Summarize this content to 2000 words in 6 paragraphs in Arabic “Memory glimmers and hints but shows nothing sharply or clearly,” wrote Janet Malcolm. The photographer and journalist spent her career dealing in facts about others. Writing her autobiography, she found her recollections to be slippery. I thought of Malcolm’s line when I made an umpteenth attempt to bring order to boxes full of thousands of family photographs, which have landed like wreckage in my attic.The collection is an amalgam of those of my parents, grandparents, great-grandparents and great-great grandparents. It has also absorbed the collections of assorted childless relatives. They span more than a century and end around the millennium.All those lives in a large wooden chest and several jumbo cardboard boxes, taking up precious London space. But a home is not only a repository for new stuff. Family history deserves some room — especially when that history is unresolved. Before I can make sense of the collection, I must identify the people in the photographs. Can I work out who they are? And if I do, should I keep the photographs for my children? Or should I put my faith in the future, make digital scans and chuck all but the most precious ones away? I start by trying to build a chronology. This is time consuming because the boxes are a mess. Victorians and Edwardians are jumbled up with flappers at parties, second world war soldiers in barracks, rows of bridesmaids in mini skirts, mourners in cemeteries wearing flares and babies of all eras. Most of the photographs were taken before 1950. They are mostly of people I don’t recognise. But I filter out my parents, obviously, and I recognise a few older family members from about 1950 onwards, two decades before I was born — or I think I do. My aunt developed memory problems, which must have been when she wrote ‘Jennifer’ on the back of nearly every baby photograph taken between 1910 and 1980 They are younger editions of themselves. Though come to think of it, perhaps some are my grandparents’ siblings. Everyone starts to look like everyone else. I am unreliable. Some photographs have writing on the back identifying the people in shot. But the captions are not always to be believed. My great aunt loved everyone but favoured my cousin Jennifer, who was born in 1971. In later life, my aunt developed memory problems, which must have been when she wrote “JENNIFER” in trembling hand on the back of nearly every baby photograph taken between 1910 and 1980. In the absence of sound evidence, could technology help? Facial recognition is used in criminal investigations. Surely it could identify my ancestors. Not any time soon, says Kurt Luther, an associate professor of computer science at Virginia Tech. Civil War Photo Sleuth, Luther’s online project, uses FR to help identify subjects in US civil war-era portrait photography — firmly outside living memory. It is one of the most robust experiments in FR to date, with a data set of thousands of portraits from public archives and private collections to help make matches. But the process requires more than technology. For a start, FR is riddled with gender and racial bias, though Luther says that is improving. “[The project] was always conceived as a collaboration between human expertise and AI,” says Luther. Even with a vast data set, FR is only “a powerful tool for narrowing down tens of thousands of possibilities”.Luther says FR measures landmarks on faces in old photographs: the tip of the nose, the corner of the mouth and so on. But landmarks only lead to a shortlist of candidates.“We also have people examining photos for visual clues: what uniform a soldier is wearing, what photographer information is available, how old the paper is, all this additional context,” says Luther. “AI never identifies everything.” He says he cannot foresee a time when it will.In other words, facts are still primary. And relatives are not the best people to make identifications. “We’re so invested in our ancestors,” says Luther. “Someone more neutral can give a better perspective.”Should I keep them for my children? Or should I put my faith in the future, make digital scans and chuck all but the most precious ones away? Some commercial ancestry sites already offer FR, albeit in a limited way. Tina LaFreniere founded Related Faces in 2020, a US-based website that helps her customers identify ancestors using technology. But she stresses it can only suggest pairings and connect users with others who may have more information. The more people use it, the more useful it will become. “I tell people not to throw away their photos,” says LaFreniere. “Box them, preserve them, put them away. Look at what tech has done in the past 10 years, from flatbed scanners to a phone camera with millions of pixels — you don’t know what it is going to bring tomorrow.”I imagine a future international ancestor-matching service powered by next-level FR and overseen by a neutral authenticator. So I am labelling photographs by hand with whatever information I am certain about. In the meantime, what if I want my attic back? Relying on scanned images would be risky: technology becomes corrupted and obsolete, files get lost, people move on. “If you don’t have heirs, [photographs] still have historical value,” says LaFreniere. “They tell the story of a place, not just people.” If I want to offload them I could offer them to local history associations or people’s archives. But I am keeping the collection. When I find a photograph of strangers or if my memory is unclear, I leave the backs blank. They are mysteries to be solved another time. Find out about our latest stories first — follow @FTProperty on X or @ft_houseandhome on Instagram
