How Google Maps allocates survival across London's restaurants
laurenleek.substack.com398 points by justincormack 5 days ago
398 points by justincormack 5 days ago
It's always annoyed me that zooming in on a building will not reliably show the business that operates there. I understand that at low zoom levels you may need to filter what is displayed based on the high density, but when I zoom in I want to see everything that is there. Sometimes I am forced to go to street view to read the sign, then type the company name into the search box to force the business marker to show up and get clickable.
I've found Apple Maps is a little better in this regard. They show a higher density of business markers at any given zoom level.
> It's always annoyed me that zooming in on a building will not reliably show the business that operates there.
8-10 years ago it was way more reliable. The decline started with them adding the option to promote a business. Frustrating.
Yes, I've noticed their results are definitely becoming more opaque and driven by what they want to show you. (This is even when there isn't a sponsored option on the map.)
Yesterday I was having the same issues as the top commenter except I was having trouble getting Google to label various mountain peaks I had zoomed in on.
It would be nice if they'd fix the missing labels on roads, even at the highest zoom with no clutter. Likewise, highway speed limits that were changed over a year ago.
advertising ruins everything, users don't want to change to other services, news at 11.
A few days ago I was trying to see if a anything new had taken over a vacant restaurant space yet, previous occupant had closed in July.
When I zoomed in, it would still only show me the Permanently Closed business listing for the old restaurant.
Searching by address, they do have a listing for its replacement. But they were prioritizing the dead restaurant on the map because why would I want to know current info from a map when they can be useless instead?
And it's not like this is a restaurant in the first floor of a tower with a bunch of businesses stacked on top of it competing for map space. It's a single floor, there's only one occupant.
OpenStreetMap-based maps tend to be much better in this regard. Although this is counterbalanced by the fact that they tend to have less data on businesses in general.
Which is not surprising, as those two have very different priorities.
- OSM want's a detailed and reliable map.
- Google maps tries to either sell your data to clients, or make you buy from them.
Their business data is their priority for maps. You can see that clearly when you look at location history changes over past decade or so. It used to be actual user location history and it was glorious. Now it's "near what businesses you were more or less, help us rate them".
It's a great moment to again remind about existence of low-friction tools that you can use to add business data (among others) to OSM, like StreetComplete app, available on F-droid and Google Play :)
https://streetcomplete.app/?lang=en
In my region OSM business data starts to be on par with google, better (more up to date) sometimes.
I have recently tried to navigate with OsmAnd a few times where I live. Once I ended up in the wrong location, and a few times I have had to look up the business in Google Maps to find their address.
I would love to use OsmAnd more. StreetComplete sounds great and looks like a nice way to be able to contribute fixes to OSM. Thanks for the recommendation!
It is smooth and kind of "I'm doing my part!" but with low friction.
> a few times I have had to look up the business in Google Maps to find their address
Exactly my point - Gmaps taught us to expect *businesses" on maps. Not addresses. Pins and stars, instead of streets and numbers. Arrival time and traffic, instead of distance, elevation and road type (size).
I use gmaps still, mostly for businesses, but to actually know where I am I have better options. Gmaps hides most of typical map features - you see less of trees, water, buildings, height elevation. On Comaps/Osmand you suddenly can correlate map with things you see (without street view! :P).
If you just want to add POI data, then Every Door is a good choice that also works on iOS
CoMaps would be a good map app, and it will also display when POIs and opening hours were last confirmed (the only OSM app to do so AFAIK)
> It's always annoyed me that zooming in on a building will not reliably show the business that operates there.
It's actually much worse than that.
I will often see the business name as I'm zooming in, but if I zoom too far, it's no longer available. You have to find "just the right zoom level" for displaying the given business.
As if it were some weird mind game they were playing with you.
A lot of these place names are user-created and I’ve definitely seen completely wrong and bogus place names on Google Maps. It seems that they hide a lot of these when the business owner doesn’t actively take control of the business page. I suppose it’s partly for accuracy, partly to encourage businesses to verify the listing on their maps.
Click on the building, it populates “businesses at this address” - at least, when I’ve tried.
Just tested - slightly different UI but still works the same. Also useful for taller buildings with a lot of tenants.
Businesses inside hospitals, businesses at shared addresses, businesses underground, all sorts of great uses.
Even trying to see the street name has a very high probability of failure, so I don't know what you expect.
There are two 40-floors buildings nearby to each other in Tbilisi, Georgia, that are missing on Google Maps. All businesses have to put POI just "somewhere". One man from Google told me that there are staff members responsible for Georgia maps but they are chilling :)
The most annoying thing is when you search for instance for "Chinese restaurants" and Google maps shows me Japanese restaurants while hiding actual Chinese restaurants.
In Tokyo when I search for convenience stores, a lot of the time Google Maps will also show ATMs, assuming that's the reason I want to go to a convenience store. Inversely, if I search for a bank branch, it'll show convenience stores. The fuzzy search results can be very frustrating sometimes.
My search for thrift stores did not include Goodwill. Had to search for Goodwill explicitly.
Clever.
I don't know about other countries, but in Japan, maps will show underground passages from e.g. the metro, with exit annotated with their numbers...
Unfortunately, not all numbers are shown, even when all the exits are non-overlapping at the displayed zoom level.
information density of online maps is, in general, quite low compared to old paper maps: https://x.com/patrickc/status/1738646361128792402
I guess there's various reasons, ranging from "it's hard to make auto-layout algos produce stuff as dense as painstakingly handcrafted maps" to "let's make it harder to scrape/copy data"
Back then it was dedicated map makers that created maps. Now it's mainly programmers. So its not surprising that quality tanks when you go from disciplinary expert staff to IT day laborers.
Its not possible to be better because its not possible for even Google or Apple to verify anything anyone claims which is not static btw. The info keeps changing all the time with biz disputes/divorces/inheritence wars etc etc.
Nobody is asking for the data to be perfect at the margin. Just for it to be readily visible at all.
>I am forced to go to street view to read the sign, then type the company name into the search box to force the business marker to show up and get clickable. I've found Apple Maps is a little better in this regard.
the way you juxtapose them calls for pointing out, Apple Maps don't have streetview which makes Apple Maps a lot less convenient.
Apple Maps has had "Look Around" (their implementation of Street View) for a while now.
Where are you? Apple street view coverage isn't as extensive as Google's but there's a binoculars button for it if they do for a given location.
Hardly anything unless in a major city, no way to easily tell if there is any coverage other than randomly clicking until it shows, also doesn't tell you the date taken.
Google street view has the 2d overlay letting you know where there is coverage, shows the date taken along with previous imagery, and they have coverage nearly everywhere in the US at a least, although some of its pretty old.
Apple Maps does seem to have more up to date satellite / aerial imagery though.
Hard to overstate how valuable all that street view coverage is on the Google side.
My little Swedish village has full Look Around coverage, and clicking on the ⋯ icon shows an “Imagery” menu item that tells me the month and year the coverage was last updated. I think you’re underestimating where they’re currently at.
In the US is has basically zero coverage outside any major city. Google on the other hand has exentiqive coverage into rural areas, albeit some of it old, at least its there, where it has newer coverage it usually has multiple one at different times allowing one to look back in time as well, very useful.
I just double-checked my village. Every single road and cul-de-sac that I could find, with no exceptions, has full coverage on Apple. Google on the other hand, has coverage for maybe 50-55% of the roads. The worst example is a residential area on the outskirts where they’ve driven the car in, down one side-street, then given up and gone home.
On the other hand, they do have historical coverage, have to give them that.
Yeah so not sure why but Look around coverage is much better in Europe than the US for some reason which seems odd since Apple is US based.
You can see the very poor US coverage here: https://brilliantmaps.com/apple-look-around/
Of course compared to Google Street view there is no comparison on a world wide basis as you can see on the same page.
In areas with partial coverage Apple Maps has basically the same overlay showing where Look Around is available. It just doesn't have a great indicator as to why the option is greyed out when there's no coverage.
I mean in Google Maps you can drag the little man over the map and it has a map layer that highlights all the roads available, so you can easily see where it is and is not. Not randomly picking a point and seeing if indicator is available.
Actually… last time I checked some local addresses Apple Maps had newer streetview data than Google.
As interesting as StreetView is, it's such a colossal privacy invasion, it's absurd. In my neighbourhood, you can literally see in peoples windows, into their living rooms.
And how is this any different from walking down the sidewalk? They're on the road, they're not stuffing cameras into your living room window to try to catch you walking around nekkid or anything. It is literally documenting what public view looks like.
The biggest difference is that you would have to actually travel there and look, rather than scanning the whole city from your recliner.
I never understood why the "collaborative filtering" approach never took off with most review options. Google Maps shows you what the average person thinks is a good restaurant, meaning the rich get richer faster and tiny statistical noise converts to durable competitive advantage.
Instead, I'd love for Google to understand me well enough to show me which restaurants I would disproportionately love compared to other people based on its understanding of my taste profiles. That way, the love can be shared amongst a much wider base of restaurants and each distinctive restaurant could find its 10,000 true fans.
On top of that, it actually gives me an incentive to rate things. Right now, you only rate from some vague sense of public service instead of "this can actively improve your experience with our product".
It's not just Google Maps, Netflix used to operate on the model of deep personalization that they've slowly de-emphasized over the years. I'm still waiting for Letterboxd to introduce a feature to give me personalized film recs based on the over 1000 ratings I've given it over the years as a paying customer but they seem in no hurry to do so. Amazon used to take your purchase history into account when ordering search results but I think that's also been significantly de-emphasized.
About the only arena this is widespread is streaming music services like Spotify.
I have a theory: They realized the right approach is to focus purely on the yes/no of what you choose to consume, rather than trying to optimize the consumption experience itself.
Remember how YouTube and Netflix used to let you rate things on 1-5 stars? That disappeared in favor of a simple up/down vote.
Most services are driven by two metrics: consumption time and paid subscriptions. How much you enjoy consuming something does not directly impact those metrics. The providers realized the real goal is to find the minimum possibly thing you will consume and then serve you everything above that line.
Trying to find the closest match possible was actually the wrong goal, it pushed you to rank things and set standards for yourself. The best thing for them was for you to focus on simple binary decisions rather than curating the best experience.
They are better off having you begrudgingly consume 3 things rather than excited consuming 2.
The algorithmic suggestion model is to find the cutoff line of what you're willing to consume and then surface everything above that line ranked on how likely you are to actually push the consume button, rather than on how much you'll enjoy it. The majority of which (due to the nature of a bell curve) is barely above that line.
I think Netflix realized that reducing ratings to a simple thumbs up/down was a bad idea after all. A while back they introduced the ability to give double thumbs up which, if you can treat non-rating as a kind of rating, means they're using a four point scale: thumbs down, no rating, thumbs up, double thumbs up.
Netflix are right that 5-stars is too many, it translates to a 6 point scale when you include non-rating, and I don't think there is a consistent view on what "3 stars" means, and how it's different to either 4 stars or 2 stars ( depending on the person ).
For some people 3 stars is an acceptable rating, closer to 4 stars than 2 stars. For others, 3 stars is a bad rating, closer to 2 stars than 5 stars. And for others still, it doesn't give signal beyond what a non-rating would be, it's "I don't have a strong opinion about this".
Effectively chopping out the 3-star rating, leaves it with a better a scale of:
- Excellent, I want to put effort into seeking out similar content
- Fine, I'd be happy to watch more like it
- Bad, I didn't enjoy this
- Terrible, I want to put effort into avoiding this
- I have no opinion on this
These are comparable to a 5 point Likert scale:
"I enjoy this content"
- Strongly agree
- Agree
- Neither Agree nor Disagree
- Disagree
- Strongly Disagree
It would be interesting to conduct social science with a similar scale with merged Disagree and Strongly disagree to see if that gave it any better consistency.
When given a 5-star choice “very bad/bad/ok-ish/good/very good”, I rarely pick one of the extremes.
I suspect there are others who rarely click “bad” or “good”.
Because of that, I think you first need to train a model on scaling each user’s judgments to a common unit. That likely won’t work well for users that you have little data on.
So, it’s quite possible that a ML model trained on a 3-way choice “very bad or bad/OK-ish/good or very good” won’t do much worse than on given the full 5-way choice.
I think it also is likely that users will be less likely to click on a question the more choices you give them (that certainly is the case if the number of choices gets very high as in having to separately rate a movie’s acting, scenery, plot, etc)
Combined, that may mean given users less choice leads to better recommendations.
I’m sure Netflix has looked at their data well and knows more about that, though.
I apply my own meaning to the 5-star rating, and find it to work really well: 1 = The movie was so bad I didn't/couldn't finish watching it. 2 = I watched it all, but didn't enjoy it and wouldn't recommend it to anyone. 3 = The movie was worth watching once, but I have no interest in watching it again. 4 = I enjoyed it, and would enjoy watching it again if it came up. I'd recommend it. 5 = a great movie -- I could enjoy watching it many times, and highly recommend it.
> The current Netflix scale effectively merges Disagree and Strongly Disagree, and for matters of taste that may well be fine.
I'm a bit skeptical about this.
To me there's a big difference between "This didn't spark joy" and "I actively hated this": I might dislike a poorly-made sequel of a movie I previously enjoyed, but I never ever want to see baby seals getting clubbed to death again.
Every series has that one bad episode you have to struggle through during a full rewatch. Very few series have an episode bad enough that it'll make you quit watching the series entirely, and ruin any chance at a future rewatch.
YouTube doesn't have ratings any more, because people disliked the wrong things which made Susan very sad.
I stopped rating things on Netflix, because after doing so for a long time, Netflix still thinks I'd enjoy Adam Sandler movies, so what's the point?
YouTube got ratings, you may still up- and downvote. They however don't show down votes anymore.
Yes, you can vote but only the uploader can see it, making it pointless and equal to no ratings.
They're only useless in that they aren't displayed for your peers, but that was always the least-useful function.
Being able to see a counter that reads as "Twenty-three thousand other people also didn't like this video!" doesn't serve me in any meaningful way; I don't go to Youtube to seek validation of my opinion, so that counter has no value to me. (For the same reason, the thumbs-up counter also has no value to me.)
But my ratings remain useful in that the algorithm still uses the individualized ratings I provide to help present stuff that I might actually want to watch.
As we all know, investors and advertisers love growth; Youtube thrives and grows and gathers/burns money fastest when more people use it more. The algorithm is designed to encourage viewership. Viewership makes number go up in the ways that the money-people care about.
Presenting stuff to me that I don't want to watch makes the number go up -- at best -- slower. The algorithm seeks to avoid that situation (remember, number must only go up).
Personally rating videos helps the machine make number go up in ways that benefit me directly.
---
Try to think of it less like a rating of a product on Amazon or of an eBay seller; try not to think of it as an avenue for publicly-displayed praise or admonishment. It's not that. (Maybe it once was -- I seem to recall thumbs-up and thumbs-down counts being shown under each thumbnail on the main feed a million years ago. But it is not that, and it has not been for quite a long time.)
Instead, think of it as one way in which to steer and direct your personalized recommendation algorithm to give you more of the content you enjoy seeing, and less of what you're not as fond of.
Use it as a solely self-serving function in which you push the buttons to receive more of the candy you like, and less of of the candy that you don't like.
I have literally not rated anything at all, ever since YouTube removed dislikes, and my recommendations are working fine. Ratings indicate(d) if a given video was likely to be a waste of my time or not, and in an age of AI slop, this feature is more desirable than ever.
Someone should make a SponsorBlock/Dearrow-type addon to flag AI slop.
> I have literally not rated anything at all, ever since YouTube removed dislikes, and my recommendations are working fine.
How can you know how green the grass is on the other side of the fence if you've never even seen it?
Isn't it like Shrodinger's Grass, or Green Eggs and Ham, at that point?
(And if your recommendations are working fine, then what is this "AI slop" that you're complaining about? I don't find any of that on my end.)
You only assume recommendations are based on ratings, but you don't know. And I have seen your metaphorical green grass, because actual ratings were a thing up until about 4 years ago, remember?
>I don't find any of that on my end.
Good for you. The true crime genre has been hit hard by AI slop.