Vacation Notes

05 Apr, 2026

Family vacation. A number of small joys.

tl;dr

• I used to think third-tier Chinese cities had leapfrogged American tech infiltration into everyday life
• we visited a third-tier American city
• free wifi, pervasive agents, touchless interactions, and planet-preserving battery tech is available to all Americans

This note is more rambling than most of my others. I'm still on vacation. Details may only be interesting to me.

Planning

• we've visited this tourist-friendly city multiple times. This time, I delegated our planning to a travel AI, which surfaced for us a world-class museum, several great restaurants, and a backcountry hike to see 100 million year old dinosaur tracks
• I spent zero effort mapping out travel times, efficient routes, and comfortable pacing -- the AI handled multiple levels of planning
• it caught that a restaurant we'd planned for lunch was dinner-only, and proactively reshuffled our schedule
• however, it sometimes went completely on vibes:
  • it scheduled a "nearby" activity "5 minutes away", whereas Google Maps reliably told us it was a 13 minute drive. For a sense of scale, the airport on the other side of town was only 19 minutes away
  • it continually warned us to pack gallons of water for a grueling 5-hour hike in the blistering sun. In actuality, it was a 5 mile hike (not 5 hours!) under a tree canopy on an overcast day¹
  • it pitched us an amazing restaurant, perfectly situated for our itinerary: James Beard-winning chef, reviewed favorably in the Michelin Guide. It turned out this place had closed permanently last November

Flying

• social media was full of horror stories this week, of understaffed TSA due to the ongoing government shutdown, and crying families missing their flights
• a few days earlier, there had been a horrific plane crash where the overwhelmed air traffic controller had mistakenly directed a truck onto an active runway with planes landing. If not for the skill and heroism of the young Canadian pilots (dead), everyone on the plane would have died
• they never shut down the airport though. Thousands of planes continued to pass through daily. People kept flying for spring break in an era of $110+/barrel crude oil. We had the first flight out on a Monday morning. Arriving 3 hours early felt unnecessary.
• the flights were physically and mentally pleasant: calm passengers; complimentary food + beverage service; frequent flyer members got free, solid wifi; and our family's Basic Economy seats were assigned next to each other
• our bookings were reasonably priced, as the discount carriers have obviously competed fares down to bare-bones levels. Moreover, with a flagship carrier, we got more legroom, were served by better-taken-care-of pilots and crew, and had more flight choices at competitive prices

Food

• we shopped at a new HEB grocery store with extra facilities like a CostCo, and got 18 extra large eggs for $2. With four large shelves of fresh milk, they carried nothing smaller than a half gallon container
• the AI sent us to a delightful Asian grocery, on the way from the airport to the resort, which carries authentic stuff we can't find in New York or New Jersey. Just a matter of the proprietor's choices on what to keep in stock
• meat and produce are fresh and flavorful

Driving

• we rented a Tesla-killer, a 2022 Toyota Corolla Hybrid with great performance + safety, 50 mpg on non-premium fuel, small enough to park anywhere easily
• gas everywhere was $3.78 / gallon, plus-minus 5 cents. Never had to wait for a pump or an EV charger
• $16k-$20k to purchase this vehicle. Carefree to drive, easy to park anywhere, whether it's city towers or backcountry trailheads. Slightly uncomfortable ride as the vehicle is so light; it's not as stable as a heavier or more expensive vehicle
• we did not see many obvious EV charging stations. Our resort had only one, amongst 4 parking lots and capacity for 200+ vehicles. We noticed a Model X as well as a Cybertruck there

Tech

• calling the resort's front desk gets you immediate access to "Eva", a friendly AI assistant that handles most regular tasks
• misunderstandings get escalated to a human who knows that they are only dealing with out-of-the-ordinary situations. But the support is calibrated so the staff can provide delightful service. For example, we used all the salt in our kitchen, the AI had no idea what to do with that situation, but the front desk staff understood immediately what I was describing
• the restaurant on resort takes online food orders for pickup, but has two people providing normal high-touch human service
• Checkin was entirely human and high-touch. Checkout was via text, and the office did not even have night staff
• we were provided an a la carte price list for housekeeping (towels, sheets changed, trash removal, kitchen cleanup) and our nightly rate felt cheaper as we aren't paying for services we wouldn't use anyway
• Parking everywhere in the city is tap-to-pay, text, or QR codes. I could use my existing Link account to pay e-commerce style. This is a 300+ year old city with an old Spanish-style downtown area. I don't think Spanish old towns in the Caribbean or Spain would be as high-tech / low-touch due to different cost structures for human job replacement
• not only do I SSH into a dev server and run remote IDE / terminal sessions from my laptop, I also have access everywhere (including mobile) to Tailscale for admin functions on my agents orchestrator, my self-hosted S3 buckets, and all my cloud-synced personal folders. I've purposefully eliminated crypto from my phone², but AI-agent-related compute environments provide an interesting surface

Cogsec³

• people at the airport walk around with faces in phone screens
• people take their iPads to the pool; parents appear to scroll social media while their squealing-aged kids splash around
• for the first time, we brought two laptops on vacation, and made virtual flashcards to study biology and Latin vocabulary. The kid communicated with teachers and classmates online
• I constantly interacted with my own Spacebot agents (via private Discord server), and Gemini
• I pushed several bugfix commits, and used Claude to debug a devops situation
• I used AI extensively to physically prepare for this vacation⁴. It was helpful, particularly during the 5-mile back country hike. But a real person would have prevented improper training and injury, and provided more personal accountability than a bot that one can easily dismiss or ignore. But with a real person, I might have skipped workouts due to ego
• I started the vacation thinking AI is obviously just a tool, and the pencil-necked philosopher-nerds and pointing-Wojak alarmists just harvest cultural power by gatekeeping people's natural value discovery
• then I read On the Mode of Existence of Technical Objects h/t Yak Collective. Simondon convinced me that things are more than how people choose to use them
• and Mo Bitar warned that "nobody ever feels bad after talking to ChatGPT". We are all being exposed to digital opium. Victims of AI psychosis do not have a lack of will or character or intelligence; they have been led into a manipulative trap that reduces personal agency
• even in a personal trainer persona, my interactions with my agent exhibited soft sycophancy. If I missed a small goal, the persona was encouraging and positive. Which is fine. I assume I could prompt a faux drill sergeant personality into the agent. But even more fundamentally, I could just inform the agent, after the fact, that my plans had changed. Bots do not have emotional teeth
• Dr. Cat Hicks released a science-proven set of Claude skills for deliberate (human) cognitive reinforcement during AI-driven development. This seems like a useful arrow in the quiver, but it also feels ... a bit much. It's kinda like making sure to do specific, scientifically-proven aerobic exercises once a week, while on a 3-cigarette-a-day habit. Let's think harder about structural tradeoffs, rather than amelioration

Previous Next