Better Community Health Through Improving Building Social Interactions Via Architecture

0 10

Imagine if the architecture of buildings was designed from the beginning to aid occupants to get to know one another, to become acquaintances and possibly friends, to look out for one another. The benefits can be remarkable. In public health and social sciences, the evidence is clear that mental and physical health are improved by social interaction. Simply saying “Hello” to someone can improve their day and their health.

If you are an architect or a developer focused on architectural solutions, you have an opportunity to help shape this.

Cutaway of a multi-unit residential building with annotation of the social syntax Human Studio has developed

Cutaway of a multi-unit residential building with annotation of the social syntax Human Studio has developed

As we move out of our isolation due to COVID-19 and back into interacting with others more regularly, the built environment is being reconsidered. But we cannot allow an over-reaction to the pandemic to overshadow the value of human interaction.

Into this space comes FLUID Sociability from Human Studio Architecture and Urban Design. It’s the brainchild of Bruce Haden, principal of Human Studio, and Ryan McQuaig, an architect who branched out to developing software tools for architects. They and the team at Human Studio, along with Professor Liz Dunn, a social psychologist at UBC whose research focuses on how time, money, and technology shape human happiness, have a vision. They wanted to develop a tool which could assess the opportunities for social interaction in a building during the design process so that sociability could be in the bones of the building. They thought that the use of simulated humans, agents, navigating through a virtual building during the design phase, would provide both insights which were hard to arrive at otherwise, and a quantifiable assessment of different building designs for the same design brief, to allow more sociable designs to have a greater opportunity to be the final ones built.

Bruce’s name might be familiar to CleanTechnica’s readership, and to those who follow the CleanTech Talks podcast. A year ago, Bruce and I had a wide ranging discussion about urban design, urban magnets, and COVID-19 which went live in two parts (part 1, part 2). At the time, my firm distnc technologies was in the early stages of discussion with them about realizing this vision.

The vision was formed years ago when Bruce and Ryan were engaged in designing a building here in Vancouver BC. Their research and professional experience had led them to understand that some building designs enabled more interaction, but they didn’t have a rigorous way to prove it. And in the absence of a rigorous yardstick, their visualizations and presentations weren’t persuasive enough to overcome other design considerations, and the design that they were sure was more sociable and hence healthier was not the one selected in the end.

This is a similar challenge faced by all architects. As Frank Gehry regularly said, good design only adds 10% to costs, but that 10% is often excluded for budget reasons. Gehry’s focus, however, was on the materials and on stunning aesthetics, not the social interactions that buildings enabled.

The infamous Pruitt-Igoe social housing project in St. Louis, assessed within the FLUID Sociability SaaS tool

The infamous Pruitt-Igoe social housing project in St. Louis, assessed within the FLUID Sociability SaaS tool

And now the vision is a reality. With my distnc co-founders, David Clement and Chris Wiesinger, and our development partner, Allion Technologies, we’ve helped Bruce, Ryan and the extended team to deliver a solution. This was enabled by a generous grant from the Robert Wood Johnson Foundation in the US, a philanthropic organization with a focus on improving human health and well being, and CAF America. Regular readers will remember David as a frequent collaborator of mine, most recently on the major CleanTechnica report on machine learning and climate solutions.

We have a growing body of global collaborators, with validation starting up in Australia where they are mostly post-COVID-19, and on the UBC Campus, where Professor Dunn and her team are surveying students about their social interaction experiences in student residences on campus, residences we’ve modeled and assessed. 3XN out of Denmark is engaging with us given their focus on humanizing buildings and the greater breadth of sustainability.

FLUID Sociability is a software-as-a-service (SaaS) solution which enables architects to assess alternative versions of buildings that they are designing to get a strong comparison of how they help or hinder the occupants from interacting. The architect continues to work in their usual tools, Revit or Sketchup, uses some Human Studio families and objects to annotate their models, and uploads them into the SaaS site for assessment.

The simulator finds the suites in the building, the amenities such as mail rooms and gyms, and the possible paths through the building. It creates occupants for each suite, including children, seniors, and adults who are both externally employed and working from home. The occupants have a caution factor created for them, randomized with ranges for adults, children, and seniors, that abstracts away from potential stereotypes related to gender, ethnicity, and other factors, but respects that people will react differently to enclosed spaces, to interacting with strangers, and to initiating greetings or conversations. Children, adults, and seniors also have different and varied speeds of movement through the building. Stairs and elevators are both options for traveling between floors, with distance and diminished physical capabilities of some agents both impacting choices of routes. This creates a rich but respectful palette of occupants with different goals, different schedules, and different preferences for path finding and social interactions.

From our set of agent use cases — go to work, go to school, go shopping, use the gym, get mail, etc. — the simulation creates calendars for every occupant for every day of the simulation. It simulates every agent going about their days, selecting the paths that they take out of the many potential paths. Every second of every day of the simulation, it knows where the occupants are and where they are going. It keeps track of where they bump into one another. It does this automatically from the simple annotations added to Revit or Sketchup without the need for architects to learn a new and complex tool.

It’s possible right now to simulate buildings with 500 occupants for 365 days of interaction, and to see the webs of interactions and changes of interactions over time. We’re pushing the boundaries with a goal of being able to simulate campuses and small groups of buildings. Distnc technologies is looking at multiple overlapping uses for the approach, and has already done early simulations of COVID-19 interaction modeling, another health aspect that’s critical as we open up again.

The team has assembled and reviewed a major body of literature on sociability and human interactions. We’ve looked at how quickly people recognize one another, and in what conditions. We’ve looked at cognitive science research on facial recognition and human interactions. We’ve assessed the global literature on buildings designed for sociability, and what was empirically valid from those results. We’ve embedded this massive set of data into the software and determine how our agents encounter one another and interact. The literature is listed in the FLUID Sociability community site, along with the paper Bruce, Ryan, and David presented at SimAud 2021 earlier this year on our efforts.

Buildings are three-dimensional, and encounters can occur both horizontally and to higher and lower levels

Buildings are three-dimensional, and encounters can occur both horizontally and to higher and lower levels

And we’ve run into the limits of well-plumbed human knowledge. We know that people recognize each other first based on the internal features of the face, the eyes, nose, and mouth, and then can recognize familiar individuals at a distance based on external features such as hair, ears, clothes, and gait. But there isn’t a lot of literature on how quickly people become familiar with one another outside of labs. And there’s very little on the reasons why they might choose to greet one another, or enter into a conversation, or how often. We’re breaking new ground with our academic partners at UBC, Professor Liz Dunn, and her PhD students such as Iris Lok.

We have put a stake in the ground by creating a set of clearly articulated algorithms for how agents choose paths through the building, including weather sensitivity and innate caution, what constitutes an encounter, how agents develop familiarity with one another, and under what circumstances they will greet one another, and further, under what circumstances they will have a conversation. The last two are increasingly into the realm of the speculative, but have the merit of being somewhat supported by research, and further by providing a transparent and testable framework.

Visualization of the embodied algorithm for when two agents greet one another

Visualization of the embodied algorithm for when two agents greet one another

One of the most interesting things for David, Chris, and me in distnc technologies is that we have created agents with individual minds. Every agent in all of their interactions sees the interaction solely from their egocentric view. It’s possible for one agent to see another agent multiple times and consider them to be familiar, but for the other agent not to have the same degree of familiarity. Similarly, each agent with its different caution factors and familiarities will have a different propensity to greet and converse with others. While it’s early days, we’ve actually embodied a full framework for Shannon communication model communications between the agents. Greetings actually occur between our agents and each agent remembers those greetings individually. Some of this is an outgrowth of work David and I were doing on emotional agent systems a couple of years ago, neatly folded into this valuable solution. More is from our joint reading of Kahneman and others in cognitive science, an enduring fascination.

We’ve mostly been stymied in our attempts to get fob data for buildings to validate our underlying calendars by reasonable concerns about resident privacy, but have been able to get one set for a suburban building in Canada, which improved our approach. That led to a model which allows architects to vary departures from the simulation for work, shopping, or dog-walking based on the building’s location. If it’s in a dense urban area with high transit, bikeability, and walkability, it’s easy to have people exit on foot. If the building is in a suburban area where everyone drives, it’s easy to annotate designs to have most people exit via the parking lots. This enables architects to make the decisions specific to their design brief, without being hampered by stereotypes of urban vs suburban living.

FLUID Sociability score for Pruitt-Igoe from the FLUID software

FLUID Sociability score for Pruitt-Igoe from the FLUID software

We represent the results in a variety of ways. To the core mission of quantifiable results, our current summarized version is interactions per agent per day, boiling down the often gigabytes of data from second-by-second virtual months long simulations into simple, comparable numbers. Pruitt-Igoe is famous as a failed social housing project, where the bonds of community cohesion never gelled, and became instead a complex plagued by violence, vandalism, isolation, and eventual failure and demolition. The reasons are far more complex than just the building itself, but the sociability score for the building is interesting. For this simple 14-day simulation of one of the buildings, with 494 occupants, the average person saw someone only every other day as they went to work or school, went in and out for shopping, picked up their mail, did their laundry, and the like. And in large part due to this paucity of encounters, they rarely greeted someone outside of their households. We’re just in the process of taking the next step of simulating greetings turning into conversations, and adding the impacts of weather in exposed settings to the solution.

For those interested, here’s a video I recorded panning around the Pruitt-Igoe 3-dimensional model, turning layers of suites and navigable space on and off, and showing the layers of representations of encounters, greetings and where there is more versus less traffic in the building, per our simulation. As a brief technical aside, the solution uses the GLTF open standard for browser-based viewing of sophisticated 3D models encoded in json datasets with full panning, rotating, and zooming.

Among other things, it shows that this one building is actually three buildings, and that it has no patios or balconies where people might have lingered in view of passers-by or those on other balconies or patios. There were no courtyards or open walkways that might have led people to see others as much. To be clear, there is no intent to say that the building itself was the cause of the social challenges which occurred in the complex, but the architecture didn’t help.

Visualization of which occupants interacted with which other occupants, grouped by suite

Visualization of which occupants interacted with which other occupants, grouped by suite

One of the most aesthetically pleasing visualizations has come from the Vega graph work which Ryan McQuaig has been strongly focused on. This interactive Vega graph visualization allows architects to explore which occupants in which suites encountered which other occupants from other suites in the building. Combined with the 3D rendering of the building, it’s intended to allow architects to see how their architectural choices might impact the lives of the agents in the building. As another brief technical aside, Vega is an open standard for data visualization from json data sets.

At the beginning of this piece I teased that architects had the opportunity to help: here’s how. The FLUID Sociability SaaS software has just entered beta. You can sign up, annotate one of your Revit or Sketchup building models, and simulate sociability for free right now. And provide feedback and insights to Human Studio and the team. Bruce and the team welcome more collaborators as we take the wraps off of this year of intense effort after several years of considering how to make their vision real.

As for developers, the Robert Wood Johnson Foundation grant is aimed at public good, and part of it is the release of any software built to open source. We’re in the process of defining the subsets of the solution and intellectual capital developed in the past year that this applies to, and it will be up on github under an appropriate license within the next few weeks. Reach out if you are interested.

This has been a rich and rewarding journey for me personally. I’m very pleased to have been part of this effort, and working with people like Bruce, Ryan, Pete, Carmen, Andrew, Altaira, Liz, David, Chris, and the rest of the international team has been intellectually stimulating. I’m very pleased that our efforts are being released into the wild, and look forward to hearing of buildings shaped into better machines for living through our efforts.

Appreciate CleanTechnica’s originality? Consider becoming a CleanTechnica Member, Supporter, Technician, or Ambassador — or a patron on Patreon.



Have a tip for CleanTechnica, want to advertise, or want to suggest a guest for our CleanTech Talk podcast? Contact us here.

Leave A Reply

Your email address will not be published.

This website uses cookies to improve your experience. We'll assume you're ok with this, but you can opt-out if you wish. Accept Read More