The rest is just about not removing all the streets. You also need enough roads for emergency vehicles. But simple two-lane roads for arteries and one-lane roads everywhere else are enough for deliveries, maintenance and (European-sized) emergency vehicles. All the other lanes are just there so everyone can travel downtown in their personal car instead of using more efficient transport options.
You need space for pedestrians anyways, and with traffic restricted to the necessary minimum you don't need bollards everywhere
This isn't some kind of rocket science that has never been tried before, it's the reality in many cities around the world. You need an entire concept instead of adding half-backed solutions to an existing system, but the solutions are all well-tried and for the most part very straight forward.
Streets may be one way because it discourages unnecessary automotive traffic and it allows the rest of the space to be dedicated to bicyclist or use for residential and commercial purposes (outdoor seating, space for games, etc).
So even if the streets are one way, there will often be space for the ambulance to maneuver around blocked vehicles. And if there's not, the ambulance can take another street or road. If it's the only route that can access that location, then with any sane amount of urban planning then the ambulance will be within a short walking distance to the location (and more often than not within eyesight).
I used to work at a warehouse, the whole street of warehouses had a street width of at most 10m (30ft) wall-to-wall. People vastly overestimate how much throughput is needed purely for shipping, due to all the passenger cars on the road.
Heavy equipment is also less of an issue than you'd think, because it tends to expand to fit the space it's permitted - like all engineering projects.
American city centers are overprovisioned on roads by probably a factor of 10 or more. Ironically, fixing that would be excellent for local businesses, and they’re usually the ones opposing it because they think there’s no way to get customers besides having them drive directly to the business and park on their property.
So how many 40 lane roads are you going to build instead of single train lines?
Look at any car-free area and there are very reasonable exceptions to make them function. The important thing is to remove cars as the mandatory way to access locations, and to cease prioritizing cars over more efficient methods of transit. Forcing cars into our cities took a massive amount of changes to our legal and planning system, and huge huge subsidies. Just going back to thinking of cars as one of many ways to get around, rather than as the primary and prioritized method of getting around, would allow for much more efficient systems.
New York City has more than adequate public transportation, except for the private vehicles slowing it down.
SF has adequate public transit, but without all the cars in the way and with higher ridership it would be amazing.
And it wasn't about "the will of the Outer Sunset" [1], and some (local) people there rely on car or rideshare access, plus all the rest of us from everywhere that isn't the Outer Sunset also want to be able to drive there to visit. The existing post-Covid time-shared road/park setup on Highway 1 was already incredibly friendly to bikes, pedestrian, pets, children, surfers. And SF Measure K only passed 54.7:45.3% [2].
> "increasing maintenance costs due to erosion are another huge reason it’s a good idea)"
You could argue the exact same that SF cablecars could be cut because they net lose money (federal subsidy), or about all of us constantly paying for the upkeep of Highway 1 down at Big Sur, which is regularly closed/subsided/storm-damaged, and costs $$$ to repair every winter, but the wishes of the (small population of) locals have to be balanced against the wishes of everyone else (CA + out-of-state residents, tourists) who use it.
In any case re the viability of transit at such low neighborhood densities, Seamless Bay Area [3] have for years been pointing out the ongoing insanity, duplication, waste, lack of ticket interoperability etc. of bankrolling 27 different agencies for 9 counties in a metro area of only <8m people [4], smaller than NYC/LA/Chicago. These operating costs are soaking up any passenger revenues, preventing it from being spent on services. SF was particularly disproportionately hard hit by the permanent Covid change of both WFH/remotework + exodus of tech employees and companies + record-high SF downtown office vacancy.
[0]: https://www.kqed.org/news/12014573/sf-muni-dire-need-funding...
[1]: "Do voters want to close S.F.’s Great Highway? New Chronicle poll shows where they stand [Prop K - 46% in favor of closure, 44% against, 10% undecided]" 10/2024 https://www.reddit.com/r/sanfrancisco/comments/1gaaj36/do_vo...
[2]: 2024 SF Measure K https://sfelections.org/results/20241105w/index.html
[3]: seamlessbayarea.org
[4]: "One journey, one ticket", key learnings from Swiss public transport" https://news.ycombinator.com/item?id=38686115
Or close that to regular car traffic but keep SF Muni buses and trams and maybe high-occupancy taxis/ shuttles/ rideshares.
You also need public transit options into it as well as transportation solutions for “out of towners” who descend onto downtown for some event (eg public parking lots outside the city that has public transit to the event).
It can be an easier and more politically popular idea to just use streets to solve this problem at the individual level even if less efficient rather than requiring efficient central management planning by the government.
And just to be clear, I’m a fan of the changes SF is making, just highlighting some of the challenges, not to mention that SF isn’t going quite so far as to rip up existing streets with the exception of the latest vote where the entire city basically voted against the will of the outer sunset to convert the highway to a park and not just closed on the weekends as it’s been since COVID (increasing maintenance costs due to erosion are another huge reason it’s a good idea).