“If you can’t trust yourself, who can you trust?” This is the question asked by Digital Me, an online experience developed by Helios Design Labs and launched in October 2015 through BBC Taster. To date, 12,101 people have tried the experience, and 1,494 have shared it on social media. The concept behind Digital Me is that individuals sometimes have online personas that are different from their actual personalities. By pulling information from the user’s Facebook and Twitter accounts, as well as determining location based on IP address, an artificial bot pretending to be the digital version of the user has a conversation with the actual user. This is an interesting concept, but technical limitations and some design shortcomings severely stunt the execution’s effectiveness.
The experience starts with the following message:
It’s hard to know which way of meeting me “privately, one on one” makes me most uncomfortable. Certainly sending my face to some anonymous entity through my webcam is disconcerting, but giving it access to my Facebook and Twitter accounts is almost equally intimate. Years ago, I tried an experience called Take This Lollipop, which takes information from your Facebook profile like your photos, locations, and relationships and embeds them in a video in which a creepy, dishevelled man stalks you online. It’s absolutely terrifying.
Despite my flashbacks to Take This Lollipop, I press on. This “digital me” communicates through an instant message-like interface. It goes on to tell me that it knows some things about me, but that I first must turn on my webcam before it tells me. I say yes, and I see myself on the screen. On a second run-through of the experience, I say no. “Digital me” refuses to continue the experience until I turn on my webcam. Having used the webcam on the first run-through of the experience, I was surprised that the webcam is required, as it did not have much of an impact.
The experience then asks me to connect to my Facebook and Twitter accounts. I link Twitter first, and it has some difficulties. The Digital Me tells me that his “Best Follower Forever” is Maggie Knight, one of my friends from high school. This designation is given because Maggie has mentioned me in her tweets twenty-four times. This appears significant to Digital Me, but in reality, Maggie just uses many mentions for many people in her tweets. A stronger algorithm would have detected this, and would have been able to identify more important mentions.
Other parts of the experience work better. I especially liked when Digital Me told me that he liked “food, good focus, and London,” as these are things I definitely can see myself having tweeted about. More confusingly, Digital Me told me that Martin O’Malley, Donald Trump, and Netflix & Office (I’m assuming it tried to say “The Office”) make him unhappy. Again, this is the sign of a bad algorithm. I love one of those things, have neutral feelings toward one of them, and am only unhappy about one of the three…it shouldn’t be difficult for my readers to figure out which on their own.
Linking my Facebook account also had issues. Digital Me told me, “I’m really interested in I’m not sure what I talk about,” which, with that garbled syntax, is actually quite believable. He also tells me that he wishes he could visit London, which does not make sense because I have been posting to these social media accounts from London for the past two month, made more damning by the fact that Digital Me is currently tracking my location based on my IP address. Maybe Digital Me wishes he was a physical entity so that he could be in London in this way, but that seems much loftier than what Digital Me is trying to get across.
Looking back, the parts of the experience that were most interesting to me had nothing to do with the Digital Me character. Instead, I poured over the parts that compared my online persona with others. I was fascinated by the sentiment analysis of my Twitter stream, which revealed that it is less positive than most people’s. Similarly, I was surprised to learn that I have an “astounding” number of friends compared to other Facebook users. This information was more interesting to me than any of the random, often incoherent pulls from my feed.
The creators of Digital Me stated that the purpose of the project was for people to learn about themselves in an age where “our digital personalities are now as real as our other personalities.” This is an important goal, but the many glitches within the experience made the Digital Me feel less like my digital presence and more like a poorly made chatbot. It reminded me of a mix of social media integration and Oliverbot, a website popular in the early 2000s that used learned behaviour to hold conversations with users.
Further investigation revealed that Sandra Gaudenzi, one of the co-creators of Digital Me, was inspired by Take This Lollipop (the creepy experienced that Digital Me initially remind me of). Her blog post mentions that the team spent six months researching APIs and data mining protocols, which makes it extremely disappointing that these are the elements that detracted most from the experience. I agree with the creators that our digital presence is rapidly becoming as important as who we are in real life, but the many problems with Digital Me only reminded me that real life still has far less glitches.