Random Walking into a podcast?
An experiment that my dad finds more compelling than my writing
Reading vs. listening
Writing to share a head’s up on an experiment I’m spinning up this week. The target audience - you don’t like to read, but you’re interested in checking out my content.
No judgement! I’m here (with the help of AI) to help. Though, if you’re subscribing you probably do read. Maybe this is more for your friends who don’t like to read?
Anyway … personally I like to read and look at the occasional cartoon. That’s just me - except when I’m driving, in which case I prefer audiobooks (and not crashing)1. I keep hearing that people love podcasts. It’s gotten to the extent that there are now shows about people making podcasts. Enough that I think everyone will assume I’m referring to a different one. To go with that flow I’m trying an experiment. You guessed it …. podcasts!
But Rich - didn’t you say you were on a podcast once about 13 years ago and you still wake up sweating at the horrific realization that you sound “like that” in real life?2 Yes - this is actually true. A shocking realization that I don’t have the late night DJ voice that my brain induces as a self delusion to keep me sane when I have to speak to other people. Apparently I sound like … not the late night DJ (at all).
Podcast?
What if I could have all the fame and fortune of a podcast without leaving my office and without doing 99% of the work? Well - thanks to Google (and I assume copious appropriation of the internet’s voice actor’s future livelihood ) you can!
Probably not going to achieve the fame and fortune part. Hopefully it will be valuable to those not interested enough to read a long rambling essay - while having a lower bar for their 20 minute drive to the gym? I guess we’ll see…
But how - and isn’t that a terrible, terrible idea?
During a break from my occasionally unhealthy relationship with Claude Code3 I was listening to an online class from Google/Kraggle. Materials included an AI presentation overview of each day’s assigned reading. That reminded me that I’d read Google’s NotebookLLM4 product can product podcast form outputs from a collection of research material.
Rather than go back to my on-again off-again relationship with Claude I thought “maybe I should just goof off.” But I guess I must have already done Wordle so instead I ended up investigating if I could do a podcast from one of the articles I’d written which I wished had gotten a bigger audience. A few prompts later I was listening with my mouth hanging open. This seems to happen when I try new tooling these days - I probably should invest in some tape to keep it closed at key moments.
At first one would think an AI generated podcast is just the sort of slop that’s ruining the Internet. But somehow this treatment creates some special magic that makes this material really come alive. Even if it filters out my snark and some salty language. Or more likely because of that. Perhaps an AI editor, or any editor just does me good. FWIW my dad called me late at night when I sent a sample to him, saying “gee - that’s a lot better than your original article!” Truthfully - I told him I didn’t think he was wrong.
It’s also possible that just hearing a model say nice things about my work makes it seem better than the result really is. But since I’d already done 85% of the work I figured I’d just turn it into an experiment. I am truly interested in what folks think as materials are published. I’ll stay close to the andon cord just in case.
The future
I’m going to pick roughly an article a week that I feel is closer to “practical, useful” than “self indulgent drivel” and try creating a deep dive podcast on it
I’ll start with Google’s tooling but I’ll probably experiment over time. Likely releasing both a longer form (around an hour) and a shorter form (~15 minutes) each week. Let me know what you think and please share it with friends and co-workers if you think it’s helpful.
Maybe, someday I’ll extend this to me talking with some of the amazing people I’ve met through my career5. But I’ve still got a good amount of therapy ahead to work through that whole hearing myself speak thing first. In the meantime I hope this experiment will land for some folks.
Looking at my Audible usage I actually probably listen more than I read too.
If you share my future podcasts and it’s like a huge runaway success I may considering divulging my earlier, one-off podcast appearance. I did manage to locate it, though it seems the link (probably thankfully) is broken.
I mean seriously, I told you like 100 times that you adding a project breaks the Xcode build so why do you keep doing it?!?!
If you haven’t tried Notebook LLM for some reason it’s a pretty simple way to wall off a discussion with a model (mostly) constrained to materials you cite for it. I hadn’t used it that much before but I’m inclined to do so more now. Though I did notice it’s not completely immune to stretching beyond what’s there in the source material set at times.
Especially if I can reconnect with a former coworker who famously explained a situation to me as, “people would be surprised how much of Amazon runs but by the grace of God.”