It’s an awkward encounter many have experienced. You ask a person how old they are and they respond: "How old do I look?"
Go on Age Guess, a research project disguised as an addictive game, and you can ask and answer that question all you want—without making anyone uncomfortable.
Once registered for the site, users can upload their photo, and start anonymously guessing away at other people’s ages, collecting points based on how close their guesses are to the correct answer. It’s fun, but it’s also science: Every participant is contributing to the work of two researchers, Dusan Misevic and Uli Steiner, who are investigating whether the difference between how old a person "looks" and their actual age can be used a marker of aging, health, and life expectancy.
Since Age Guess started in 2012, there have been more than 75,000 guesses from 1,632 users in more than 130 countries. On average, people flip through at least 50 photos on the site making guesses, but one power user scrolled through 1,750 images, says Misevic, who is currently a researcher at the Robustness and Evolvability of Life Lab of the French National Institute for Health and Medical Research.
So far, Age Guess hasn’t collected enough data to even begin to give insight into broad scientific questions about aging. But the site's preliminary statistics, shared with Co.Exist, are a fascinating sociology study:
- It is easier to guess a man’s age. The average error for a male photo is 4.8 years, but 5.6 years for photos of women.
- Both genders are equally good at guessing the age of a male.
- But women are better than men at guessing the age of females.
- The highest accuracy of guesses is for photos of people between the age 25 and 45. But the average age of site users is 32, and the average age of people in photos is 31. It could be that people are better at guessing ages close in age to themselves. But, in an early analysis, it looks like that’s not the case.
As with any citizen science project, the trends above should be taken with a grain of salt. Since they are gathered from random people on the Internet, there's no way to control a wide variety of variables, such as the quality of the photo, whether or not a person is smiling, and how closely people are paying attention when they are guessing. However, Steiner and Misevic prefer their method to a more exacting experiment that involves paid college students evaluating generic photos at a computer lab. "People have done that. But this is more fun for everybody involved, and I think there is the potential to generate a lot more data than you could ever in a classical lab setting," says Misevic.
The larger aims of the project are also fascinating. While human life expectancy has dramatically increased over the last 150 years, what's less clear is whether people are living longer and suffering longer, or whether they are truly gaining in quality of life. "Does an 80-year-old always look like an 80-year-old? Or is an 80-year-old today biologically comparable to a 95-year-old 15 years ago, or 20 years ago, or 50 years ago?," asks Steiner. "If I look older, do I have a higher chance of dying? . . . Is the biological clock for some people ticking faster than for others?"
These are all huge questions that Steiner, a scientist at the University of Southern Denmark's Max Planck Center on the Biodemography of Aging, doesn’t pretend to have the answers for. The researchers think their data set could eventually help study people’s perceptions of aging, but for that they’ll have to get people to upload photos of themselves at different ages, and maybe photos of their parents and grandparents—so perceptions of aging can be compared over time.
For users of the site, it can be hard to get feedback on how old we really look. "There were several pictures of me that got overestimated by a couple of years, which didn't exactly make me happy, but is still quite interesting" says Misevic.