By Andrew Walker

Do you say what you mean in your emails?  Probably not as much as you think…

The words we use in interpersonal communication don’t fully convey the meaning of what is actually being said. This is true of written and spoken language. This remains a huge challenge for social computing applications in business and email is no exception. I recently detected that someone had hacked one of my email accounts. Rather than moan about it, it is an excellent opportunity to discuss how the social context in which we use technology shapes our knowledge, or to put it

in words my hacker might understand "Knock yourself out. You won’t learn anything.” Here’s what I mean.  Language is always partially encoded. This is a social phenomenon. For example: "I love you". If you say that to a parent, it conveys a different meaning from saying it to a lover, a sibling, a friend or someone you’ve just met in a bar.

There are social and personal variables in the communications equation that alter the meaning behind the words, even though the words are the same. As a result computers find it very hard to process the social and emotional content of what we say and do, and sometimes so do people.

Take the concept of Facebook "Likes".  I "Liked" the Starbucks Soy Pumpkin Latte page. People who don’t know me might assume that means I like non-dairy vegetable coffee drinks.  Actually the drink itself sounds horrible, but it used a cool Facebook photo sharing promo, which I “liked” because I’m interested in Facebook marketing campaigns.  My friends would know that about me.

Personalities and relationships aren't explained when you interact through digital channels, so the meaning of our interactions is easily misunderstood.  It is also affected by emotions, for example you might laugh at a joke about a dead parrot one day, but not on the day your pet parrot dies.

When you combine the social vector of different types of relationship with the human emotional filter – you get a system that processes errors into understanding of other people's communications. This tendency to read our own meanings into other people’s words is why doctors don’t like showing medical records to their patients, because the way they write about patients is cold and detached.  It doesn't mean the doctor feels cold and detached towards a patient, that’s just how doctors write records for other doctors to read as functional data sets rather than an interpersonal narrative.

This challenge of capturing meaning from what people we don’t actually know write (or ‘sentiment analysis’ as data miners call it) is why company PR teams worry about letting staff use social media and keep public messages very generic and often a bit boring. We love juicy conversations and use colorful language in private, but in public we tone it down to be safe.

In the world of social media for business, our language is very safe and as a result, the technology doesn't really engage us to it’s full potential… yet. In the future, we'll see public social networks evolve into a multiplicity of private channels where social relationships are more exposed so that we can extract more value from them.

We're already there in some respects.   If you posted family holiday snaps to a LinkedIn group and shared a white paper on custody banking on your teenage daughter’s Facebook page, you would learn that what is normal for one platform is utterly incomprehensible in another and affects what people think about you as a result. This means that, as the multiplicity of business networks evolves, corporations will have to diversify the rules of doing business to allow staff to present different aspects of their personality depending on who the audience is.

Making business communication more informal in this way will make social network applications more effective. We've all had informal client drinks, formal client meetings and experienced long-term client relationships where we behave differently from how we behave with new clients we don’t know every well. There’s a business social network application for each of those relationships and we’ll see workplace processes evolve around them.

Consider the prospect of social networks attached to clients, products and services that only ever contain a few people at a time in a secure, private setting. That's a tool for innovative business development in the era of social computing.

As for the person who hacked my email inbox, what did he/she learn?  The science of statistics suggests not much. Without context can only really get value from sentiment analysis using large samples.  Analyse the linguistic sentiment in 10,000 private emails on the same topic and you’ll learn something about what the people who wrote them think. Read one person’s email and you’ll only ever learn what one person has emailed.

Sorry hacker, but you won’t learn anything about me from an email without knowing my relationship to the recipient and how I was feeling when I wrote it.  I’m a variable, not the equation.