The physical observability problem
On the places that exist but cannot be found and what we are creating to solve this
There is a park in San Francisco that Google Maps says is a park. It has a green polygon, a name, maybe a couple of photos, and some grass. If you search for things to do nearby, it does not appear. To every algorithm that shapes human movement in that city, it is essentially invisible.
Cue a warm Saturday afternoon. It is full of families with coolers, couples picnicking, kids, and dogs roaming. A man who sets up a chessboard. A rotating Mexican food cart without a website. The park is alive in ways that no database registers.
The underlying is that the maps show us the places, but don’t tell us what is really happening. One version of our city lives in the databases of Google and municipal planning offices. The other one is where people actually are and what they are doing to make it so serendipitous. This is what I like to call physical observability, and we have little to no tools for measuring it, until now.
What physical observability means
In software engineering, observability refers to how well you can understand the internal state of a system from its external outputs. A system with high observability produces signals you can actually read. Low observability means the thing is running, and you have almost no idea what it is doing or why.
I started using this term in a different context: the gap between institutional designations of physical space and how communities actually inhabit that space. Physical observability is the degree to which a place’s lived reality is legible to the tools, platforms, and institutions that make decisions about it.
For example, A Philz Coffee in downtown San Francisco has extremely high physical observability. It has reviews, foot traffic data, real-time hours updates, photos uploaded this week, and check-in signals from 6 different apps. The city planning office knows it exists. The algorithm knows it exists. Capital knows it exists.
The chess park in San Francisco has almost none of that. Neither does the night market that pops up in a parking lot every Friday in Manila. Neither does the community garden in Santiago that emerged spontaneously and has been a neighborhood hub ever since. Neither does the indoor badminton court tucked behind a hardware store in Seoul’s Mapo-gu district, which has been running leagues for 15 years and has never had a Google Maps listing.
These places are not small or marginal. They are the actual fabric of how people live in cities. They are just unobservable to the systems we use to navigate, fund, plan, and govern urban life.
Urban researchers have a related vocabulary. Ray Oldenburg’s foundational work on third places identified the informal gathering spots, bars, parks, and barbershops that sit outside home and work and do the social work of communities. More recently, scholars have described a fourth category: what Patricia Aelbrecht calls “fourth places,” informal spaces characterized by what she terms in-betweenness, adaptive to social activities, and precisely the kind of space that maps poorly onto commercial discovery platforms. This is the category your physical observability framework is built to make legible.
Mapping shows what exists
It would be convenient if this were simply a matter of better data collection. Deploy more sensors. Scrape more check-ins. Train a better model on street-level imagery. Problem solved.
But the physical observability problem is not primarily a data problem. It is a category problem. It is a power problem. It is a problem of whose reality counts as real.
The categories used by commercial mapping platforms were not designed to capture community life. They were designed to capture commerce. A business has hours, a category, a price range, a point on a map. A public space has a polygon and a name. The rich middle ground, the space that is neither business nor formal park, the gathering that is neither event nor institution, is simply outside the ontology.
This argument has a deep intellectual lineage. James C. Scott’s foundational work on state legibility describes how modern institutions systematically simplify and flatten what they need to govern, producing what he calls high-modernist maps of social reality. His concept of metis, the local, practical knowledge that cannot be captured in formal systems, describes exactly what physical observability attempts to recover. The informal park, the oral-knowledge gathering, the spontaneous plaza: all of these are expressions of metis that legibility-obsessed institutions render invisible.
Municipal planning systems have a related but different bias. They operate on the parcel. They classify land by permitted use. Research on Mumbai’s development plans demonstrates this precisely: a wedge forms between planned and actual land use that planners call the de jure-de facto divide. Master plans based on technical principles are structurally unable to foresee or adapt to how communities actually inhabit space.
This is not neutral. The communities that tend to have the lowest physical observability are, almost without exception, the communities with the least institutional power. Informal settlements. Indigenous gathering places. Immigrant enclaves. Neighborhoods where the social infrastructure is maintained through relationships and oral knowledge rather than through institutions and platforms. The places that are hardest to see are the places where the cost of being unseen is highest.
Rob Kitchin, whose work on data-driven urbanism is the most rigorous treatment of this problem in the academic literature, frames it this way: urban data not only describes urban systems but also shapes how a city should be governed and what it should become in the future. The partiality of data is not a technical bug. It can be a political feature.
There is a feedback loop here that I think is underappreciated. Places with high observability attract more visitors because they are discoverable. More visitors generate more reviews and check-ins, further increasing observability. Places with low observability stay low. Capital flows to the observable places. Investment follows the signals. The neighborhoods that produce strong commercial signals attract more investment, making them more commercially legible, which in turn produces more signals.
A systematic review of 319 research papers on urban public spaces from the user perspective identifies three unaddressed gaps that map precisely onto this dynamic: the interpretation of user perceptions, overlooked user demographics, and data-acquisition failures. The communities with the richest social lives are precisely the ones least equipped to produce the data signals that would make those lives visible to platforms.
This is how a city gradually optimizes itself toward what can be measured and away from what cannot. And the things that cannot be measured are not trivial. They are often the most important social infrastructure in a given community.
When Google Maps feels incomplete, it is not because it has failed in its mission. It is because the mission was too narrow from the start.
Social signals
Here is what makes this tractable, or at least tractable in new ways: social signals are everywhere now. People are constantly generating data about where they are and what they are doing there. Instagram location tags, Snapchat stories, Facebook events for gatherings of twelve people in a living room, Spotify listening sessions anchored to a place, Strava routes that reveal which unofficial paths people actually prefer, event pages on a dozen different platforms, check-ins, photos, stories.
Almost none of this signal has been aggregated and put on a map in a coherent way. Research on social media check-in data confirms that this kind of user-generated geographic information provides a new perspective on people’s spatial and temporal preferences in urban places and can portray urban structure in ways that traditional data sources cannot capture. The signal exists. The layer does not.
Community advocates working at the intersection of data and neighborhood planning have articulated this as the difference between near data and far data. Far data is what platforms and governments measure. Near data is what communities know about themselves. The research is detailed: advocates see far more value in near data to make sense of, enrich, question, and challenge the far data that governs decisions about their neighborhoods.
This is the specific problem I have been building toward solving. Not by collecting new data, but by making existing social signals legible at the level of place. By creating a map layer that aggregates live activity across platforms into a single view of what is actually happening where, right now, and over time.
The goal is not to commodify community life. The goal is to increase the physical visibility of places and gatherings that are currently invisible and to give visibility to every system that matters.
Why this matters
One thing I have learned from five years of working on this problem across four continents is that the hardest part is not building the tools. The hardest part is getting people to agree that the problem exists.
When you talk to city planners about informal gathering infrastructure, they often know exactly what you mean. They have seen it. They have watched communities build extraordinary social fabric in spaces that their own systems cannot see. But they do not have a language for it that travels. They cannot put it in a budget justification or a planning document.
When you talk to investors about the gap between the city that apps see and the city that people live in, they often feel it intuitively. Most of them have had the experience of arriving in a neighborhood that felt alive and warm and completely inert according to every app on their phone. But they do not have a frame for why that gap exists or whether it is addressable.
Physical observability is that term. It names the category of problem that all of these people are describing from different angles. It connects the community organizer in Manila to the urban planner in Auckland to the founder in San Francisco who is trying to build a product that helps people find each other in the city they actually live in.
The invisibility framework developed in critical data studies offers a useful corollary: by understanding what remains and what must remain invisible, we can better grasp both the limitations of our tools and the power relations that produced them. Naming is the first act of measurement.
The category-defining future
A city with high physical observability would have planning decisions informed by where people actually gather, rather than just by where commercial activity is. Informal parks and gathering places would have documented social functions that could be weighed in development decisions. The value of a corner where elderly residents play cards every afternoon would be legible to the system, not invisible.
Community organizations can demonstrate the scale and consistency of their programming with data. Not to prove their worth to some extractive metric, but to advocate for resources and protection with evidence that travels.
People arriving in a new city or neighborhood can find the real social life of a place, not just its commercial surface.
The map would get closer to the territory. Not all the way. The map can never be all the way. But closer. Close enough that the territory gets treated as real.
Bibliography
Aelbrecht, P. (2016). Fourth places: The contemporary public settings for informal social interaction among strangers. Journal of Urban Design.
Benjamin, S. (2008). Occupancy urbanism: Radicalizing politics and economy beyond policy and programs. International Journal of Urban and Regional Research.
Bhide, A. et al. (2014). Re-thinking urban planning in India: Learning from the wedge between the de jure and de facto development in Mumbai. Cities, Elsevier.
Kitchin, R. (2014). The Data Revolution: Big Data, Open Data, Data Infrastructures and Their Consequences. Sage Publications.
Kitchin, R. (2015). Data-driven, networked urbanism. SSRN Working Paper.
Neumayer, C., Rossi, L., & Struthers, D. (2021). Invisible Data: A Framework for Understanding Visibility Processes in Social Media Data. Social Media + Society.
Pei, Z. et al. (2025). Near Data and Far Data for Urban Sustainability. arXiv:2501.07661.
Scott, J.C. (1998). Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed. Yale University Press.
Wang, J. et al. (2024). Understanding the user perspective on urban public spaces: A systematic review and opportunities for machine learning. Cities, Elsevier.