BYOI (Bring You Own Identity): Tomatoes Tomatos

My computer thinks I’m gay
I threw that piece of junk away
On the Champs-Elysées
As I was walking home
This is my last communique
Down the superhighway
All that I have left to say in a single tome
I got too many friends
Too many people that I’ll never meet
And I’ll never be there for

– Placebo – Too Many Friends

It’s a couple of days after #Brexit , a little bit more after Italian national roundup on local administration vote and some other speculation that makes prediction made by analyst run in a corner of a room covered by shame for their total fail in predict results.

It’s a couple of days since we discovered that Facebook is use phone data to better understand your behavior and which friends, group, topics should suggest to you, even if you are not a Facebook member ‘cause even  simply applying the six degrees of separation ( this analysis payoff when your user base is around 1 billion of connected users at the same time everyday.


In my previous posts I’ve explained my point of view on why some complex projects like those who include identity sometimes fail and some other time are successful, I’ve also introduced the concept of clustered identities in homophily groups who may help you to predict certain behaviors in your ecosystem.

Still some days ago I had a (as usual) great conversation with a colleague/friend of mine Jonathan West who beside being a great leader in marketing share with me the passion of observing people behavior in order to better understand them.

The conversation was, as usual around the concept of relationships within identities and how those may be somehow influenced by common traits. Think about it for a moment, we define identity relationship as a way to describe the link (relation) between a set of identities in a form of 1 to 1 or 1 to many or even more many to many interconnection, where the single identity may define the relationship with one or more identities transferring the relationship itself of triggering other relationships based on a specific status or different relation.

It’s about the way we interact with our other digital identities defining not a specific “alter ego” of our persona but a defined state of the value of the digital identity correlated to our persona and the context (even temporary) where those identities are used.

Now if this is the starting point we should start to walk back to the root of the problem:

how a relationship between a set of identities is set?

That could be probably explained mainly by the context where those identities are used or by the final expected outcome of the relationship should be but if we dig a little bit more into it the question may assume this semblance:

Why certain individuals tend  more likely  to create specific relationships with certain set of identities than others?

Let me rephrase the Placebo lyrics:

My computer thinks I’m gay
I threw that piece of junk away

I got too many friends
Too many people that I’ll never meet
And I’ll never be there for

– Placebo – Too Many Friends

It’s a wonderful and powerful concept, a thing who define me based on whatever sort of analysis and no..I’m not meaning AI necessary.  My persona unable to offer a distinction between the physical me and the digital social identity. It’s the relationship between being me,as in physical myself and the digital identities that my computer analyze daily that define who am I in the end.

I may recognize myself in it or not, I may refuse it or not but this not depends from the data the computer “reads” but from who I am (something that a computer still miss to understand apparently … ).  Now read it again, what I see is a form of clustered  grouping of my digital identities where the machine simply “read” the daily behavior and try to match it with other similar identities/relationships.

A relationship born, live and “die” based on the context… context is defined by the need to reach a specific “goal” no matter if it is related to a better UX experience, a pre imposed rule (i.e.: you may login on the app only through a specific social identity) or any other factor (i.e.: company force you to use specific set of identities to use specific features/apps).  But…you know in my post there is always a but.. a relationship is influenced also from external factors (i.e. the “external” identity used may trigger other relationships) or it is subjected to change because of other relationships status (i.e.: I decide to close a social network identity or simply I prefer a Facebook to Twitter).

It’s not about the fact that “the computer think I’m gay”  but if this information is relevant to better predict my behavior.

What makes me different or equal from other identities?

It’s a matter of grouping the relationship. Take your identity solution and look at the various relation defined by the various accounts you have. Why those accounts are there? How did you asked for them? Why you asked for them? Are all still relevant or some of them are simply the result of a sum or imperfect processes who have assigned you identities that are not relevant to your job anymore?

Now think how you interact with these relationships. Do you do regular “house cleaning” and make it so that those relationships are in a “clear and active” state or you simply relay on the fact that “ soon or later” will come handy again?

Now for those who answered “yes” to the “regular house cleaning” question: do you do for all your identities? Yes I mean also the social identities you use as federated logins, those that have a certain number of apps with “permission” to read data out of them…eventually even which are the other federated apps or some “useful” context out of the use of the  primary social login identity.

The relationship is about a M:N combination that start from an inner relation and extend itself based on context (temporary status), need (contingent to the context of use) in other way based on the expected behavior of the user and guess what? User or better personas are somehow predictable if we are enough precise in defined them. How? What if we group in  large anonymous collection of similar identities (digital identities).

Wait no I am not meaning collection PII information but behaviors, I need to profile groups not collect their attributes and metadata’s. I need to observe the relationships  based on how those relationship evolved. Think at a onion model where the inner circles are the personal identities I use as personal accounts, now expand the onion circle to those accounts who have a relation with the inner circle one and I use to access other apps  and go on like this. Now add a direction to the circles that describe the way the relationship occurs (i.e.: the social login is mandatory to access the enterprise app vs. to access the enterprise app I decide to relay on the social login). The direction give you a specific set of behaviors that are driven, for example, from your personal perception toward security, UX,context of use, etc..

So we defined the relationship type and a set of context where the relationship may occur, if we adds to this a set of  profiling attributes (age,job role,location,etc..) we may obtain a cluster definition. Now if we would be “so kind” to share these result in a global hub we would obtain a real specific set of results who may be quite precise in predicting a certain behavior.

Like…if we have 1 billion connected users and we spoof devices (i.e. phones,laptops,etc..) and eventually we log the apps federated, the logins hours, location, what has been published etc.. we would probably able to define your relationships even before those will happen…

Placebo – Too Many Friends