Catching Bad Guys with Math

Thoughts from a retired cybersecurity AI leader.

recent posts

about

Biography

  • Yes, Apple Health, I did do a lot of walking in Japan and Korea

    I’m back from a great trip to Japan and Korea. I’ll post photos soon, but in the mean time, I had to share these hilarious screenshots from Apple Health about two weeks into my trip.

    Notifications from Apple Health about a change in steps and flights climbed

    Apple Health showing 8x increase in stair flights climbed and 3x increase in steps taken

    It’s almost hilarious looking at this graphs how much of a step-function change it was between how much walking happened in Japan and Korea, versus back home in Canada. I climbed 8x more flights of stairs and walked nearly 3x more steps while in Tokyo, Osaka and Seoul compared to my normal daily life in Toronto!

    While there were probably multiple reasons for this, I believe the biggest reason was their amazing subways. It’s easier to move around these major cities in Japan and Korea by taking their consistently excellent, efficient and punctual subway systems, rather than try to take a car. I definitely noticed that most of the locals took trains to get around over battling traffic.

    I especially appreciated Japan’s subway UX. Even though it’s incredibly complex, I was never lost or confused taking the trains in Tokyo or Osaka. The consistent use of colors, fonts, labels, and visualizations made it easy after just a couple of rides for my brain to start recognizing the patterns. Similarly, I found Seoul’s subway system easy to navigate, with lines, platforms and exits all clearly and consistently labeled.

  • Back to School!

    Back to School!

     

    I’ve decided that I don’t want to stop learning in retirement, so I’ve assigned myself an hour each day, five days a week to go to “school”. Here “school” can be anything as long as it’s some sort of learning for me; it could be watching a documentary, taking an online course, practicing guitar, etc. Continuous learning and adding some structure to my days is, I believe, important to a successful retirement!

    For my first school term, I’ve planned out a 7-week course to learn just enough Japanese and Korean to hopefully better survive and enjoy my upcoming trip to Tokyo, Osaka and Seoul this spring.

    My plan?

    First, I’m using Pimsleur language tapes (well, they’re audio files now and not tapes, but I can’t help myself!) on most days because I like their focus on conversational speaking instead of grammar, reading, and writing. Seven weeks is a very short amount of time, so it’s going to be important to just start speaking right away, and these Pimsleur courses do exactly that. I’m going to try to listen to them while taking a walk outside — but only when the weather is reasonable! It is so cold in Canada nowadays! — so I can get in some vitamin D as well. There’s one exception to the speech-only focus: I learned that the Korean alphabet is actually designed to be learned in a single day! So I’ve added a day to the curriculum to learn Hangul!

    Second, I’m (of course) going to use AI to monitor my progress, keep me on track, and also generate quizzes and flashcards that are customized and matched to my curriculum. Interestingly, AI can also have conversations with me, listen to my pronunciation and give me feedback. It’s remarkably good at this, and that I can customize the scenarios to practice conversations in is super powerful.

    I’ll report back on my progress and share what I learn about making this retirement school routine stick, as well as my language progress!

  • Mandarin Movie Finder: Coding with AI

    Mandarin Movie Finder: Coding with AI

    My dad likes to watch movies that have a Mandarin soundtrack, but Amazon Prime Videos has no “search by language” facility. Instead, you have to search on some other term (like “Jackie Chan”), manually click on each video to pull up the detail page, and see if what’s on Prime Video has a mandarin audio track. It’s super slow and painful.
     
    So yesterday I spent 3 hours doing my first AI-assisted major coding project. I wrote a “Mandarin Movie Finder” script in Python using Claude Code that would log into Prime Video, search on a term I enter, and then go through all the hits and add any unwatched Mandarin language video to my dad’s watchlist.
     
    It worked great, but it took a long time because as good as AI coding is supposed to be, it really required me to point it in the right direction multiple times. It’s particularly bad at debugging (and many, many times kept going down ratholes that were obvious to me) and building functions (AI doesn’t need to reuse code; it can just copy/paste the same 7 lines over and over again, so factoring out code just isn’t natural for it). I’m still not sure if it was faster than if I just coded it myself manually, but I have to say it was a lot of fun. It felt like a video game, where I’m trying to teach a junior who is super eager and super fast how to write a program. It was an interesting experiment for me.