City Ranking: new algorithm

Hi Everyone,

Some of you already noticed the overall city ranking , it’s like an unofficial ranking of cities based on some numerics (@thomas can you explain again what’s behind this.)
I’ve been thinking a lot of creating a new calculation behind this ranking which includes overall numbers, engagement and popularity.

So thought why not ask you what kind of stats should we include?

My first ideas look like this:

1. Numbers

  • artworks/m²
  • artists who painted in a city

2. Engagement

  • open community notes
  • processed community notes
  • ‘seen’ clicks in the app in the last 6 months
  • new uploads in the last 6 months
  • archived items in the last 6 months

1. Popularity

  • artwork “all time likes”
  • artwork likes in the last 6 months
  • city followers in the app
  • scans of an official city map QR code
  • map views

Our offcial partner-cities should have bonus points?

So what do you think?


I really like the idea of artworks per m2, as a way to even out the difference between small and big cities, but it’s difficult to actually find a decent way to do that technically.

Data sources that will give you information about the area size or population of any given city don’t seem to be widely available - the closest thing I could find is Wikipedia/Wikidata, but I’m not sure how many of our cities are covered there, I should probably give that a try and see what the hit rate is like.


I like the artworks/m2 idea too, but understand Thomas when he says it might be difficult to get correct data.

Also, if I use my hometown of Rotterdam as an example, it’s not really clean data. Rotterdam is the city with the largest port in Europe. This means that the city has a large number of m2, but most of it is taken up by ports and port companies.

Maybe artworks/inhabitants is better and easier to use?

An other thing that may be considered could be number of routes in a city?

This works for little street art villages like Penelles in Spain. such a place has to be high ranked anyways.

As Edwin said, for the bigger cities with each a different character like, the half of the territory is a part of a port will Rotterdam rank much lower probably cause of this algorithm then, while there is plenty to find in Rotterdam city itself.

Thanks for your input Edwin.

Whilst I understand the intent of the artwork/metre squared and the other suggestion of artworks/population I think both would be very difficult to implement technically and also would offer very strange and unpredictable results. For example a very small town like Woomelang with it’s population of less than 200 people and small area, and it’s 9 artworks, would leap over most major cities with their artwork totals in the thousands, by either measure. This would make the rankings a bit of a farce I think.

I also do not know where you would get the data for metre squared, I cannot find it for many Australian cities.

I also think that these measures would also create a huge disincentive for those of us who are trying to cover the large cities like Melbourne. We spend many hours (and dollars on petrol or public transport) travelling to keep our map up to date, so to me the larger area cities with limited “active” hunters should be rewarded, not punished for covering a large area.

I’d like to see the rankings page with multiple options, perhaps you could rank by artworks (total), artworks (active), a data quality measure, an engagement measure etc., perhaps a way of filtering only partner cities if required (I’m not sure they should have “bonus points” in any ranking). I think offering a number of simpler ranking choices like this is a better idea than coming up with some complex algorithm that no-one really understands.

Whatever ranking methodology is chosen though I think it needs to be explained well to those viewers of the rank screen so they understand what they are looking at, in a help screen or something.

Obviously any engagement ranking data (ie use of the app) is going to favour those countries where that has been in wider use for a longer time, and I’m not sure how hunters can influence this really.

If you are going to reward “Community Note” processing and punish “Community Note” inaction then I’d like to be sure hunters are given the information they are supposed to be actioning. For example the recent handling of a community note for Melbourne by a moderator was for a community note that neither Neda or I even saw in our Inboxes (I still don’t know why that was).


Totally agree Andrew.

But still I have the question: It is so difficult to creation a correct as possible list. Is a ranking list really important?
I think the motivation we should have, is anyways trying to have a much as up to date city we hunt.
I do understand, that it’s motivating to see your city higher in the ranking, because of your hard job. You guys, lifted Melbourne to one of our top cities on Street Art Cities, we are grateful for that.
Still, I am always a bit reserved when it goes about rankings and quantity. I don’t need a list, to keep my city Oostende maintained, and it never influenced my mood or ego at all. I am proud of what I am doing, anyways.
When pushing to hard to numbers and rankings, after a while, the quality will decrease anyways (as we experienced already in fact). Cleaning up a platform is necessary, but not fun when people start to see it as a race to see their city high ranked as possible, and using tricky tricks cause that’s indeed the danger as you mentioned: some of them are gonna invent some tricks to influence the ranking. And then indeed, we gonna compare little cities, and bigger cities, how many km we drove etc in comparison with a hunter who keeps a village always solid up to date, or even guide people around in the village. This has a value too?

I still didn’t read a real legit reason to even keep that ranking alive. It’s often a point of discussion and as we see… how many algorithms you have to include to create a probably good, even artificially built ranking? Neverending discussion. And if one hunter has another opinion, which makes sense, we need to change the algorithms again? Until another hunter has a valuable potential algorithms to implement? Its not really my thing or DNA…

But, that’s my personal opinion, probably a bit different then what the most others or even Core Team thinks.

I am in agreement with a lot of what you say. I think we also should think about who might use the city ranking, it is not just for the hunters who are maintaining the city, it is more important for visitors. I think if I was planning a new trip somewhere, it is useful to see which cities have the most artworks (which is why I suggested an option to choose active artworks as the the ranking criteria). I think in this regard it might also be useful to be able to filter on a country too.

Giving the cities a “data quality” ranking is a good idea too, but that algorithm has to be well thought out I think, and in my opinion should be separate from other rankings. I think it is the lack of clarity over what the current ranking means that causes confusion (and perhaps some frustration?). To an untrained eye (SAC user) seeing Antwerp (512) sitting above Bergen (1002) and Berlin (608) sitting above Sydney (948) with no explanation as to what it means, must be very confusing.

I think that the data currency is important so perhaps for the data quality ranking a greater “weight” could be given to artworks that are more recently “seen” or updated. Older records, untouched and unseen for years get a lower weight?

Is it ego that makes you like to see your city high on a list of “Street Art Cities”? I prefer to think of it as pride in your city rather than ego. Has seeing the movement of Melbourne up the list been motivating as Neda and I have added over 1,000 artworks between us over the last couple of months. Definitely it has and I don’t feel there is anything inherently egotistical about that. But as you say we all have different opinions, and that is what makes life interesting.

Exactly. The ranking should be for visitors and… not really for us as hunter. Good point, as I mentioned before actually. The ranking is not a race or competition and we should try to avoid this in any case when and where possible.
The current ranking causes many doubts, questions etc. The only ranking which makes sense in a way is the the total artworks.
Perhaps, we shouldn’t make that ranking seem more important than it is.

A complete list with countries- cities - number existing artworks (and maybe (historic artworks) ) is actually more than enough? All the rest is a source of neverendering discussion, comparing cities, towns, villages with totally differente character, even a totally different urban art scene (like… Belgium is not really that paste up country… 100 km away into Germany… you find plenty of that).
So, in every aspect of urban art, u need to implement algorithms then and at the end of the day… you know what?
We talk about art, this is something with a really different wide taste(I am not talking about hitting a like button on the app, but is this is not relevant in comparing with the app views in a city). Are you sure that a brilliant artwork from Roa in City X is 100X hitted by an app user? You should be surprised how less the like button is pushed in comparing with mobileviews. I barely, or better say, I never use the like button. In function of the app, seen is for me priority. If l like the work, or don’t like. Yesterday did again a little tour in my city to hit the seen button(and found a gone one yes, added 2 other older pieces… 2 places higher in the ranking… ridiculous in fact).

In the French language, there is a very well known expression
“des goûts et des couleurs, on ne discute pas”

@Streetartwerpenaar Mentioned: partner cities bonus points? Could be interesting. I had a quick peak on Vilvoorde, and have seen which kind of uploads the hunter(City Vilvoorde) did… not even a complete sprayed letters. Not even stuff which should be allowed on the map actually.


City ranking. Several before me has adresse the complexity of city ranking. pr m2 or pr km2 seems OK but brings up problem of area in a city which is non inhabitable. Non inhabitable would be mountains, water - both fresh and sea. So a flat city with little non inhabitable could score high and visa verse would score low. Then you have odd cities like Kiruna (Sweden, no1) and Tromsø (Norway, no2) - they are the largest cities (correct me if I am wrong) in the world when it comes to km2. Kiruna / 19,447 km2 and Tromsø / 2,521 km2. But do not have much people respc 18K and 77K.

Per capita can be an interesting angle. But …

In general from looking at different cities in the map - us the hunters do not use same criteria for registering pieces. Thereby the foundation is not robust. When it comes to building relations on these registrations and extract relations and make coherence comparabilities and evaluations will not create a good and true picture in a city ranking.

Artists who painted in a city. This is interesting since it tells us if a city is hot and could be a place for artists to visit. It also tells us something about how the city responds to street and urban art.

My suggestion, based on non a robust foundation of primary data, and external data which makes no distinguishing of habitable and non habitable areas, that you do not put more efforts into ranking based on m2 or km2.

Per capita would use a comparable number from public registrations. This then seen up against total number of works is maybe a relevant direction. A mathematician or economist would be able to come up with a formula for application. Then again the foundation is not robust as Stef mentioned in his lates comment.

Our lives and how we hunt will influence. Take me, I live the life documenting street and urban art on a 24/7/365 bases. I know all the artists in my area (90+) - both the incognito and on incognito. I have press card and can hang with them at night when they work - police can’t touch me. I also initiate and rund projects - at the moment we have painted more then 30.000 m2 in projects alone, so I am part of the community. Artists uses my pics pro bono. So it’s a win-win relationship giving me access to most if not all pieces. Often getting heads up prior to new stuff being made. Also I find myself as a link between society in general and the subculture. The press comes to me asking and some times joining when we are out at night. Even TV (news channels) hangs with us. Thereby our platforms for hunting and gathering are different. This adding to the “non robust” foundation, our raw data i.e. what has been registered AND if we are consistent in clean up when something is not there any more. Hard to see how one could get a fair ranking.

Bergen, Norway
On StreetArt Bergen (Redirecting...) I keep a chronological update of work added to City of Bergen if you are interested.

1 Like

Intentions are good.
My suggestion: Raw data not robust.
Leave dead :wink:

1 Like

I am glad that I am not the only outspoken one with this view. Making a list, editing till we satisfy hunters and app users is nonsonse, and waste of time. You can NOT get a fair ranking.

1 Like

This is all super useful feedback, and I think we’re all reaching the conclusion that a single number isn’t going to cut it.

I really like your suggestion @Andrew_Haysom of allowing the user to sort on multiple dimensions:

“I’d like to see the rankings page with multiple options, perhaps you could rank by artworks (total), artworks (active), a data quality measure, an engagement measure etc., perhaps a way of filtering only partner cities if required (I’m not sure they should have “bonus points” in any ranking). I think offering a number of simpler ranking choices like this is a better idea than coming up with some complex algorithm that no-one really understands”

I think we’re going to implement something similar to this, currently thinking of the following options to start with:

  1. Number of artworks
  2. Recently updated
  3. Most visited
1 Like

This is now live on the cities page.

I’ve left ‘Popular’ in there for now to not change the default order right away, but it now has a clear definition (combination of artworks count + map views), which is also listed at the bottom of the page, alongside some other details.

When you change the sort order by choosing an option from the dropdown, it also changes the text shown on the right of the city names to reveal the value that the sorting is based on.


Nice work Thomas.

I notice though that the Melbourne map icon is missing from that screen now too though (as well as the home screen). I wish that could be fixed :slight_smile:


Oh, that bug is back?!

It really has me stumped haha, is it only the Melbourne one that is disappearing for you?

It loads okay for me weirdly…

Yes, it was never really fixed I don’t think, it did appear briefly for a couple of days, but has been gone again for a while (I’ve attached screenshots), it’s not just me as I know it is same for Neda, and I just had my wife check on her browser on her iPad and missing in both spots for her too.

If it is of any help, I just had a look at the page source code in my browser and found the file names for the tiles, the two screenshots attached here show what I see when I try and load the tile for Melbourne and tile for Los Angeles.

Melbourne on my machines is OK.

Did you do something? The map tile icon seems back on my laptop now.

Yes, nice! I reset the cache on our Australia server, it appeared to still be caching a broken version from back in August.

Fingers crossed it keeps working now!