Purpose
An index for my memory
A menu of topics for my next conversation with you
A faster way to share what I’m excited about without the barrier of writing a complete blog entry on it
A skimmable way to spread content I found valuable
Experiments and experiences
Cruise
I'm in Cruise's tester program, which provides free, driverless rides, but only 9pm to 5am in San Francisco. I finally got a chance to try it out last week. It was a great experience! It arrived in 3 minutes from my pickup location in Bernal Heights. I used the app to unlock the rear doors. No one in the driver's seat! It played some instructions that reminded me of a theme park ride, and then we were off. With a max speed of 25 mph, we avoided getting on the highway, and other cars passed us, but overall the ride was smooth and uneventful. On the way to Bernal, my Uber driver talked to me the whole time. In contrast, I enjoyed the silence of the empty car... except I spent most of the ride video calling with my parents to share the experience. Highly recommend trying it if you take nighttime trips and you're in a supported area!
More antidepressant iteration
Recap: I was taking benztropine to combat the night sweats from duloxetine. The benztropine was making me dizzy during the day, so I switched to glycopyrrolate, which has a similar effect but can't cross the blood-brain barrier. Seems to be working well so far. I stopped taking anything for about two weeks while I was waiting for the new anti-sweat to come in, and had a couple of unpleasant depressive episodes.
Getting home on time
I was late coming home a few times for no good reason, causing me to miss spending time with Euda. To solve the problem, I set an alarm that goes off every day at 4:30pm. Android now lets you set a custom recording as your alarm ringtone, so I joke-screamed "BENNY! GO HOME NOW! NOW, BENNY!!" Every time it goes off, I let it play through a few times for the laughs, and then I go. Works like a charm!
AI dinner x2
I attended two AI dinners this month. One was hosted by a VC firm for ~6 CEOs and me and an OpenAI rep; the other was about 50 "influencers" put on by an AI startup.
At the smaller dinner, one of the CEOs said, "There's so much hype around this AI stuff, but at the end of the day, I realized it's still software like anything else. Still have to do design, user testing, find PMF, etc." He's right! Today, while there is value to be unlocked, it is still unlocked in a fairly traditional manner. We haven't gotten to the point where agents can take complex, multi-step, tool-using actions to save hours instead of minutes. Maybe that'll come next year 😜.
At the bigger dinner, I ran into my 2012 skip-level manager from Google, where we launched Knowledge Graph together. He's still there, but he's stepped down from his VP position to go back to IC work and retool on generative AI. It was a strange moment. In a way, LLMs are the next evolution of Knowledge Graph--a faster way to get to answers without having to click through web links.
I also met my counterpart at Google, who emphasized that companies like VMWare and Slack were category-defining, but ultimately their lunches were eaten by linux and Microsoft respectively. How do small companies like OpenAI, Anthropic, etc not only define a new category, but protect it from the tech giants with huge distribution channel advantages? In OpenAI's case, it seems like "soft acquisition" was the answer; in Deepmind's case, actual acquisition. In Anthropic's case... we'll see.
Staycation
Diana and I took a week off, but instead of going somewhere, we stayed home. My parents graciously continued to take care of baby during the day. We took naps (sometimes multiple!), took care of chores we'd been putting off, read books, went on long walks, and generally recharged. I’d planned to do some writing but decided to rest instead.
Life updates
📣 Launched Claude.ai
🧑💻 Helped host Anthropic's first hackathon
🗞️ Mentioned in NYT
📖 Started reading Chip War
Content
5 point Likert ratings for “I would recommend this content to a friend”, sorted
I don't want to talk about it 4/5
Part memoir, part pop-sci, part therapy log, all on the topic of depression in men. There were a few times reading this that I had to stop because it was too heartbreaking.
Despite feeling like my childhood was excellent compared to many of the stories, made me think critically about my childhood and my role as a father.
Struck by repetition of phrases like "after 3 years of therapy, we finally had a breakthrough." Sometimes the trauma is buried deeply.
I realized about halfway through reading that this is the same Terry Real featured in How not to hate your husband after kids
Barbie 4/5
Surprisingly good! Excellent world building, fairly strong characters, not too heavy-handed on the feminism. Not for little kids.
Favorite part: musical numbers.
Beef S1 4/5
Gave this another shot after a friend suggested it gets less depressing. It stayed pretty depressing, but got a lot more rewarding, eventually hitting strong Parasite themes.
Traces the expression of intergenerational trauma in two protagonist anti-heroes as their web of lies and delusions eventually catch up to them, their family, and their friends.
[Spoiler] Last episode is the best: after crashing their cars into a ravine in the wilderness and trying to kill each other, they eat unidentified berries that cause them to hallucinate and vomit, propelling them into a final release of their egos. They confess to each other their deepest regrets, failures, and dreams before they save themselves the next day. Catharsis!
Joy Ride 3/5
Fairly standard adventure / narrative beats.
Holds up surprisingly well after 34 years! Pretty fun.
Expected the parents to hear the baby's talking or for the baby to play a more substantive role a la The Boss Baby, but never happened. Turns out, it's a comedy drama about a single mom who falls in love with a taxi driver.
Silo E1-3 2/5
Too much action, not enough substance. World building starts strong but doesn't seem to go anywhere.