A Curious Aspect

With the rise of TikTok and similar off-the-cuff services people have become more accustomed to watching videos with a portrait orientation. This has long struck me as odd as it removes a viewer's ability to explore the periphery to better understand a situation or context. However, given the incredible popularity of the format, there is clearly something that people find attractive. This is clearly something that makes sense to explore if I plan on doing anything with online video long term.

A couple of weeks ago I thought that it would make sense to start with videos of Ayumi doing silly puppy-dog things. I went through all of my videos of her, of which there are hundreds, and found none of them that would work in a portrait orientation. Just about every video had been recorded in the more traditional landscape orientation with a 16:9 aspect ratio. Converting this to a 9:16 simply did not work as it would rarely allow more than half of Ayumi's face to appear at any given time. New videos would need to be captured.

At the start of March, the puppy and I got in the car and drove out to Izumi on the Pacific coast of Chiba Prefecture. There we walked along a beach well known for all the driftwood that accumulates there and recorded about 25 minutes of video using a handheld camera as well as a drone using both 16:9 and 9:16 aspect ratios. A YouTube Short was published within a few hours of returning home and, to the surprise of nobody, it received almost zero views.

Given I am a complete unknown to just about everyone on Earth, this comes as no surprise. This video was also the first "short" and pretty much consists of me walking around on a beach, because Ayumi was moving in and out of focus with most of the footage that was collected. Sure, people watch these videos on their phone, but fuzzy images are no fun.

A week later I went out for an evening drive after a rather frustrating day at work. Rain was softly falling and it offered an opportunity to try and play with reflections and light. The pocket camera was mounted to its suction arm and moved around the interior of the car as I drove to Funabashi and back. Again, a video was uploaded to YouTube within an hour or so of getting home. However, this one was watched by four times as many people as the first video. Clearly something worked. But what?

From what I can tell, there are only two real differences between the videos:

  • one focuses on a person while another focuses on an action
  • one is labelled mostly with English while the other contains Japanese hashtags

Perhaps the problem is the language?

Today I posted a third Short. This one involved me driving home at night with the top down on the car. The descriptions were all in Japanese without a hint of English. Interestingly enough, the view count was more than double the previous video after just three hours. Are my videos being presented to people outside of Japan? Sure, I'm geographically inside the country, but the use of English was supposed to make the videos a little more international. According to the various metrics that YouTube shares, though, over 90% of all visitors for every video are in Japan, with over 80% of them being male. So, if I want to have videos that attract eyeballs, the English should be an afterthought.

Mind you, the point of creating a YouTube channel was not to attract a huge audience, but to learn new skills while also sharing the beauty of Japan with anyone who might watch. That said, investing hours into video recording and editing only to receive a few dozen views is disheartening. Having a bit of a balance would be nice.

So, with this in mind, I think the plan for the near future will be to continue making shorts while also focussing on some of the longer-form videos. They will be made primarily for a Japanese audience, which means I'll need to improve my pronunciation a bit if there will be any speaking, and subtitles will be made available for anyone who doesn't know the language. Let's see what patterns emerge over the next few months.

Playing with ChatGPT

Large Language Models – or LLMs – have been in the news for well over a year now as people explore various ways these sorts of tools can be used in any number of situations. Like many people, I've dabbled with some of these as well. The locally-run models are interesting and terribly slow while the public services are rapid and most certainly not free1.

Last year I was curious to see whether this technology could be used at the day job in order to provide a more complete description of student development in the written sections of progress reports. This is generally an area where specificity is useful, but the time needed to write a detailed analysis of someone's growth is often greater than the amount of time that teachers have available. The idea that I wanted to test was this: given we already know a person's attendance, participation, past progress scores, and material coverage, can a 400-word summary be provided to the teacher ahead of time? This would allow them to read through the feedback and make edits where appropriate, while ensuring the students have realistic, evidence-based feedback to help them with future studies.

After a couple of days worth of effort, it was clear that there were two LLMs capable of delivering the core of what I was looking for. One even had the added benefit of being able to translate across multiple languages, making it a clear winner. I delivered my findings to senior management at the day job who agreed that it was an interesting area of study, but they decided to not pursue any sort of automated feedback system just yet. The one or two vague sentences from the bulk of our over-worked teachers was clearly sufficient in their eyes, so I set my investigations of LLMs aside given that none of my personal computers really had the power to use them in a realistic fashion2.

However, as LLMs, image generators, and video creation engines become more powerful, this isn't something that makes sense to completely ignore. There are a number of areas where these sorts of tools would be incredibly useful, particularly the Large Language Models which can be trained on the writings of an individual.

Jordan Peterson has mentioned on a few occasions how people he works with are doing things like putting the entirety of Christian texts into LLMs in order to ask the machine questions. Peterson Academy will do something similar, putting entire courses into an LLM so that students can "ask the course" a question. Dr. Peterson has even had all of his writings put into an LLM so that he could essentially ask himself a question. His glowing reviews of this budding technology has made me incredibly curious.

Could a Large Language Model be used to generate course material that is customised to an individual student? Could it take into account their strengths and weaknesses? Could it recommend further study on matters the student may require to successfully complete a course? This would be an incredibly interesting area of study, but it's a bit beyond my abilities at the moment. Before getting into something like this, it might be better to start testing LLMs with personalised data.

Dabbling with ChatGPT

Looking at the hardware requirements to run a Large Language Model system locally, I would need to invest about $3,500 at a minimum into a workstation. Several avenues were explored to reduce costs, including starting with a used system that could be upgraded in sections. Unfortunately, the cost always worked out to somewhere in the vicinity of $3,500 to $4,000 in the first year, not including the cost of electricity and network bandwidth. This is several thousand dollars more than I am willing to invest at this time, so a cheaper option was required.

Fortunately, OpenAI has a $20/month offering for a fully-featured ChatGPT account. Last night I created an account and started playing around by asking the machine questions, hoping it would provide answers that I already knew3. In each case the response was correct and, in one instance, the answer included information that I had never heard before. Within five minutes I discovered two other sources that corroborated the new information and was impressed because ChatGPT taught me something new while I was testing its ability to understand my words.

Next I wanted to see how well the DALL-E image generator worked. The first request was "Show me a paper airplane flying over the surface of Mars with the Curiosity rover in the distance". This is what I received back:

Paper Airplane on Mars

I then uploaded a picture of Ayumi and asked "Show me a picture of Ayumi playing a game of Go against a cat". This is how DALL-E responded:

A Shiba Inu playing Go with a Cat

Very interesting.

But could it write something the same way I might? Would it know to use an excessive number of commas, ellipses, and relative clauses?

Today I tested this out by uploading my entire public blogging archive dating from 2006 to today, as well as every public social post I have ever written to Twitter, App.Net, and Nice.social. This amounts to almost 4,000 blog posts and over a quarter-million social posts. Replies and any post shorter than 100 characters was excluded, as they would lack any context to work with.

To test, I kept the request vague: "Can you write a blog post using my writing style?"

What came back was interesting:

Embracing the Unfamiliar: My Journey with Ayumi and the Art of Coding

In the ever-evolving landscape of life, I find myself continually drawn to the unfamiliar, seeking growth at the crossroads of personal adventure and technological exploration. This journey, accompanied by my faithful Shiba Inu, Ayumi, mirrors the unpredictable nature of coding - a realm where each challenge is a step towards mastering the art.

Ayumi, with her spirited demeanor, teaches me the value of curiosity and resilience. Our travels across Japan, from the bustling streets of Tokyo to the serene landscapes of Kyoto, are more than mere physical movements; they are a metaphor for navigating the complexities of coding interfaces, databases, and APIs. Each destination, much like every coding challenge, presents its own set of puzzles, demanding adaptability, patience, and an eagerness to learn.

The philosophy of coding, much like life itself, is an endless loop of learning and unlearning. It's about embracing the bugs and errors as stepping stones, not stumbling blocks. Ayumi's unwavering spirit in the face of new territories and her joy in discovery remind me of the satisfaction found in solving a particularly stubborn bug or deploying a feature that perfectly serves its purpose.

Our adventures are more than just tales of a man and his dog; they are a narrative of the symbiosis between technology and human experience. As Ayumi explores new landscapes with fervor, I delve into the intricacies of coding, each of us on our own quest for understanding, yet together on a journey of mutual growth.

In this dance with the unknown, Ayumi and I find harmony. The challenges of the road and code, rather than daunting, become sources of inspiration. They remind us that at the heart of every adventure, whether it be through the forests of Japan or the virtual forests of code, lies the opportunity for discovery, learning, and, ultimately, a deeper connection with the world around us.

Screenshot of Request

Lots of commas. Relative clauses galore. Even a pair of semi-colons. The only thing missing was an ellipsis in the first or second paragraph. The intertwining of disparate ideas in a semi-unresolved manner also seems to capture some of my longer-form blog posts from years past where I tried – and failed – to hone the craft of writing thoughtful articles exploring a philosophy that I could not yet lay down in words.

Absolutely remarkable.

Bringing this whole thing back to the idea of using LLMs with personalised education, using a tool with this sort of perspicacity would make it possible to construct personalities for one or more "teachers" who could interact as realistic avatars with people studying a topic. While this would not eliminate the isolation a person feels when learning something on their own, it could drastically reduce the need to find secondary or tertiary sources for answers. An LLM with an entire library of materials on astrophysics, for example, could answer questions from students in the persona of the teacher leading the course or – if such things were permitted – through the voice and mannerisms of Stephen Hawking, Michio Kaku, Neil deGrasse Tyson, or Edna Krabappel. This could drastically improve "engagement" in a learning environment; something that is often a challenge with students who are easily distracted or bored.

At the moment, no artificial teacher can replace a real human who is invested in teaching people a subject they are passionate about. We still have a long way to go before technology can completely replace the benefits of learning from physically present educators. However, a well-trained digital assistant could go a long way to reducing the workload of a teacher or professor, answering the simpler questions or repeating the same points for students who need the repetition. This would allow digital systems to be where skills and knowledge are acquired so that classrooms can be dedicated to the application of the skills and knowledge.

A pie in the sky dream? Perhaps. But we are a lot closer to this being a reality than a lot of people might realise.


  1. Yes, there are free offerings out there, but these services generally offer the previous version of their software or an incredibly limited subset of functions. When exploring a tool with as much press coverage and potential as an LLM, it's generally better to invest some resources in order to build a better understanding of what potential may exist.

  2. I do have a decent computer … I think. It's just not decent enough to run LLMs with any sort of velocity. A single question can take upwards of 30 ~ 45 minutes to answer. Maybe one day I'll invest in a dedicated machine for this purpose … but not for a long while.

  3. Do not blindly believe anything found on the Internet.

Our First Anniversary

One year ago today I took a detour on my way home to stop by a pet shop in Adachi to see a young puppy with white fur. Later that afternoon, after a bunch of paperwork, she came home with me.

Ayumi and I – Last Year and This Year

Looking back at the early photos and videos, it's hard to believe that Ayumi used to be so small. We've been together for almost every hour of every day, unless I've been at church or out to a store, which makes it a little difficult to see her grow with time. Fortunately, cameras are incredibly ubiquitous now and we can very easily capture a moment to enjoy it at some point in the future. Will she continue to grow larger this year? I wonder ….

To celebrate this lovely occasion, Ayumi and I hopped in the car and drove to Noda to walk around the giant Shimizu Park that dominates the west side of the city. We crossed bridges, followed a river, walked through a bamboo forest, and met a bunch of people. After returning to the car, we drove over to Tsukuba to explore another park that is mostly known for its soccer fields and a very, very large grassy hill. Along the way we enjoyed a bit of food and we even shared some yogurt, which is one of her favourite treats.

As per usual, while we were on the road people would see us in our topless yellow car and smile. A number of people took pictures of Ayumi as she looked over the passenger-side window. Pedestrians on the sidewalk waved. A man on a scooter pulled up alongside us at a red light to say "hello".

Ayumi Hangs Out

So much has happened over this past year as a result of Ayumi and I being together that it sometimes seems unreal. How can so much take place in such a short period of time? A better question might be what does the future have in store?

At the tail end of 2022 I had made a list of things that I needed to do in order to position myself better for whatever might happen in the months and years to come. Ayumi was not on the list, but she provided a great deal of incentive to get to work on those objectives while also demanding some new ones. 2023 was remarkably productive, interesting, educational, and rewarding as a result. The first six weeks of 2024 has followed the trend with the added benefit of meeting new people just about everywhere we go.

Ayumi's name consists of two kanji: 歩 + 光. Progress and light. May our future together be long, interesting, and joyous.

The Missing Piece of the Puzzle

Car audio systems that come pre-installed from the factory are rarely anything to write home about. The stereo often has a UI that is confusing, the speakers are often entry level, and the road noise washes out most of the fine nuance in the music once the vehicle moves faster than 40km/h. Because of this, I opted to not have any audio system installed in my car when it was delivered as my plan was to install very specific components to achieve the sound that I was after.

In October, less than a week after taking delivery of the yellow Copen, I got to work installing everything my initial research said was needed:

  • a Joyn bluetooth stereo kit
  • a Pioneer TS-C1630SII speaker kit, consisting of two woofers and two tweeters
  • a Pioneer TS-WX140DA powered subwoofer
  • and enough sound damping material to silence a toddler

Audio gear awaiting installation in the Copen Cero

The installation itself took a little over two hours and has worked pretty well since it was first powered on. However, when driving at highway speeds, music still tended to wash out. Podcasts were even harder to enjoy if one of the participants mumbled or had a thick accent. Suffice it to say, the initial work was an incomplete solution. Something else was needed to improve the audio quality.

Over the next few months I toyed around with different methods to improve the sound quality within the car. Additional padding was installed. Vibrations were chased down and eliminated. The trunk panels were bulked up with damping sheets. At one point some thick blankets were kept behind the seats to help muffle any sound that would come from that direction1. Some of the work resulted in improvements. Some of it did not. What I really wanted to do was find a way to build an active noise-cancelling system for inside the car. Unfortunately, this proved to be cost prohibitive2.

Last week I decided to approach the problem from a different angle.

The tweeters are located in the pillars that hold the windshield. The woofers are located in the door. The subwoofer is behind the passenger seat. This means that the main sound is coming from the front of the car while the lower-frequency elements are coming from behind. In some of the nicer cars I've been fortunate enough to ride in, the key to "luxury" is having a surround sound system badged with a Dolby logo, a Bose logo, or both. Given there is nothing inherently wrong with the Pioneer speakers, and the Joyn stereo is really the perfect accessory for the car, it made sense to look into transforming the five-speaker system into a seven-speaker solution.

One of the many challenges of installing parts in a two-seater convertible car is space; there is simply not very much of it for third-party additions. I've seen some people on X add speakers behind the seats in order to improve sound, but this would not really solve the problem that I had with sound clarity. I've even seen a creative person replace the rear window platform with a custom-speaker base. Both of these solutions made use of standard car woofers and could be done without spending too much money. However, I wanted something that would offer improved clarity while also being useful when the top is down. What I needed was a pair of unobstructed satellite speakers.

With a plan in mind, I started looking into which models might work for my car. Again, because the Copen is a rather compact convertible, mounting options are incredibly limited. Had I bought a car with a fixed roof, there would be dozens of potential solutions. I needed something compact that would also match the stylings of the car. Fortunately, Pioneer has the perfect solution: the TS-STX510-B satellite speaker kit.

A quick search online showed that Amazon had these on sale for 29% off list price. A few clicks later and a pair was scheduled for next day delivery.

The original plan was to find a way to mount these speakers to the seatbelt guide. This idea was scrapped when the sleek items arrived, though, as they were simply too heavy and awkward to place there. I would need something a little more sturdy … like the space between the roll bars.

In most convertibles there is a wind deflector between the roll bars. This is said to accomplish two goals:

  1. Air from inside the car – such as the A/C – would hit the deflector and stay in the vehicle
  2. Air from outside the car would not be able to enter over the trunk as a result of wind channel patterns

I do not know if either of these are actually accomplished by the 5cm tall piece of clear plastic that sits between the driver and passenger roll bars, but it almost doesn't matter. The speakers have to go somewhere, and between the seats strikes me as the most logical.

So that's where they went.

Pioneer Satellite Speakers

And they're perfect.

This evening I went for a little trip around Aichi prefecture to test the new sound setup and came away incredibly impressed. Not only can I hear many more details in the music that I listen to, but the road noise has also diminished quite a bit. This is most likely due to the proximity of a bare speaker to my ears, but it's a nice little bonus. Sirens from ambulances and police can still be heard, as can the horns from impatient or frustrated drivers, but the endless droning of tires on pavement has been masked reasonably well. Hopefully they're just as effective when listening to spoken-word podcasts.

Later this year, after the temperature warms up a bit, I hope to install a bit more sound damping in the car to reduce a bit more noise3, but I really feel that the audio setup is just about perfect as it is.


  1. These would also come in handy in the event the puppy dog and I needed to keep warm for some reason … like waiting to watch the first sunrise of the year from the windy shores of Shizuoka Prefecture.

  2. I was not willing to spend over $800 on more electronics that would muffle exterior sounds with the help of microphones that required holes to be drilled through the body of the car in various places. I know "it's just a car", but I'm not yet ready to start poking permanent holes into the body.

  3. I am surprised by how much styrofoam is in the car. Sure, it's light and doesn't mind getting wet, but it's a horrible material for sound absorption or dissipation.

Smiles Everywhere

Over the past few months I have found it a little uncanny how often people will look over at my car and smile. This might happen if Ayumi and I are out and about, as she tends to get attention everywhere she goes. It might happen if it's just me in the car with the top down despite the season. However, the pattern is always the same: people see the bright yellow car, they look inside, and they smile.

Ayumi with a paw hanging out the passenger-side window

Given the overall reaction that people have, I'm beginning to wonder why more people do not have fun cars and happy puppies.

Last month it seemed like Ayumi and I were approached by at least one person everywhere we went who wanted a photo of the puppy hanging out the window. It happened at Bentenjima, Lake Yamanaka, Oarai Sun Beach, Nokogiriyama, and Hakone1. During the workweek, we would sometimes head to Teganuma Park at lunchtime. People would see Ayumi poking her head out of the car and remark at how cute she is or wave to catch her attention. Phones would be pointed. Pictures would be taken and, presumably, shared. It's wonderful to see.

Ayumi poking her head over the door in Hakone, with Mount Fuji in the background

One of the reasons I chose my car was because it looks happy, and it's hard for me to not smile when walking up to the vehicle from the front. It's headlights and big grille make it look a bit like a cartoon character; forever in a good mood and ready for silly adventures. While most people invest their money into practical cars that can get them from Point A to B in a safe and reliable manner, it's a shame that so many vehicles on the road are dull and utilitarian. Not everyone can get by with a two-passenger convertible, but car companies could do a little more to make their vehicles a bit more interesting for the average person. Offering vibrant colours would be an easy place to start.

Maybe if more people could smile – and were smiled at – while in their car, we'd see less road rage and fewer frowns throughout the day.


  1. Ayumi and I had a very busy month, it seems.

A Letter from Daihatsu

Late last year, right before Christmas, Daihatsu Japan released a news statement reporting that some of the safety certifications for 64 vehicle models and three engines were fraudulent. The issue has apparently gone unnoticed for years and, as a result, cars may not be as safe in a collision as the public was lead to believe. In response to this news, the management of Daihatsu has shut down production of any vehicles identified as having false certifications and processes will be put into place to ensure that such things cannot happen again going forward.

Daihatsu's Statement and Apology

This response is pretty boilerplate for any Japanese company found guilty – or even legitimately accused – of wrongdoing.

Japanese social media was buzzing with speculation about what sorts of problems might exist with the tens of millions of Daihatsu vehicles that navigate the roads every day, but I noticed an interesting pattern with many of the people I follow, who all own Daihatsu's plucky little Copen: "We're concerned, but won't stop driving the car."

In my case, I'm not concerned at all and most certainly will not stop driving my car. Heck, I don't even want any sort of refund or special service as a result of this situation. I knew that buying a car as small as a Daihatsu that it was not going to protect me from serious collisions. I've seen the aftermath of an accident involving a Daihatsu at busy intersections over the years. To say it's "not good" would be an understatement. This doesn't mean the vehicle is unsafe or is unable to protect an occupant from most types of crash, but it does mean I should pay just a little more attention to where other vehicles are while I'm effectively "queued up for a T-bone"1, which I should be doing anyway.

When people ask me why I'm so casual about this given the very real possibility of a new driver, young family, or elderly person being in a serious accident, I point to the overall track record of vehicle collisions since 2010:

Fatalities in Car Crashes in Japan - 2010 to 2022

Between 2010 and 2022, the number has almost halved. This is not because people are all clamouring to buy larger vehicles or because people drive more carefully today than they did before the advent of smart phones, but because vehicles – even those with falsified safety test results – are better able to protect its occupants today than ever before. Yes, every death is a tragedy, but not every death is preventable.

When some scoff at this, I follow up with a question:

Is there any company with more than a dozen people anywhere in this country that is 100% perfect in everything that it does? If so, they are lying to you.

We all know how situations like this come to pass. Senior management tells their people to accomplish a certain goal with very tight constraints. Middle management tells their people to make those goals a reality while having an egregious number of meetings with each other to make it look like they're busy. The people who report to middle management do what they can with the resources they have.

And then comes the actual problem: someone in the chain decides to muck things up a little bit. This can happen a number of ways. Perhaps someone is simply lazy and doesn't want to do their work. Perhaps someone is overworked and simply does not have the time to complete everything they're expected to do. Perhaps someone does not have the necessary training. Perhaps someone wishes to create havoc simply because they can.

Or maybe the issue is in management? A middle manager wants to get ahead of their peers by completing their work first. Or, more accurately, demanding their people to complete the work first. Or the manager wants to do something under budget, so does not talk to HR about bringing in one or two more people to help an already over-extended team. Or the manager is a walking example of the Peter Principle and is clueless about what's actually required and how to communicate it down the chain.

Fact of the matter is, anything that is created by humans is flawed in at least one area. The more complex something is, the greater the risk of flaws. I accept this risk and get out of bed each day not knowing whether my apartment will collapse in an earthquake, whether my car will be in the parking lot when I go outside, whether my shoe laces will remain tied, or whether my coffee will be hot by the time I take the first sip. However, based on historical patterns, I can assume that my apartment will withstand a common earthquake, my car will not be stolen, my shoes will remain on my feet, and my coffee will be hot for at least one sip.

I appreciate that Daihatsu is going to do better going forward, as any organisation that takes pride in its work should. However, I am not going to be angry or upset that such a thing was possible or that it affected the very car that I received just a few months ago and am still paying for. To demand perfection from everyone else when I cannot offer it myself is foolish, and life is too short to waste it on being upset over something like this.


  1. Stopped with a fully-exposed side of the car, making a situation where an oncoming vehicle can slam into the driver or passenger door head-on more likely.

Not Worried at All

My mother asked me a question during our most recent phone call that tech bloggers and pundits have discussed for well over a decade: Is it a good thing to teach kids how to code in school?

On the surface, the most obvious answer to this question is a simple "Yes", but I wanted to know why my mum asked this question, given that she doesn't even own a computer and sees them as conveniences, but not essential.

If more people can write code, then why wouldn't a company just replace programmers after a couple of years to keep costs low?

Ah … the IBM Way™ of running a business.

The question came about as I was describing the philosophy that I adhere to when writing code and how it can create friction with management teams who expect absolute obedience from their people. So, with this in mind, the question could probably be phrased like:

Aren't you worried that you will be replaced with a younger person who knows the basics of coding and won't question authority?

Short answer to this question is "No" and I would encourage everyone to learn even a little bit about how to make computers work for them rather than the other way around. There are professional data analysts who earn six figures a year just because they discovered the power of VLOOKUP in Excel and launched themselves into a new career as a result. There are musicians who learned a dozen commands in Python to create a custom pattern synthesiser which resulted in a long career. There are kids all over the world with beat up computers who discover that a little bit of PHP can be the start of a career writing plugins for many of the most popular website engines that are currently in use. With just a little bit of knowledge, computers transform from "magical black boxes" to "tools that enable work to be done". If someone learns a little bit and makes a little money, they are free to decide whether they take the plunge to build their skillset or move on to other things.

An example I gave to my mother went along the lines of "How is learning to code any different from learning a language?" Is anyone threatened when there's another human who can speak (or sign) a language? Most of us can communicate with our mouths, hands, and even eyes. If a billion people were to learn Esperanto, would that be a net negative for anybody who is currently fluent in that tongue? At the very least, it would open up a new market for art and culture.

Going further, we all have our first languages. Some people use it better than others, and those who put it to great use earn the greatest benefits.

So … no. I am not worried at all when schools teach kids how to write code. I am not worried at all if management is upset when I refuse to do something the way they want because it conflicts with a philosophy I adhere to. I am not worried at all if I am replaced, as it's bound to happen eventually. What I am worried about is people not having the opportunity to learn a skill that may help them solve a problem they'll face at some point in life.

January Goal: Five-Minute Shot Lists

Believe it or not, the average amount of footage collected for each YouTube video that I have published to date consumes about 215GB of data. Some of the longer trips have exceeded 500GB while some of the shorter videos barely used 10% of that maximum. Of course, it goes without saying that the longer a trip is, the more footage there will be simply from the fact that a camera affixed to the windshield will be recording everything it sees while the car is in motion. However, it has become increasingly clear over the past couple of months that having cameras record things willy nilly is no way to make videos that people want to see. There needs to be a little more thought put into the effort.

Generally when I am exploring the map in search of destinations, I make use of satellite imagery and street-views to plan stops. Parking lots will be marked, photo locations will be tagged, and basic ideas will be considered. If there are not too many people and the location is far enough away from buildings and airports, I'll make some basic plans for the drone. If the weather is forecasted to be nice, I'll find places where the 360˚ camera might be mounted to the car for situational shots. Will the walking path be interesting? That's where a gimbal will be used to reduce any camera shake that might spoil an otherwise unrepeatable moment1. The rough plan then sits in my head until we start reaching locations.

This has been semi-workable but, unfortunately, a number of shots tend to be completely forgotten about until after the opportunity has passed. This results in videos consisting mostly of dashboard footage, which is primarily used to convey a sense of distance and space to anyone watching on YouTube2. Something better is required.

While investigating various ways to improve the stuff I share on YouTube, I started watching videos from people who work with video professionally. Each one of them says the same thing: make a shot list.

While I have been doing this mentally, they strongly suggest putting it down on paper and writing out a sequence. This allows a person to look over the shots to ensure a consistent flow. If there is a story being told, doing this will also allow a person to consider whether a particular scene should be shot one way or another … or both ways with the final decision being made in post-production. These are perfectly sensible suggestions. So, with all this said, one of my goals for this month will be to make shot lists for all YouTube videos going forward.

There are some considerations to take into account, though. The lists need to:

  • aim for at least 5 minutes of video
  • dashcam footage cannot exceed 1/5th the length of the video
  • there must be "action shots" of Ayumi doing her thing at each destination
  • if the drone can be used, there should be at least one shot that could not be done with another camera (not just something recorded from "on high")
  • at least two cameras must be used

This list will likely get longer as I continue to hammer down the format of the episodes, but it's a good place to start. There have been a number of times where I have kicked myself during the post-production process because there was a particular shot that I had made a mental note for but completely missed while at the site, usually due to some sort of distraction. With a written down shot list, there will be less risk of missing an opportunity to record something interesting. If I can maintain this for the month of January, there's a high probability that I will continue to make lists for all future productions, improving the quality of the videos that are shared online.

At some point, I will need to also make a plan for how to get better shots while using extension poles, as there's a lot of interesting things that can be done when you have a camera on a stick.


  1. It's hard enough asking a person to do something twice. Good luck asking a puppy to repeat an action!

  2. A lot of the dashcam footage is sped up and sliced to make it more interesting, but there's only so much of it that can be used.

Domino Effect

People have tried to make a lot of analogies for life over the years and so many of them fail to capture the entirety of the human experience. Despite what Tom Hanks might suggest, life is not like a box of chocolates. Nor is life a dance. Nor is it briefer than a ballad sung by legendary rocker Meatloaf. Life, if it can be compared to anything, is a giant domino effect.

This line of thinking ran through my mind today as I wondered what would have happened had I not moved to an apartment north of Tokyo.

  • I would not have joined the church I attend, which means …
  • I would probably not have met some remarkable people, which means …
  • I would probably not have joined various Bible study groups, which means …
  • I would probably not have had the opportunity to develop meaningful friendships with the people who I look forward to seeing several times each week.

The argument could be made that I could have met other people and joined other groups had I attended a different church, and that's certainly true. It would be a different domino chain, though. Because, had I not joined the church I attend, then …

  • I would not have met Ayumi, as she was at a pet shop in Adachi, between the church and my apartment. Which means …
  • I would not have learned that she dislikes public transit. Which means …
  • I would probably not have invested the time and money at a driving school. Which means …
  • I would probably not have earned a Japanese driver's license. Which means …
  • I would almost certainly not have a bright yellow Daihatsu Copen.

Another argument could be made that I would have likely chosen to get a car if I had a different dog, as there are no vets around my house nor is it particularly easy to get around with public transit in that part of the country. This is a possibility, but it's not likely that I would have stumbled across the plethora of videos on YouTube featuring people driving their Daihatsu convertibles with a puppy in the passenger seat. Before settling on the Copen, I was seriously considering a Honda Fit or a Daihatsu Mira e:S as they are cheaper and easy to find at a dealership.

Of course, if I had not picked up a Copen, then I could not have …

Again, some might say that I could have brought Ayumi – or a different dog – to different events in a different car, and that I could have started a YouTube channel irregardless of the vehicle … but there are a whole lot of IFs required to make it all fit together. Had any one of the dominoes that resulted in a decision above not fallen, then this year would look different.

Heck, had I taken my sister's advice and returned to Canada for a couple of months from February, then pretty much nothing that happened this year would have come to pass as it would not have made sense to move into an apartment so close to flying half a world away.

Life sure is interesting.

Not Yet

Nozomi passed away almost a year ago. Every so often I find it hard to believe how quickly the time has gone, but the feeling is always the same. She deserved a longer life and a better ending than what she received. I can only pray that all dogs truly go to heaven and she is at peace now. Every so often people will see my phone and remark that the dog they see is not Ayumi. I let them know that her name was Nozomi and that she was a very special friend.

Nozomi on the Phone

The picture on the lock screen was taken on August 2nd, 2014 and it remains one of my most cherished photos. It was taken during a particularly difficult period, but it always brings a smile to my face. Nozomi's bright eyes and ever-present smile communicated everything I needed to hear. It still does. The photo on the main pages was taken six years later, on November 22nd, 2020. Nozomi was ten years old by this point and was quietly dealing with a medical condition that would not be diagnosed until four months later, when it almost killed her. For so much of her life, she was incredibly patient. She rarely put up a fuss and she looked forward to every opportunity to spend time with the people around her.

Her photos remain on the phone because I haven't yet been able to give her up to the past. Yes, Ayumi is also a special puppy with a special place in my heart, but she's not a replacement for Nozomi. Close friends cannot be replaced, nor should they be. But every so often I wonder if I should change the lock screen image for the first time since 2014, or the main image, which has been static since the day it was taken three years ago. Is it time to move on?

The answer is obvious, but the execution is difficult. Part of me knows that changing these images, which captured just a moment of tranquility with a good friend, would be akin to leaving much of the last 16 years in the past. Perhaps that's the thing I should be doing, but it's not the thing I want to be doing.