Big Data Bang Goes the Theory

Like it or not, our world is driven by computers.

The details of your life, my life, and the world we live in

are being recorded and kept in vast stores as digital information.

And now, a new generation of technology

is analysing our data in ways that are already changing our lives.

We now live in a world of big data. Computers talking to each other,

sharing our information in ways we never believed possible,

sending out a stream of 1s and 0s.

So tonight, on Bang,

we find out exactly what Big Data is and what it's really

capable of doing, the good and the not so good of this brave new world.

Maggie will be looking at the frightening things people can do

with personal data.

And how some of us might be leaving ourselves vulnerable to crime online.

This is about someone else being really careless with your data.

Kind of a shocking thing to see.

I will be looking at how big data technology

can improve our lives...

From getting you to your holiday destination safely...

..to helping us to save lives.

This is really going to make a huge difference, isn't it?

And Jem will take us back to basics of what data actually is

and how it's become so powerful.

That's Bang, on data - and the new digital revolution.

For most of us, data means the digital information

that we personally use every day.

But there is another kind of data that we rely on without even

thinking about it.

For example, every time we take to the skies.

We're used to seeing streams of vapour left in the wake of planes,

but as we have been hearing in the news, they can also leave a data trail -

a stream of data that monitors the plane's performance.

At the headquarters of Rolls-Royce in Derby, engineers make nearly

half the world's passenger jet engines, including this,

the Trent 1000 - the engine that powers

many of our transatlantic flights.

The temperatures in the back of the engine are staggeringly high -

we talk about the temperature

as being half of the temperature of the surface of the sun,

and in fact it's 200 degrees above the melting point

of the metals that we use.

The only reason they don't melt is that we pass cooling air

through special passages

and channels that keeps the gas away from touching the metal.

The engine is full of vital components -

all engineered with absolute precision,

including an on-board computer.

This relatively unassuming box is the brains of the engine.

Not only does it control it,

but it also performs another crucial function.

It receives data from sensors buried deep within the engine,

measuring 40 parameters 40 times a second including temperatures,

pressures and turbine speeds.

All of the measurements received by the computer are stored,

and then streamed via satellite back to base, here in Derby.

And that's not just true for the Trent 1000.

It's the same for the entire fleet - that's thousands of engines.

A Rolls-Royce-powered engine takes off or lands every two

and a half seconds, somewhere in the world.

That's a very cool factoid.

And wherever they are up in the air, there's information coming back

about the functionality of this engine back to Derby, to here.

Absolutely, and they're constantly monitored

using clever data analytics that are looking for anything

going wrong in the engine, or any sign that it might need to be

serviced early or something like that.

So wherever you're flying to,

while you're 30,000 feet or higher,

thousands of streams of data are constantly sent back to base.

Here, computers are programmed to sift through it

for any anomalies.

So you've got 11 engines that have flagged something up on your system.

That's not necessarily an emergency

but something that needs to be looked at. Is that how it works?

Just an example - this has been flagged due to that -

a step change in the behaviour of the oil pressure parameter

of around ten PSI.

So when we start to see something that is not what

we would expect to see, that's the first trigger point.

-And that's when you guys step in.

-Absolutely.

Analysts here take a closer look at any problem data,

and get on the phone to the airlines immediately. The result?

Technical faults are dealt with before they become a major

problem, preventing delays.

Plus, the working life of the engine is dramatically improved.

One of these engines will fly around the world 450 times before it

needs to be overhauled - that's a hell of a mileage.

And just as big data works for plane engines,

it can also work for healthcare and the human body.

We've all seen computerised systems in our hospitals.

In intensive care, vital signs

need to be monitored frequently at the bedside.

Now traditionally, this information is noted down on paper,

but here at King's College Hospital,

a new technique is being trialled

that records all this information and important new data

in a way that could mean the difference

between life and death for a huge number of people.

Brain injuries are the most common cause of death

and disability in young people.

In UK hospitals, we treat over

220,000 patients with them every year.

Jordan Ball was admitted to King's after a motorbike accident last year.

His recovery is as rare as it is remarkable.

When I woke up, my mum and my brother came to see me.

Only one eye was open, and I was just staring at my brother.

-And a tear came down my eye.

-Don't. You will make me cry.

And he was like, "Mum, he knows it's me! He knows it's me!"

90% of people with his injury don't even wake up.

The 10% of people that do wake up, they need help with

eating, walking, they don't tend to recover like Jordan has.

He is on the top end of recovery.

-Do you realise how lucky you are, Jordan?

-I do.

-He does do, don't you?

-It's incredible.

Most patients are a lot less lucky - and for them,

some of the most serious problems can occur in the following hours

and days after admission to hospital.

Secondary brain injuries are a serious concern in patients

that suffer trauma to the brain.

If you imagine that this is the initial point of injury...

one of the biggest problems

is that the electrical activity in the tissue surrounding this point

short-circuits, and storms through all the cells,

using up all the energy supply, the glucose.

Eventually, the cells stop working and they die, never to be replaced.

If it was possible to know when these secondary events were starting

to happen, doctors could intervene and potentially limit the damage.

When Jordan was in ICU, he suffered seizures

that the doctors understood very little about.

So obviously, the doctors were doing everything they possibly could

but there was a big part of all of this that was an unknown.

If there was anything that could detect brain activity

and that doctors were then able to make a move to stop that happening,

that would have been wonderful.

King's College Hospital have been working with

Professor Martyn Boutelle

and his team at Imperial College on a big data early warning system.

This bolt can be fitted to the skull by a neurosurgeon without

even having to go into theatre,

turning what's going on inside the brain into data.

You can see, sticking out, we have a number of different sensors.

One of them the measures brain electrical activity.

Another one measures the pressure and tissue oxygen

and brain temperature as well.

And then the last one measures chemically

what is going on in the brain tissue.

So if all of the indications from a probe like this

-are letting you know that there is trouble ahead...

-Yes.

That gives you a chance to act before the secondary brain

injury really takes effect?

Exactly. That is the idea.

This data could produce vital new insights,

but it's recording between 16 and 32 channels,

each being measured up to 200 times a second.

Very quickly you can't see what's going on

when there's that much data.

Doctors in ICU need an automated solution to turn all this

available data into something immediately useful.

So Professor Boutelle turned to Cybula - big data specialists

who also worked on the engine monitoring systems at Rolls-Royce.

So, obviously,

an engine is completely different to a person's heart or a person's brain,

-but you're using the same programme to pinpoint problems in each.

-Yes.

-How does that work?

-Exactly.

Because it's doesn't matter to us

whether it's brain data or whether it's an aero engine.

It is really just the data that's the issue.

We are able to look at the patterns

in the data that characterise those events

through those shapes.

Here is an example of a brain event that we are actually looking for.

In this section here, we are looking

for this kind of spreading kind of wave.

The liquid goes into here...

So, with a big data solution at its heart, this prototype brain

monitoring system works in near-real-time.

Importantly for the busy critical care staff,

they could see that something was happening here so they can see,

"Yes, it has started to happen. We need to do something."

Big data is being used in flood alert, transport, natural disaster

response systems...

You name it, it can provide vital new information.

And always as a collaboration between research groups,

engineers and experts in data collection and analysis.

The big data revolution isn't just about storing more

and more unconnected information.

It's also about programmers

designing new software to spot patterns and make connections in the

data - and for that they need access to the data in the first place.

Many believe that if we could make more data

freely and openly available,

then we could crack problems in ways that were previously unimaginable.

The Open Data Institute is encouraging businesses

and the UK government to share more of their information.

If you make this data available,

it could be used to show more transparently what's

going on, it can make people accountable

for the performance of, for example, public services, health education...

So give us a few examples of the difference that can be made.

The data that was held by the Department of Transport

on bicycle accidents - within three days, that data had been taken,

turned from one form into another and somebody had written

an application that basically avoided

the bicycle accident black spots around London.

Just a very obvious thing that never occurred to the people

that had the data in the first place to do.

Sharing data clearly has some advantages but when it comes

to personal information, it can be very controversial.

Unless they opt out, people in England will soon

have their medical records put into a digital database,

and the NHS there plan to make them available for research.

It could lead to medical breakthroughs - but some

people still worry about releasing sensitive information like this.

If I'm undergoing a medical crisis,

I'd really like that my medical records could be shared

between the appropriate services in an appropriate way,

but blanket data publication, you have to be very

cautious about...in this area when it's relating to individual data.

And I think that both corporations

and governments have to be extremely careful to respect

and maintain the privacy of an individual.

The NHS say that details that could identify individuals will be

removed before the information is made available.

But medical records aren't the only sensitive data

we have to be conscious of.

Today, it seems everyone wants to know what you're doing.

Take a train or a bus and your journey is tracked.

Use a points card when you're out shopping,

and they keep track of what you're spending

and use that information to market yet more stuff to you.

And your bank account

and spending patterns are monitored by financial services.

Now, that's for your protection to prevent fraud,

but it's also for future credit checks.

It doesn't stop even when we think we're having some time to ourselves.

We now spend nearly half our waking hours in front of some screen

or other - but it's not always just between you and the computer.

Take e-mail, for example - now,

if you use a free service like Yahoo or Gmail, you'll be very

familiar with those targeted ads that appear on the page.

So have a look at this - this account belongs to

a member of our production team,

and you'll see that the ad that's appeared

is offering flights to Australia.

No surprise, because he uses this e-mail address

to book most of his travel.

Some free services automatically scan your search queries,

social networks and even e-mails to get a sense of who you are,

so they can target their adverts better.

Sometimes,

when a service is being offered for free, WE are what's being sold.

In this case, as a potential customer for a targeted ad.

This is what we sign up to in return for free communication,

map services and a world of knowledge at our fingertips.

Internet giants like Google, Facebook, Yahoo

and Twitter don't release our personal details directly to

advertisers, but they do generate an income from our profile.

So how bothered are we, really, about sharing this sort of data?

Let's find out from some volunteers at City University London.

I've got some cards for them to choose from - where red means

they share a lot online, green means they're sharing very little,

and yellow is somewhere in the middle.

OK, so let's start by asking you to choose a card.

-I'll go with one of these, actually.

-OK.

Yellow, but I'd like to be green.

I'd like to think that I protect myself.

I do tend to try and go through privacy settings

on things like Twitter and Facebook.

I never save card details on any of my accounts.

I'm a bit paranoid so...!

Later on we'll find out whether they share

as little information as they think.

But first - internet security expert Professor Alan Woodward

is showing me a murky corner of the internet

where personal details are bought and sold,

known as the Dark Web.

These are bulletin boards where people

are discussing selling now,

not just credit card details,

but all sorts of different personal information.

There is actually quite a black humour side to this -

he's saying how professional he is -

"This is the result of three years' hard work".

You can get very specific.

There you are, date of birth, 15.

Why is that important? Because in the UK, for example,

a name and a date of birth, to a credit agency,

can be considered a unique combination.

So how much information does someone need to glean

for it to be really useful?

Anything you can add that starts to make your reference more unique.

So you don't need much of that.

Including, for example, put your home address down,

your date of birth,

if you can get hold of something like social security numbers,

National Insurance numbers, in the UK.

I could be whoever I want, from this list of people I can buy.

And to give you an idea of how easy it is

to collect data once it's out there,

Alan's asked James Lyne from Cyber Security giant Sophos to join us.

He's using some legal and freely available data harvesting tools

to gather information about our volunteers.

What these tools all really do,

is, they take individual pieces of information,

that in themselves would be completely innocuous,

so a name, a social media profile,

an e-mail address,

and they combine them together using these Big Data techniques,

expanding the information massively

and make a very accurate profile of what that person looks like.

Surprisingly, James doesn't need much information

to build an accurate profile.

People don't realise that often photos, tweets

and other data they may upload, contain GPS coordinates, by default.

So you might not give away your address or postcode,

but you're giving away your location to plus or minus 10-15 metres.

You might see 160 tweets that correlate to that location.

Plus, the tweet content may talk about being at home,

doing something for the kids.

It gives away very clearly that's where they live.

With potentially two to three years' backlog of data,

that's enough to build a profile of anyone.

And we'll be letting the volunteers know

what we've found out about them, later on.

From keeping planes in the air to stealing identities,

if you've got access to data

you can build some incredibly powerful tools.

We first found this out long before the Big Data revolution,

over 70 years ago.

Jem takes up the story.

During World War II,

brilliant minds gathered in these buildings at Bletchley Park

to decipher encrypted German messages.

The results helped shorten the war by two years,

and saved countless lives.

At first they used human computers, real people sat at desks

cracking the codes by pen and paper.

But by 1943, the engineers had realised machines might be able

to do a much better job -

machines that processed with simple on/off switches.

So how do you link simple switches to answer a problem?

Well, I've got two of them here.

Essentially, it's a tiny computer -

I now need to programme it.

Now the input to this I'm assigning to "Is it Monday?"

so if it is Monday - it gets a positive input.

If it isn't - it gets nothing at all.

This switch, the input for that, "Is it 7.30?"

And the output here, I'm assigning "Good time to watch BBC1?"

Right. Let's start using the computer. Is it Monday?

Yes. Is it 7.30?

Yes.

The computer says it is an ideal time

for checking out some science on your telly.

At Bletchley Park,

engineers hooked up their own network of on/off switches

to crack the German codes.

Where I've used two switches, this machine used over 2,000,

and it was aptly called Colossus.

Back in the day, Colossus was revolutionary

because it used these electronic valves

for its fast and reliable switching.

Fast and reliable for its time, because within ten years,

that same job was being done by transistors, considerably smaller.

Now, I pulled this out of a modern computer.

The central processing unit - the chip that does the switching.

And on there, there are 54 million transistors.

And it's that kind of miniaturisation

that has revolutionised what we can do with computers.

Using switches to process ons and offs is how all computers work,

but today they're known as 1s and 0s.

You might not think you can get much subtlety

out of a switch just being on or off,

but there millions of them at work for you right now,

sending out a stream of 1s and 0s,

sequentially telling every pixel on your screen

just how bright or dark they need to be.

Your holiday snaps? A sequence of 1s and 0s.

Your MP3s? A load of 1s and 0s.

And every letter on a keyboard

is an eight digit code of 1s or 0s, to a computer.

For Colossus, data was fed in on paper tape.

Each punched hole, or unpunched space, acted as a 1 or a 0.

Today, individual 1s or 0s are called bits.

Nowadays we reckon eight bits are a byte.

And to match the storage capacity of something like a hard-drive -

250GB - your piece of paper would need to go to the moon

and back, and probably back to the moon again.

So how can we pack so many bits into such a little box?

Well, most hard-drives work using magnets.

Computers magnetise an area of a disc

like I'm magnetising these bolt-heads.

I'll use magnetic North for 1 and South for 0,

which can then be detected later.

In a real hard drive the magnetisable areas

are sitting on a spinning disc.

Quite literally, on there, there are millions and billions

of magnetisable areas,

each of them so small, that they're smaller than a virus.

10,000 of them would fit across the width of a human hair.

And this is spinning around at 100 times a second or more.

And yet still the computer is extracting

a phenomenal amount of data incredibly quickly.

Now just because this all seems like

some ridiculous fantasy piece of engineering

doesn't mean you shouldn't have a go at building your own.

Now where's that MDF?

As a team, we are putting together a massive four byte hard-drive.

Four rings of eight magnets on a spinning platter.

As the disc spins past the electro magnet

it reads each bit as a 0 or a 1.

I've left Chris and Jim to secretly encode each ring

as a sequence of eight bits - enough for a letter on a keyboard.

That's a zero.

And I decipher the code back into letters.

What takes me 30 seconds,

a computer does at nearly the speed of light.

Oh! I mean, that's just brilliant! What can I say?

Milk and two sugars, please.

Our ability to store vast quantities of information digitally,

a bit like this, and process it with tiny, lightning fast switches,

is what's driven computing,

and opened up this whole field of Big Data.

And as engineers develop even better storage,

and even faster processing,

Big Data applications are going to have a bigger and bigger influence

on our everyday lives.

Back at City University, London, it's results time

in our personal data experiment.

First up, those who chose green,

believing they put no personal data about themselves online.

Could we find anything about them?

You've got my mobile number in there.

Which I'm a bit surprised about but I'm guessing that might come

from a shopping website or something like that.

It was from somewhere that you'd published it.

But it's not just his mobile that's public.

Do you use that e-mail address for resetting certain accounts?

Uh...yes.

I think when I said I was green

I'd forgotten how I put some of those things in.

It's amazing how much this information just stays there.

Absolutely.

There's actually an astonishing number of cases

where people thought they were really, really secure,

and gave nothing away, but in reality,

posted an awful lot of information online.

And even for those in the red group, who knew they had data online,

there were still surprises.

Under my name there's only my phone number and my e-mail address.

Whereas with the other guys it's their full home details,

addresses, many phone numbers for them.

This isn't about you being careless with your data.

This is about someone else being really careless with your data,

and all those other coaches, and the names of those children.

All the names just listed out there,

it's kind of a shocking thing to see.

In fact, all of our groups were quite shocked.

Can anybody who is not a member of this website just access it,

or do you have to, you know, become a member of it?

It's all accessible. We didn't register to get that.

That's good then(!)

So what, do you think, constitutes safe online behaviour?

So, firstly, don't be too paranoid.

I use Twitter, I use LinkedIn,

I enjoy online services.

But we have to think a little carefully

about the information we upload.

Do we want to give away the location of this photo, in our back garden,

that contains the location of our house, plus or minus 10-15 metres?

Secondly, consider lying online.

Now, I know that sounds like a strange thing to say

in the real world, but when a service provider says,

"What's your date of birth?",

don't tell them, and if they demand you give them that information,

give them a fake answer.

Keep note of that for future purposes, for a reset,

but don't tell them the truth.

And remember, once something's on the internet,

you really can't delete it.

So think before you put anything there in the first place.

So we should be wary about any information

that's out there and unrestricted.

But as Liz is finding out, Big Data's offering up

more than just new ways to reveal our identity.

It's also offering up

a new generation of facial recognition techniques

that eventually, may even be able to tell how we're feeling.

Two dimensional facial recognition systems have a wide variety

of very useful applications, but they're not completely foolproof.

If I hold up a picture... of Jem Stansfield's head,

because the system only analyses in two dimensions,

this flat picture can fool it into thinking

I'm someone completely different.

2D systems mostly work by measuring the distance

between your key facial features.

But the technology can be easily confused.

Here at the Centre for Machine Vision

at the Bristol Robotics Laboratory,

Mark Hansen and his team have made a system

that can see in 3D, like we can.

This booth uses a high speed camera,

and five near-infrared flashes to build up a 3D likeness of my face.

So it's captured all the images.

God, I look hideous!

That's an awful photograph!

That is a 3D image of my face, and it's saying, "Access denied."

Why is it not recognising me?

Because we haven't enrolled you on the system yet.

I walk through a few more times and Mark programmes the computer

to recognise the face its detecting as mine.

Yay! Good afternoon, Liz. Excellent.

We're extracting the key features of your face...

The height of my cheeks, the bump on my nose,

is that way it all boils down to?

Absolutely, yep.

This is Big Data facial recognition,

matching patterns captured across my entire 3D image,

with what it has already learned about my face.

It's more robust than 2D systems,

and you'd need a twin,

or a 3D print-out of someone's head, to fool it.

But the most advanced on display today isn't 3D.

It's actually 4D.

This system can process my reactions to a series of YouTube clips,

in real-time, guessing what I'm feeling.

These kinds of technologies aren't ready to leave the lab quite yet,

but this is how robots could see us in the future,

and identify what we are thinking.

Whoa!

And you can't help but imagine what else might be on the horizon.

Can Big Data predict the future?

This may seem a little far-fetched,

but in many ways it's already happening.

Police forces in the UK are trialling the use of data,

like weather forecasts and records of break-ins,

to predict where the next crimes might happen.

And online retailers are planning to pre-package our goods

before we've even ordered them.

Whether we like it or not, Big Data is here,

and it's going to change our world

in ways we could never have imagined.

Next week on Bang Goes the Theory, we look at the science of ageing.

And we'll be joined by Sir Terry Wogan.

Meanwhile, if you fancy working in Big Data,

check out our careers guide at bbc.co.uk/bang.

And for information on keeping your data secure,

follow the links to the Open University website,

and play their interactive privacy game.