For many thieves who like stealing stuff (just a little bit tho), it’s not fun to be stopped by a lock.

After some thought: 10 years feels very fast, I’d think 20-30 is more realistic, I’ll see more what people estimate in the literature.

There are existing analysis in the same range, e.g. buildingcongress.com/report/2025-...

I've also taken the 50% usage over 10-year assumption from their report.

A common complaint is "this is too expensive at $5B"/

I'd frame this as: To be worth it, just due to the new housing, one unit of housing needs to be worth >$25,000

Over coming posts, I'll provide more data on whether that's likely true.

Ah that would've come in useful when I was arguing with someone about decongestion pricing in NYC and they said something like "We're just shifting congestion from roads to transit".

People do believe the opposite of this theory sometimes!

And you might ask, why does this matter?

These estimates often end up being very wrong. Lots of cases of highway expansions not meeting the expected ridership demand.

And there's a reason stuff doesn't get checked: No on earns money from being correct on public projects, ten years in the future.

In both cases, I've gone through huge documents that didn't break down the estimates into simple base assumptions. So the public (e.g. data-oriented readers like me) cannot validate them in a reasonable way.

In public discussion, such numbers are treated as facts:
"115,000 daily riders"
"City of Yes to create 80,000 new units"

But they really are super-uncertain estimates. And estimates are only good if the assumptions are good.

Yes that’s the goal, aggregate by anything and it shows the map: Borough, Neighborhood, Zip, block.

Thanks for the feedback!

Oh I’ve got it at the lot level and as an interactive app, just making sure I find intuitive ways of releasing this. Coming soon!

This one is actually a recommendation from a Building Code (Sydney, Australia).

Very nice. If I ever get the time or AI gets good enough, I can repeat this with building outlines.

And also agreed that we should only count rectangular buildings, because otherwise I can cheat with a narrow entrance for a triangular building.

And the fully technical answer is that this search isn't conclusive, because I'm searching over *lots* rather than *buildings*.

Searching by building would be a lot more work. But if you find a more narrow building, please let us know!

This is 8.4 feet, but the lot and building are very much not rectangular.

maps.app.goo.gl/c6UhBbAvDsHw...

Here's the #2, in Glendale, at 9.6

maps.app.goo.gl/q3HnTWFXDehW...

But let's get serious:

This one looks like it might win, right? At 9.3!!!

(But the lot isn't fully rectangular, which is cheating)

maps.app.goo.gl/P8KqYVpS9DB8...

There are marvels like this 10.3ft townhouse in Brooklyn

t.co/IcbzWhilZe

So what's the verdict? 75 1/2 Bedford St still holds strong.

9.5: 75 1/2 BEDFORD STREET, MN
9.6 (so close!) : 77-03 75th St

Odd contenders:
9.3: 176 Brighton Beach
9.3: 71-60 72 PLACE
8.4: 2625 Stillwell