reminds me that he and the Spanish guy took me to Torikizoku for my birthday this year.

Around the time they all left, I was about to start at my previous company. The share house became very quiet. For a while, a group of Indian residents would gather together, but after the person at…

Read More & Comment →

Build me in public

I've been updating my personal website more often lately. I want to gradually put more of myself on it: the different sides of me, the things I think about. It feels a little like I'm building myself in public.

Or maybe it's more like building in semi-public. A personal website sits somewhere between social media and a notes app. It doesn't feel quite so exposed, but it isn't entirely private either.

After all, far fewer people go out of their way to visit someone's website than scroll through social media these days.

People come with different expectations, too. On social media, you can say whatever's on your mind or browse lighter posts. There are more challenging topics now and then, but without an emotional hook, the algorithm rarely gives them much attention. Someone who comes to my website, though, is here to read something I made.

I haven't felt like opening Substack much lately, and I've switched its in-app notifications to email. Substack still has likes, replies, and shares: all those immediate signals that people are responding. I've noticed that they affect me, and sometimes I envy writers who get more engagement. That's not what I want to chase right now. I want to focus more on the quality of my writing, especially since I don't think I'm writing enough yet.

One thing about a personal website is that feedback comes slowly. I rarely know who's reading. I've added emoji reactions and comments, but not many people use them yet. With my recent Stream posts, I can't even tell whether anyone has seen them. For now, I imagine most readers are other writers or friends, though I do occasionally get an email or hear from someone.

Wiwi and Lee Wei are part of why I've been updating more often. Since pulling back from social media, I've spent more time in my RSS reader. I still catch myself wanting to scroll through something. Both of them post pretty frequently, so their writing keeps showing up. Seeing them use their own websites to document their lives and thoughts made me want to do the same.

So I finally made a Stream section. I can post here without feeling like every piece has to be as complete as a Life article.

Writing in a notes app and publishing something feel very different. Knowing someone might read it makes me take more care in…

Read More & Comment →

Putting in the Reps to Ease My Anxiety

Lately, when anxiety hits, I tell myself to do something first.

I've tracked my time on my calendar and with a Pomodoro timer for years.

Looking back at the numbers recently was humbling: I hadn't put in nearly as much time as I thought. Yet I'd spent plenty of it worrying and telling myself I wasn't good enough.

So I've decided to put in more hours before judging whether I'm good enough.

That brought to mind an experiment I read about in Atomic Habits. Jerry Uelsmann, a photography professor at the University of Florida, divided his students into two groups. One would be graded on how many photos they submitted; the other would submit just one photo and be graded on its quality.

The students graded on quantity ended up making the better photos. They had more chances to experiment, make mistakes, and improve. The students graded on quality spent so much time studying theory and chasing the perfect shot that they barely experimented, and their results fell short.

Of course, not everything can be measured, and more hours don't guarantee better results, especially with AI in the picture. Still, I need to do enough work to see what's working and change course when it isn't. The same goes for AI: to learn how to use it well, I have to actually use it.

I came across Manny's take on Google Lab's Vibe Coding game platform:

An LLM is like a seed of intelligence. Put it at the entrance to a new field, and it can open the door for you. Sure, you can open Claude Code or Codex yourself and do just about anything. But if you never start, it's all talk.

This week, I gave myself a simple target: "At least 40 hours on my Pomodoro timer this week, so I can see how much time I actually spent focused." That's eight hours a day, five days a week. If I can't manage eight hours each weekday, I'll have to give up some of my weekend to make up the time.

When I start worrying about everything, I check my Pomodoro timer and calendar: how much work have I actually put in this week? If I haven't even hit my time target, I can keep going instead of getting lost in my thoughts. If I'm worried I won't find a job, I look at how much time I've…

Read More & Comment →

The White Bear Effect

In English, it's called the White Bear Effect. It can also be called Ironic Process Theory (the theory of ironic rebound).

Proposed by psychologist Daniel Wegner, it describes how trying to suppress a thought can make it even more vivid. In the experiment, participants were told not to think of a white bear, but the image kept appearing in their minds.

The idea is that when our minds try to suppress a thought, they also have to spend some effort checking whether we're thinking about it.

This might also explain why willpower fails. When I frame a rule around what I must not do, it takes twice the mental effort, and I still keep thinking about it.

If I tell myself, "Don't eat snacks," "Don't scroll social media," or "Don't watch another episode," my mind fills with snacks, likes and comments, and the shows and anime I want to watch. Willpower might win a few times, but I can't keep it up forever.

A better approach is to find an alternative, accept where I am, and use more positive phrasing.

Instead of "Don't scroll social media," I can ask, "What articles or books do I want to read?" I don't have to fight the urge to scroll; I can give myself something else to do.

That's also why I started this Stream section. I can post the scattered thoughts I'd usually put on social media here first, then share them there later if I feel like it. Knowing that different people will read them here helps me stop chasing the dopamine hit of likes and shares and focus more naturally on the quality of what I write.

I feel like this applies to a lot of things, especially the part about phrasing things positively. The way I put things into words really does shape how I think over time. I'll share more when something else comes to mind!

Go To Comments →

I Can't Have It All

I've always been afraid of making choices. I'm afraid of putting too much into one field, hesitant to go all in on anything, and worried about the opportunity cost of choosing wrong.

But I'm not afraid anymore. I've realized that I need to take the path in front of me seriously before I can gradually open up other options, while continuing to create or explore other possibilities and giving them more of my time and attention.

I can't want a salary above the market rate without putting in more effort than average.

I can't say I want to create things if I spend every evening after work seeing friends or going to events.

Of course, some people can have it all. But that's because of their circumstances. Maybe they worked incredibly hard earlier, so things look easier for them now. Maybe they're simply better at what they do. Or maybe they have the resources and drive to do many things at once.

In the end, I have to come back to my own circumstances, see what's possible for me, and make my choices within those limits.

Given where I am now, I need to take care of my main job first. If I want to create, I'll have to make time for it out of the hours I would otherwise spend resting, socializing, or entertaining myself. If I'm unwilling to pay that price, I shouldn't imagine I'll get the result.

Go To Comments →