Can AI help trace the identical residential buildings in the post-war housing stock?

The documentary Identical Strangers (2018) tells the deeply tragic story of a social science experiment in post-war New York in which newborn triplets were separated and placed for adoption with poor or rich families. In the documentary, we see how the triplets encountered each other "in the wild" - and from there unravels the web of the social experiment they were subjected to. The brothers immediately become a media sensation, as they share many similarities despite their very different upbringings. They are strangers to each other, yet somehow identical. They are identical strangers.

A lot less macabre is the similar situation in the post-war housing stock. A considerable amount of houses were built between about 1965 and 1990. This was a time of enormous housing shortage, but also a time of a new industrial way of building. Parts of the houses were made in factories and assembled on the building site. This enabled cost-efficient construction and reduced the need for skilled labour. Architectural firms worked closely with construction companies and contractors to build this way. As a result, many houses look alike, both inside and out. But who built where and when is largely unknown. Hypothetically, hundreds of thousands of houses could be 'identical strangers'.

The potential of this undiscovered "alternative data" is huge. With Identical Strangers, we deploy AI to track identical homes. To this end, we work along three lines:

• Build an alternative dataset of similar dwellings through systematic empirical research Google Streetview.
• Explore the building culture and practice of this period through a series of interviews of the architectural and contractor firm partners of this period.
• Develop various algorithmic strategies to understand which approach helps discover identical unknowns.
• Explore potential use cases for sustainability, circularity and cultural heritage.