Posted by justincormack 12/9/2025
WTF? One and zero are not probabilities, and 6 out of 5 is not a rating.
This is where I stopped reading.
> At minimum, ranking algorithms with this much economic consequence should be auditable.
"At minimum". Immediately preceded by a paragraph starting by "For policy", with sentences like "If discovery now shapes small-business survival, then competition, fairness, and urban regeneration can no longer ignore platform ranking systems" or "tools of local economic policy".
That's perhaps not an outright call for regulation, but it's certainly suggesting it's warranted.
That's not what the author was suggesting (or indeed, what they built). They were trying to untangle the positive feedback bias showing up first in the rankings gives.
I think there's probably a lot more to untangle, but as a first pass it's super cool!
> the most important result isn’t which neighbourhood tops the rankings - it’s the realisation that platforms now quietly structure survival in everyday urban markets.
For any service like this, _of course_ ranking is at the core of it. A more honest article could have started there, eg "since you can't display all results, and doing so is useless to everyone, the heart of these products is their ranking algorithm and choices. Let's examine Google's."
Don't hate the player, hate the game. I hate the game too, fwiw.
Including, of course, the way many popular chain restaurants got there is they make food a lot of people like.
A1 steak house.
AAA1 steak house.
00AAA000 steak house.
I think this is very likely false if you mean compared to the status quo ante. Before Maps, a well-loved but hard-to-find venue just wouldn't ever be seen by most people, and the absence of reviews made branding more important because it was all you had to go on. I'd be very doubtful if the proportion of independent cafes and restaurants decreases when Google Maps enters an area. (Couldn't find any causal research designs though....)
The more general point that the algorithm is not neutral (and probably never could be) must be right.
(I asked ChatGPT but it ended up with: "We have almost no clean exogenous variation in Maps rankings or feature rollouts at fine geographic scales that would let you estimate impacts on entry, survival, or market structure in a neat DiD/IV way.")