To assess the truth(iness) of a message, one of the most useful pieces of information is who is the source? Is it someone you trust? Distrust? Never heard of? What personal stake does the messenger have in persuading you? The claim that taking mystery tablet X will give you a glowing complexion and concentration to match is most trustworthy when told to you by your close friend (who gains from doing you a favor), and least when told by a tablet X salesperson (who gains commission on your purchase and loses nothing from your disappointment with the product) or your practical joker colleague (who gains if you have an unfortunate reaction).

Online Identity and identification are topics of perpetual controversy. Most recently aired in the “real names” issue, via Google, Facebook, etc. The argument for insisting on real names is that people act more responsibly when their words are identified. The arguments against include the safety of dissidents and whistleblowers and the general privacy of anyone who, say, wishes to review their experience with breath mints without having their halitosis problems featured in their increasingly prominent online timeline. Another argument against is that for the most part, knowing that a posting was written by “Sam Smith” tells me just about nothing about it. If I want a system that helps me understanding the context of the message and assess its veracity, attaching a decontextualized name to it just doesn’t do it.

A name is useful as context when it is an index into greater knowledge about the person. As such, it needs to persist in time, but it need not be a “real name”, tied to a real world identity. A persistent pseudonym can serve this same function, with the significant privacy advantage that its reach is bounded. One can create a long-term history as a credible reviewer of embarrassing personal products or dangerously dissident political positions, safely walled off from your day to day persona.

For the reader, the person attempting to assess the veracity of the rumors, gossip and early-breaking stories that flow through their news feeds, what is needed is a better way to see this history, to be able to, at a glance, look at a source and get a feeling for their trustworthiness.

In the physical world, we do this with all sorts of cues: a person’s gaze, their clothes, the color of their skin (aka profiling, another quite controversial issue that we will simply elide in this posting. But for further reading see for example Leslie Zebrowitz on face perception or Ambady and Rosenthal on quick personality judgments or the ACLU on racial profiling.

But online, there is no body. History, instead, is the virtual body. And one that has the advantage that the cues that are read from it about you are based on your actions, not your skin color or gender. But it is not a body that is easily seen. And that is where we can make a big difference, improve our ability to distinguish truth from truthiness. By visualizing one’s online history, we make the person – the source of some new information – perceivable. It can be words written pseudonymously – enough such words have a heft and complexity of patterns to give that identity life. And, for this discussion, lets limit it to words and interactions produced in a public context, meant to reach a broad audience, and understood to be permanent, searchable, graphable, etc.

Let’s use twitter as an example. How can we create a portrait of Twitter users so as to have a better sense of who they are – what is their worldview, their underlying agenda, their similarity to me? One might start with a picture of the user showing what she talks about and the temporal pattern of her postings (see Alex Dragulescu’s Twitter portraits as an example) . And let’s surround this with a depiction of who she follows and who follow her. We can make a horizon line so it shows the ratio of followers to followees. Those she follows, we’ll show a visualization of representative words and phrases, for this is the information that informs her, at least in this context. Those who follow her are the amplification of her voice. We want to get an understanding of her reach. Are her posting read (does she have followers who follow few people beside her?) and are they amplified – does she have followers with big followings? Strongest are followers who follow few but have a big following. Here to, we’d like to see if she is quote, replied to, retweeted – and what she herself repeats and retweets. How interwoven is she with this scene? Has she often retweeted posts from banned accounts?

I’d like to see such portraits both at a large scale, to peruse an individual, and also as thumbnail sketches, highlighting key patterns. What is the signature pattern of spam? Of bots? Of ordinary posters, journalists,marketers, celebrities, etc.? (see this paper for some work in this direction, from back in the heyday of MySpace)

There are excellent cues in the online space to help us assess trustworthiness, but often they are hard to perceive. Along with portraying the participants, maps of information flow would be an excellent tool. Fantastic work is being done in this area by Gilad Lotan, e.g. on the tweet that broke the story of Bin Laden’s death and the one that alerted almost no one about Whitney Houston’s.)