Human waste in San Francisco

, , Leave a comment

Some maps should not be published. Not because they deal with touchy subjects ‘human waste’? iiirk. Or because the data they present is wrong. Simply because they just don’t add anything to the discussion. I think maps are really powerful; compared to text for instance, they give a spatial context in a split of a second, they cross language barriers, and so they are easy to take for face-value… and therefore to be misinterpreted. As a cartographer, there are maps that will de facto be wrongly interpreted, or just won’t give the whole picture.

Wasteland app

Take this map whose intention is to highlight the lack of toilets in San Francisco.

Screen Shot 2016-07-06 at 17.51.29

The look

On the interactive map, you can select a neighbourhood, enter an address, or select monthly data. You can also switch from point to aggregated data (HEAT – as shown on the picture). It is a beautiful interactive map – fully customized (background map colour and markers), and with a lot of advanced features (geolocation of address, with calcul of the number of points in 2.5mi buffer, and heatmap). It’s a GoogleMaps API, which positively surprised me; it’s nice to see that it can also fully customized. Usually highly customized maps use Leaflet, because it’s currently the most easily accessible to programmers that come from javascript rather than GIS.

The storyline

So what does it really DO? It clearly shows that there is a problem of human waste in San Francisco, and that this problem is localised in one area (the Tenderloin district – which would surprise nobody who visited San Francisco or heard about homelessness issues in San Francisco). It shows very little less, especially because of the choice of to aggregate the data with that transparency (it highlights the places with very high concentration but the shadow makes it difficult to see where there is nothing) or, when not aggregating to use a silly poop picture (it is cute, but that’s a little bit too big and so hides the story).
The website however links to three articles that raise the real issues highlighted in that map (not around the data, but let’s talk about it later). There are a lot of homeless in San Francisco, which means there are a lot of people that do not have places to take care of their basic needs. The city on the other hand does not have enough public toilets or homeless shelters to compensate for the lack of affordable housings. Instead of telling us where we might step into a poop or highlighting the areas where poop was found in a given month, maybe this map should not just highlight the results (=poop) as well as the causes.
There is a direct cause to all these wastes; and that is the lack of public toilets. Apparently more public toilets have been installed recently, although the problem hasn’t been solved. One of the articles linked on this website also shows that homeless are using Drop in Centres for instance for their basic needs. It also shows how difficult it is to solve this problem – free self-cleaning toilets had been misused so much for doing other things that they eventually became useless.
So there is a problem, and what would a map be powerful enough to do? Reading through the articles, I realised that there had been many different trials. That the latest ideas are only dealing with the ‘pee’ issue (with innovative urinals – at least these will not be misused for prostitution or drugs) or in an expensive way (airpnp is supposed to be an airbnb of toilets. But really $3 to go to the toilets? That’s more for tourists than homeless).

The data

There is a lot to be said about the data – who is collecting it? is it a systematic task or relying on self-collection? There is an obvious biais on data collected non-systematically – it might be due to the population density (VGI efforts like OSM are more comprehensive in dense areas), or the digital divide, NIMBYISM, and so on. For instance, clearly there is not much data in parks – is that because there is nothing to collect, or because nobody reports it there?


More context with more data

Showing only one dataset shows only one side of the story. Other dataset could have been included
It was difficult to find open data about toilets, as they are not included in the SF open data portal. A quick google search found a CSV that was used during a SF Hack Day and which contained 63 geocoded toilets. OSM has also about 60, and has more complete metadata (opening hours depending on the day of the week, wheelchair access, location of the toilets, cost, and so on). However it has duplicates, which means the data needs to be cleaned.
I also looked into homeless shelters, and found 8 in San Francisco. The map below for instance shows these all together on Carto (ex. CartoDB – not sure I’ll get use to this name!)

Showing a story rather than a picture

It is easy to think that interactive maps tell stories; users can ‘query’ the map and therefore ‘read’ better the data. However, reading through the links to other articles that were given on the wasteland website, it is clear that many solutions to the human waste issue in SF have been tried – free toilets, or strict enforcement, or fancy toilets.
Instead of mapping all the years together, it would be quite valuable to see the different strategies adopted. An interactive map should be able to show the state of the streets at different strategic periods: the one before JCDecaux toilets were installed, the one when it was made illegal to poop in the street.
I need to quite my job to get the time to do all this 🙂