I’m moving on from my current job, taking a break before starting a new role in January.

This is my final day, so after the frantic period of activity running up to this week’s staff awards event I’ve settled down to write my handover notes.

Distilling two years’ work into a few pages is proving quite difficult. What’s struck me most is the frequency with which I’ve suggested my replacement “speak to so-and-so” to get a particular task done.

My email account will be closed and eventually deleted after I leave. That means the many detailed, lengthy and sometimes just plain weird discussions I’ve had with colleagues will vanish into the ether, just as the results of face-to-face conversations I’ve had will leave when I do.

This all underscores the value of human memory. I had no handover notes at all when I started here, so learning how to get even simple tasks done was a long and complicated process.

As people leave their employers they take with them detailed knowledge of people and processes, built up over years or even decades. While replacement staff may be easier to find in the current job market, their knowledge of the organisation will take much longer to develop.

Employers, as well as new employees, would benefit from finding improved ways to capture this organisational memory.

Internal social networking can enable that inter-generational transfer of knowledge between new employees and old-timers.

It needn’t be technologically complex, though. At an event I attended earlier this year, Euan Semple spoke about talk.gateway, the bulletin board he introduced at the BBC.

“Staff members shared more information outside the organisation and with people in other countries than they did with each other. We had to give them an infrastructure or mechanism to talk to each other online,” he says. “I wanted to introduce social computing tools on the intranet and started with a bulletin board.”

talk.gateway allowed staff to ask questions, find solutions and connect with each other. Crucially, though, it’s archived and searchable, which means discussions can be viewed even after the people involved in it have moved on.

More and more organisations are introducing internal Facebook-style social networking, including some in the public sector. Carl Haggerty’s innovative internal social networking pilot in Devon Country Council led to a sharp decrease in helpdesk calls, as employees solve problems by using each other’s knowledge.

Networks like this also enable newer employees to ask questions of and learn from longer-serving ones, helping people settle in and get up to speed with the job.

My (as yet unappointed) successor will have to make do with twelve pages detailing my key processes and projects. I wish them well, and look forward to the next challenge – watch this space!

As I watched, I tweeted my thoughts, which became part of the huge stream on the #bbcqt hashtag. Looking at the hashtag search, you’d easily come to the conclusion everyone thought Griffin came across very badly.  Tweetminster reported that 99.9% of tweets were negative about Griffin.

Yet 24 hours later, a poll for YouGov found support for the BNP had increased following the show.

So why was the Twitter barometer of social attitudes wrong?

Quite simply, that’s because Twitter hashtags only tell us what people on Twitter think about something.

The comments on the BBC’s own Have Your Say forum, or submitted by viewers hitting the Red Button on cable or satelite had a far less critical view of Griffin’s performance, with a sizable number saying they agreed with his views.

So what this demonstrates is that what people say on Twitter should not be taken as what people in general think or feel.

There’s been a lot of discussion lately on the value of real-time search, with social media monitoring services selling sentiment analysis as an accurate method of understanding what people think.

This isn’t strictly true. It merely tells you what people on Twitter think. People on MySpace might think differently, and people not on the internet at all might have different attitudes altogether.

This is what ethnographer danah boyd (she does not capitalise her name) describes as the Not So Hidden Politics of Class Online:

‘For decades, we’ve assumed that inequality in relation to technology has everything to do with “access” and that if we fix the access problem, all will be fine. This is the grand narrative of concepts like the “digital divide.” Yet, increasingly, we’re seeing people with similar levels of access engage in fundamentally different ways. And we’re seeing a social media landscape where participation “choice” leads to a digital reproduction of social divisions.’

So for instance, boyd found that while discussions about social media tended to focus on Facebook, as this was the platform used by social commentators themselves, at least as many young people were using MySpace.

Our choice of social network, boyd argues,  isn’t about features or functionality. It’s a result of what sociologists refer to as homophily, the social phenomenon which means we choose to socialise with people like ourselves.

Because of homophily, the platform on which we choose to socialise online is inextricably linked with factors such as race, education and socio-economic status. This is reflected in the stats, which consistently show Twitter users are older, wealthier and better educated than people participating elsewhere online.

The phenomenon of homophily is evident not just in choice of platform, but who you interact with once using that platform. In following people who are interested in the same things as us on Twitter, we inevitably choose to follow people who are quite a lot like ourselves.

This means it can be something of an echo chamber, with views and opinions like our own reflected back at us.

This makes it inherently unreliable as a social barometer. It only reflects a certain strata of society, while other platforms may vary from this considerably. Let’s not forget that 10m adults the UK are not online at all. They have views (and votes) too.

Those of us working in engagement (as well as lazy journalists) would do well to remember that the views that echo through our own Twitter streams do not neccessarily represent everyone.