Recently Microsoft announced that they’re releasing a new XBox which was apparently big news and was reported on at length around the globe. One article in particular on the BBC News website caught my attention because it contained what I thought to be some really bogus reporting
“from first appearances tech writers were more impressed by what they saw than the public posting on social media.
Every time you say Xbox One, you feel like to have to add the word "new" in front of it to avoid confusion. @MurraySwe
The one thing I came away with was that Microsoft has trouble with numbering. @FinleyNathan
Am I the only one who thinks the new Xbox One looks like a VCR? @dandahlberg“
Now perhaps its true that reaction on social media was negative but the BBC’s attempt to justify that claim by cherry-picking some random tweets from a sea of millions is, in my opinion, misleading lazy and pointless. And I said so:
So here’s an idea. Instead of picking random tweets what if the BBC instead attempted to gauge opinion by analysing the overall sentiment of those tweets instead. Its not even that difficult, there are free services such as What does the internet think that do this for you. Here’s a screenshot I took from that service at the same time as I was reading the BBC report:
What does the internet think is doing something smart but ostensibly very simple – it is aggregating tweets in order to give a better reflection of the sentiment toward “xbox”. Now imagine if the BBC reporter that wrote the above article had chosen to measure the sentiment before and after the announcement, would that not have given a better reflection of the reaction from social media rather than a few random tweets? Its certainly more interesting and newsworthy to me, I couldn’t care less about the opinions of individuals that I have never met but measuring the overall sentiment – well that means something. Moreover, I’d like to know if the announcement has affected that sentiment positively or negatively. As any data analyst (or BI guy) will tell you its not the numbers themselves that are important, its whether they are trending up or down that matters more. Now what I would love is a service that showed you the trend of sentiment on social media over a given period of time so that I could look back historically; actually, collecting that data seems like a great use case for a cloud-based ETL tool, I wonder if I could build it with Datasift?
Interestingly in between my writing the bulk of this blog post and then hitting publish on it I stumbled across How The Xbox Reveal Affected Microsoft's Flagging Reputation
on Forbes.com. In it the author, Haydn Shaughnessy, opines:
At the time of the reveal I … said Microsoft had a store of public trust that could help develop the Xbox One market, and should be nurtured across its other brands. Microsoft has managed the Xbox reveal very well, despite criticism the following day and more recently. Optimism and the reputation for innovation have been enhanced
How did Mr Shaughnessy arrive at this conclusion? He did exactly what I suggest above, he aggregated and analysed sentiment from social media:
I arrived at that conclusion by examining big data around Microsoft sentiment. The data is drawn from tens of thousands of news and social media sources and is filtered by “emotion”.
He also used a graph to show the trend over time. As they say, a picture tells a thousand words:
The chart depicts an uptick of optimism toward Microsoft after the XBox One announcement.
Take a read of the article for more insights, it really is very interesting. I’m very impressed with what Mr Shaughnessy and Forbes have done here. They sought out evidence based on sentiment then analysed it to draw a conclusion rather than deferring to perceived popular opinion. The credibility of Forbes has risen in my consciousness (and perhaps with anyone reading this blog post too) and I’ll probably seek out Forbes articles in the future. Certainly in preference to the BBC anyway.