A dance with data

Yesterday, Google reminded me that I had a half hour left on my shift.

Along with this it showed me a map highlighting my current location, my destination and my route of choice for getting there – complete with up-to-date traffic information so I could see how long the journey might take.

The thing was, I had never told Google any of this information – nor did I ask for a reminder of my clocking-off time.

I’ve never told Google my work hours, I’ve never told it where I work, I’ve never told it where I live and I’ve never told it my route between those two locations.

That is to say that I’ve never actively told it any of this.

What I have done, however, is switch to an Android phone for a few weeks while my iPhone is out of action.

During the set-up process I casually activated the Google Now feature, and thought no more about it until yesterday’s information.

However in doing so, I had given Google permission to monitor pretty much everything I did with the phone and present me with information based on what it “learned”.

With that in mind I can only assume that – via the GPS on my phone – Google had noticed me moving between these two locations every day of the week – generally at the same times – and “learned” that this was where I work and when I tend to finish up.

There’s no other way it could have figured it out, from what I can think – I don’t put my hours into Google Calendar, nor do I have them detailed in Gmail.

This bit of knowledge was almost certainly based on my movements alone.

I should say that this post isn’t necessarily here to bemoan the menacing hand of modern technology – I had to activate Google Now and should have known better than to be surprised by this result.

However, Google’s prompt has made me better appreciate the trail of information we passively leave behind us every day; well in excess of the obvious flash-points of public Facebook data and the Twitter feed.

It also made me better appreciate just how sophisticated computers are becoming at interpreting data and making fairly-accurate assumptions from there.

After my brush with Google Now’s cognitive reasoning I was about to turn the feature off to avoid a repeat performance, but I’ve since decided to leave it on – for now at least.

I currently using Google’s phone OS (until the iPhone is resurrected, at least), its mail and chat client, its search engine, its mapping software and (in work) its web browser.

I’m equal parts intrigued and terrified to learn what else it figures out about me.

Apple’s US manufacturing bet

Apple’s CEO Tim Cook has revealed that one product line in the company’s Mac range will be manufactured in the US next year, a blip in the decades-long trend that has seen such activity move to Asia (and in particular, China).

On the surface it may seem like another fine piece of PR from the Californian company (and that’s because it is), but the reasoning behind the decision is far more nuanced than that.

While likely to remain relatively low-cost for some time, manufacturing in China is beginning to get more expensive. The middle class is growing and workers are beginning to unionise, which will ultimately lead to demand for higher wages there.

Coupled with this is a rise in distribution costs, which means that getting the finished products out of China and to consumers – more often than not in the US – is gobbling up more of Apple’s (very handsome) profit margin.

Apple’s decision to begin manufacturing a small amount of its products in the US is most definitely a toe-in-the-water tactic, but it is also a defensive manoeuvre that will give the company a head-start if (or indeed when) Chinese manufacturing loses its edge.

Doing it before anyone else does is the icing on the PR department’s cake.

Of course those optics of it are not to be sniffed at either – Apple will now be championed for creating jobs in the US and can soon vaguely state that it has “begun” manufacturing in the country. This is despite the fact that it will be one line in a product category that accounted for just 14% of the company’s revenue in Q3 2012.

Perhaps a more important political consequence, for a cold-eyed businessman like Tim Cook at least, will be the opportunity to sell products to the US Government (which requires that they are manufactured in the US-of-A).

Finally, having a manufacturing base in the US plays into the Apple desire for control.

Having a factory down the road means higher up-front costs but it also means product quality can be more easily monitored, which could reduce faults and – by proxy – returns costs.

It also allows the company to keep a closer eye on factory conditions, thus reducing the chance of Foxxcon-like scandals which are damaging to its coveted brand.

China will remain a manufacturing powerhouse for the foreseeable future, of course. However Apple’s decision is sure to make some people sit up and take notice.

Coincidently, when I interviewed PCH International’s Liam Casey (known as ‘Mr China’ for his work in helping companies get products mass produced in the country) at the 2012 Dublin Web Summit, the topic of rising costs in China came up.

He sees the demographics changing but said the country will ultimately remain the place to manufacture because the expertise, speed and scale to do so is there.

Despite today’s revelation it can be assumed that even Tim Cook agrees with this, at least as long as the iPhone, iPad – and the majority of Macs – are getting churned out from Chinese production lines.

Obama’s data machine and the future of political campaigns

A lot has been said about how data was used to predict the result of the 2012 US election.

A lot has also been made of the Obama campaign’s ability to use raw data (and online platforms) to its advantage.

This article, however, is the first time I’ve seen some real detail on exactly what that data machine looked like and what it achieved.

It’s a long piece – and it spends a lot of time looking at the personalities behind the programming – but for some really eye-opening stats skip down to the ‘what they actually built’ section on page 2.

Some choice facts and figures from there -

  • 30,000 Reddit users registered to vote through Obama’s site after he did an AMA there
  • The campaign used data from DVR boxes (akin to a Sky+ box) to target ads at viewers who had watched certain things, for example the TV debates
  • They designed a Twitter tool that could calculate a user’s influence, cross-reference it with other data (like if they were in a battleground state) and target DMs accordingly
  • Facebook fans were told if friends certain hadn’t voted and encouraged to get them to do so
  • The campaign created a ‘quick donate’ function similar to make repeat donations as easy as buying from Amazon or iTunes