It's imo scarier we're not finding giga-brained bugs yet, boundary pushing helps bound your "unknown unknowns", human-tier attackers, and weaker-LLM script kiddie attacks.

Still just finding "haha oopsies!" bugs means you're safe from neither the fruits of incompetence nor the beasts in the forest.

In the immortal words of the reply to "I could've done that myself!",

"...but you didn't, did you?"

Paragraphs that you realize afterward are "first or last 20 pages of a sci-fi novel to establish setting mechanics" voice but it's real life

"nah babe, Claude wrote it! you see, we live in an era where robots are able to write most of our code quite effectively, all I did was explain the functionality I wanted and Claude successfully built it! It's a bit like having a secretary who builds my lego rocketships."

"wait no come back-"

Freudian cornflake

Everything reminds me of her…

I think you’ll see mostly stuff like the NYC congestion pricing,

Other taxes would prob be charged on refuels at fast charging stations, on self-driving hardware, at purchase time on cars that support it, on car tabs as a “self driving fee”, and baked into additional payments on owner’s insurance.

I also think (vibes based, not an auto mechanic) depending on how hardware costs come down and difficulty, if it does work, there will be a market for aftermarket-by-car-dealers “add the sensors and hardware to add self driving to your current car”, and this would get significant uptake.

I agree we’d see slow process though on the transition, range is long and there will be significant overhang.

We did about 15% hybrid/electrification in the last every 5 years with rising speed, and I expect self driving to catch on much faster, so eg <=14 years vs <= 35 years seems plausible to me

May more fully unpack that at another time but tldr “it works, it’s agi, it’s kinda just a smarter dude in your computer or robot, it (mostly) doesn’t wanna kill you more than its user, & working with it’s a patchwork PITA as 1:1 slapping robots or agents into business process has MAJOR frictions.”

Wearing my prophetic hat a lot today; my default model isn’t quite “nothing ever happens” but “you tend to get realities that displease everyone interested in a specific one”

I suspect this midterm period paradoxically looks like “boring agi singularity”.

So I kinda think “if it can be done, demand will be overwhelming and supply will quickly figure it out, because if you’re selling a “iphone adoption curve” product, you and your competitors can compress many years of learning into “GET SOMETHING OUT NOW”

I think stuff like the above will be RAVED about online ala “how tf did I live without it” with massive downmarket uptake as people now basically have a fetching servant, at cost of fuel/electricity

I expect auto cars to massively drive car electrification, as this lets cars fuel at destinations.

I think this includes

- summon via app
- last mile semi-intelligent self park (eg w/ parking space markers)
- order groceries online from a “cloud kitchen”-esque store, a picker loads your car, it drives home,
- integration with food apps and fast food (store brings food out, car drives home)

My mental model is “around 2032-2040 top car models start supporting self driving for semi-arbitrary independent legs.

I think the net traffic increase still happens because it seems like “as long as it’s net profitable to field more cars, this will happen till breakeven is reached on these rideshares”, but I’d probably downweight empty leg ridership more than I did

I think you’re like 70% right tbh

Pure vibes, I kinda think we get a similar summoning legs problem for 25-40% of their driving, but kinda do think they’ll be much more popular than pure “summon your own car”, w/ the threat of people switching to that keeping prices attractive.

Ambivalent on this. Think you’re right on sufficiently short leg transit, but mentally thinking about “no traffic-has plugs and wifi Amtrak” and tbh I still hate the extra 2-3h delay enough to fly even for $100-200 more, same destinations.

Not sure when exactly speed matters but it do.

Tradeoff spectrum of “I think this will be a massive improvement on human thriving”

vs “traffic is going to get so bad holy shit what we have now is baby mode, it’s going to be the black hole nuclear urban problems of the 30s just wait bro holy shit”

Many of these would already have been taken, in both cases, but many wouldn’t, as riders would’ve made other arrangements

eg “I couldn’t leave work to do the ferrying thing”, “I wouldn’t have drove to the festival if I had to park”, I would’ve bussed back if I couldn’t double transit my car.”

(This could change if cars start to get seen as wholly unnecessary to own in hyper-abundant availability, but I expect demand for sharing one’s own car with ungrouped unknown randos to be low so we won’t usually be able to compress summoning legs as an efficiency)

I don’t think any of these individually add to much, but in sum, the ability to summon one’s own car means many more summoning legs, and possibly more than human legs

Each one is ~an empty car that will take up road

-“Doubled transit”, eg lets say I like bussing or biking to work in the mornings but don’t like it at night. I bike to work, but order the car to my office while I bike, so it can drive me home.

Consider that the car still needs to drive both legs it would’ve before, but now there’s a cyclist too.

- “Ferrying”, eg I’ve got the car at work, but I want to send my kid to soccer practice or karate or whatever.

so car leaves my office, goes to school, picks them up, ferries them to karate, then returns to me at the office for 2-to-1 no human legs.

- rideshare approaching TRUE marginal breakeven of perfect competition (if I’m uber or an auto-driver driver all, I spin up as many auto-ubers as I legally can, and I DGAF as long as each one’s marginally profitable post-risk assessment and depreciation), most have no human inside.

I see it as “many legs look like”

- interpersonal car sharing: car commutes me, comes back, commutes wife, repeat in reverse for 4 daily no human legs.

- “active idling” or “auto-swarms” during eg sports events w/ cars driving loops once local parking is exhausted, or new MASSIVE pickup swarms.

Buried lede of “transit will be better” is “The traffic bomb of self-driving is coming, and people are NOT ready for roads with twice as many cars, and more than half with no drivers.”

So there’s going to be MASSIVE pushes in most metros to consolidate in this space for shared company commutes, likely first trotted out as a corporate incentive but later more broadly.

Further this is likely to rise in importance as cheap scaled auto-drivers work.

If marginal cost to get a ride go down bigly (eg “call my car like a taxi”), congestion will get MASSIVELY worse, too.

Many won’t care if a car takes 2h in rush hour to reach them if they didn’t need to sit through it