Numbers

 

At university (distressingly some twenty-odd years ago now), I quite quickly realised that I was a qualitative, rather than quantitative, kind of guy. Oh yes, I knew how to live the student dream.

To explain – I studied Sociology. And in the realm of the social sciences there is a long-standing debate about whether they are a science or not. The obvious answer to just about anyone is “not”, the exception to that being those qualitative types of people who believe that anything and everything can be boiled down to numbers.

If you read these pages more than occasionally, you’ll have realised that this cult of measurement is a particularl hobby horse of mine, one of the thoroughbreds from the Ballantine Stables. When it comes to people, measuring things tends to lead to some very odd, and often dangerous, side effects.

All of this is constantly playing in the work that I do, especially when I see that much of the power of technology we have at our fingertips seems to be increasingly described in terms of access to data (big or otherwise), and when people talk about data they are inevitably talking about things numeric. But here’s something – maybe it’s the qualitative stuff that is actually changing the world around us.

Take, for instance, the transparency under which  most institutions now find themselves. Want to know what it’s like to work at a company? Well, even notoriously secretive organisations like Apple now find themselves open to scrutiny through services like Glassdoor (see the near 2,500 employee reviews here: http://www.glassdoor.co.uk/Reviews/Apple-Reviews-E1138.htm).

Now with services like Glassdoor, or recommendation engines in trading platforms like Amazon, whilst there are headline numbers (usually out of five or ten), the interesting and qualifying information comes in the form of the prose text. And you can usually get a feel about how valuable an individual’s scoring is based on what they have written too.

Now there is a whole world of automation trying to plough a furrow through the world of qualitative analysis – look for the words Sentiment Analysis and you’ll find the types of product I mean. But machines aren’t very good at this kind of stuff because, ultimately, reducing qualitative data down to numbers is just forcing quantitative techniques into a domain it doesn’t belong.

There are two key reasons for this: firstly machines still get troubled by the context in which prose appears (computers have an irony-bypass, for example). See Matt Baxter-Reynolds’ recent foray into the sentiment analysis into understanding whether ChromeBooks are a good thing or not: http://www.zdnet.com/does-sentiment-analysis-tell-us-whether-chromebooks-are-a-good-idea-7000023101/)

But secondly, good qualitative analysis is a learning process, not a scientific hypothesis/test processes (which is at the core of sentiment analysis from my experience). As a researcher, as you read, you learn, and that takes you down new paths appropriate to your understanding. Think about how you use the web, how you surf, and how you make decisions and leaps based on the content you consume.

That everyone now has the ability (if they so choose) to contribute to that content is the fascinating power of the internet, social networks and our increasingly connected world. It’s also why it has such profound consequences for so many facets of our modern lives.

 

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s