Final Project: Remote or On-site?
If we no longer need to (literally)go to work, how should our city look like?
1. Project Overview
Remote work under the pandemic has initiated transitions towards different styles of work and living, bringing different evaluations on existing building types and blocks.
And this trend is very likely to persist with technologies like VR.
For example, since most employees were conducting remote work, many office buildings in Lower Manhattan were mostly empty.
This part of Manhattan was selected considering its proximity to various parks and its dominant program type as residential buildings.
Goals and Metrics
I would like to explore how people might decide on whether to commute based on their various conditions and how a different type of building might change those decisions.
- Set up a series of metrics and analysis tools to help predict each residents’ decision of remote work.
- Use an agent model and persona input to combine all the metrics applied to each agent.
- Use a new procedural type to replace all the existing buildings on the site, while keeping all the statistics of areas consistent.
- Evaluate performances of current building types and a new procedural type.
What Did I Discover?
=> Distributed programs might be a fair solution to large-scale remote work.
Nowadays, different programs within an urban region are often unevenly distributed, especially residential and office clusters.
It might be a better idea if we offer better coverage of “micro-scale” office/retail /commercial spaces in residential districts.
2. Computational Design Model
Composition of the computational design model. Metrics were adjusted during the last two weeks but the overall structure was followed.
The input layer handles the GIS data imported from the PLUTO database, generating streets, lot boundaries, area statistics.
The building layer creates building forms out of the data provided in the first layer, including residential, park, office/retail spaces. Then a series of metrics will be calculated, including proximity to parks/subway stations, viewing distances, etc.
The agent layer further extracts the number of agents and their spatial distribution.
Finally, the metric layer uses basic metrics from the building layer to predict their remote/commute decisions, leading to the average prediction on each building block.
Analysis Tools
First of all, every agent will be given a building block and corresponding floor number(which affects their cost of get in/out of building, noise from the streets).
Each metrics above will be remapped to (-5, 5) domain, and be used to calculate remote/commute decisions based on the persona below:
Negative weights pull the agent towards the decision of remote work, vice versa.
Then the overall remote work rate will be calculated.
Procedural Types
The first, “current” type rebuilds the current massings as close as possible.
Using the PLUTO data, each block’s statistics, such as total residential/commercial&retail/office areas, number of units, number of floors, can be read precisely.
“Home office” consists of “Office”(retail/office space below) and “Home”(residential towers above). The distribution and the density of residential towers, height of the “Office” part are determined by attractors.
Design Space
3 attractor options control the new procedural type:
The closer the block to the attractor is, the denser residential towers are.
3. Analysis and Results
Exploring the Design Space
Live in Scout.
Results
Conclusions and Next Steps
Generally, for lazy people, the current form’s remote working ratio(80~90%) is significantly higher than new types(35~50%).
For the other two types of people, the differences are reversed.
Points to be improved/further studied:
- Persona Weights: tricky numbers to be decided. A deeper study into decision models might be needed to produce more reliable results.
- Are there other procedural types suitable for remote work?
