Hi I’m Graham Mossman from EXASOL and today I’m going to talk about “The Big Data Internet of …. What ?”
I’m going to talk about Hype in general – in fact I’m going to teach you some of the tricks of the trade. Which is probably going to get me thrown out of the Magic Circle – we aren’t supposed to tell you how the tricks work.
We’re going to look at the Gartner Hype Cycle which shows how enthusiasm for a technology varies over time
We’re then going to look at where Big Data and the Internet of Things are on this cycle currently – which is incidentally, right at the peak of maximum hype craziness.
I’m then going to look at four things that the Hype merchants say about these technologies that really aren’t true and I’m going to close by looking at the future of these technologies, beyond the current hype.
By saying that Hype is everywhere, I’m not saying it’s a bad thing. Everyone exaggerates their good points to get a job or a mate. I’ve got to admit to using Hype myself – if everyone’s yelling, sometimes you need to raise your voice.
It’s just sometimes the Hype is a bit overwhelming and drowns out the message
And sometimes you wonder, like with Henry VIII’s codpiece, whether if under all that flattering padding there’s anything of substance at all.
The Gartner Group (as well as their famous Magic Quadrant reports) produce a report called the Hype Cycle. They have noticed that over history a promising technology will emerge from the labs and be taken up by early-adopters and an increasing amount of noise will be made about how this technology will change the world, even though it is at quite an early stage of maturity. This noise will reach a peak called the “Peak of Inflated Expectations”. Some kind of reality will then set in when it appears that the technology isn’t quite ready for prime time yet, isn’t the appropriate technology in all cases and in short, some more work is required. This is where some people stop making noise about this technology and start banging-on about something newer and hotter that has more recently emerged from the labs.
That’s not the end of the story though – the technology continues to be developed by those who are in it for the long haul and at some point the mature version of the product starts to gain admirers in a more gradual and sustainable manner, until it hits the Plateau of Productivity where it is an established technology that is as boring as wi-fi or radio or any one of thousands of technologies that were once hot new inventions but are now just taken for granted.
Last year, Big Data was at the Peak, this year it is the Internet of Things. Gartner predict both of these technologies are 5-10 years away from becoming established technologies and they also predict that the current mad hype around these technologies will lessen – in fact they say that some reality about Big Data hype has already arrived, and I think that’s a positive sign.
For example, if we look at “Virtual Reality” which sits near the bottom of what Gartner call the Trough of Disillusionment, this was a technology that was white hot a few years ago, but has been slow to made the breakthrough that once seemed imminent. At one stage it was said that there were more international conferences about Virtual Reality than there were paying customers. However, it’s now a more mature technology with a more focused set of Use Cases and so recently it has attracted increasing attention and is showing signs of a “second coming” – this time as a more mature technology with more reasonable expectations.
Here is my definition of Big Data – I’ve tried to keep it as straightforward and sensation-free as possible
And again an unsensational definition of the Internet of Things, you’ll see that it is very much like Big Data, except for the first paragraph which describes how the data is collected. Big Data and the Internet of Things have many ambitions and techniques in common.
There is no one single Nobel prize-winning discovery behind Big Data and the Internet of Things – we’ve had embedded systems in the 1960s way before the term “Internet of Things” was coined. And way before “Big Data” we had something called … “Data”
What has happened recently is that a number of incremental developments in a number of technologies have combined with cheaper hardware and software to make new techniques economically and technically viable.
It has been suggested at various points in history that a new technology has rendered the old business rules obsolete and you need to value businesses and investments in a new way. This happened in the 1920s during the explosive growth of Radio in America and in the 1840s during the British railway boom, but it happened again in recent memory during the “Dot Com” madness of the 1990s.
Investors and entrepreneurs were persuaded that you couldn’t measure the worth of a dot com by old-fashioned metrics such as do they sell stuff for more than it costs and are they going to be able to pay their staff next month. Sounds ridiculous but actually it took a long time for people to realise that “old-fashioned” business rules like that did apply even to the newest sexiest technologies.
Big Data and the Internet of Things are not exceptions to this – entrepreneurs and investors will need to pay heed to the boring and old-fashioned idea of a Business Case
Here’s a very early and rather trivial Internet of Things product, a wifi connected kettle that can have water available at the perfect temperature without you having to wait for it to boil. Can’t help noticing that you still have to fill it, but take a look at the bit I’ve outlined in red – “Saves over 2 days a year in waiting time”. Wow, get me three of those and let’s really start saving time. Two days of my time is worth more than 99.99 so it’ll pay for itself in less than a year.
Except that it doesn’t – when I’m doing something valuable, I don’t use the kettle. Then, usually when I need a break, I don’t mind waiting a minute or so for the kettle to boil. This isn’t wasted time, it’s break time. Or I’m talking to a co-worker or checking social media. The time is not wasted and shouldn’t be valued at my hourly rate.
So, in my view, the business case for the iKettle doesn’t hold water (boom boom)
I was an accountant before I became whatever I am today, so I’ve prepared some flaky business cases in my time. If you watch “Dragon’s Den” you will have seen most if not all of these tricks in play – sometimes all in the same pitch. You need to look carefully at the business cases around Big Data and the Internet of Things - astronomical amounts of money are “proven” to be “in play” but some of the assumptions are just variations on some of these tricks. I don’t have time to get into this in detail now, but invite me back sometime and I’ll do a whole 20 minutes on Bogus Business Cases for you – it’s important that you can spot when somebody is lying to you like this.
The last trick on the previous slide (“growth will be exponential for ever”) is epidemic in the field of Big Data and the Internet of Things. Here is the single most overused and abused diagram in Technology marketing. Moore’s law talks about the doubling of the number of transistors per square inch of integrated circuit. It has held for the last 40 years or so and Marketeers love this diagram because it allows them to introduce exponential function into their evaluation of what a technology will be worth in the future. But you can’t just say “Because Moore’s Law” – Moore’s Law is a very specific observation about transistors on an integrated circuit.
Some people misuse the rule by saying that “computer power doubles every two years” – this is nonsense – disk speed, clock speed and network speed don’t double every two years – computing power is more than a matter of transistors per square inch.
And once you’ve misused the rule a little – why not misuse it a lot and suggest that everything to do with a particular technology will double every two years “Because of Moore’s Law”.
And finally … just because it has held for the last 40 years, why should it definitely hold into the future ?
Air speed records start out with the Wright brothers and a bicycle-powered plane capable of 7 mph. For the next 30 years we have a linear growth before WW2 and the invention of the jet engine. Things do briefly get exponential as we break the sound barrier, but that curve doesn’t keep on going – in fact it stops dead in 1976 with the Blackbird spy-plane which continues to hold the airspeed record.
Another thought – the youngest man to walk on the moon will celebrate his 80th birthday this year. Manned Space Travel is another technology that, ironically, lost momentum.
There are a lot of reasons why growth stops being exponential and sometimes stops altogether. There is no economic case for supersonic flight – the Concorde was too expensive and we don’t need faster spy planes now that we have satellites. Google and Microsoft would be even bigger if it weren’t for the efforts of various governments preventing the creation of damaging monopolies. There is wide-spread social opposition to nuclear power in many countries which means that the early growth has not been sustained. With mobile phones, if almost everyone already has one, it’s harder to sell more. In particle physics, the American version of the Large Hadron Collider was scrapped because the American politicians thought there were better things to spend money on during the last recession. Similarly, the Wall Street Crash hampered the development of the telephone – because nobody was making the big capital investments required to sustain explosive growth.
And sometimes a technology will hit a wall. Maybe not permanently, but certainly enough to stop exponential growth. Quantum computing has threatened to be a major force for some time, but it turns out that there are unforeseen difficulties that have been harder than expected to overcome.
So all I’m trying to say here is that if someone tells you that a technology is “growing exponentially” – you need to ask yourself – “is it really ?” and “for how much longer ?” because in a finite world, nothing grows exponentially – apart from maybe Technology Hype.
Here is a slide I used at the last Dataconomy London presentation to describe what you need for a successful Big Data application. You need lots of badly –behaved data – you also need a well thought-out question, in other words, a well-defined business benefit you want to achieve. You need a Mathematical algorithm for getting some sense from the badly-behaved data, you need an IT platform capable of getting fast results (and yes, that means you should be looking at EXASOL) , and then you need statistical knowledge to be able to properly understand what the results are telling you.
There is some good news here : most businesses these days are able to gather and store large amounts of “Big Data” in an economical manner. There are Open Source platforms for data movement and storage, for example Hadoop and there are also a new generation of In-Memory databases that allow fast analysis for not all that much money (such as EXASOL). The cost of the IT hardware required to support these platforms is also dropping – I will resist the temptation to say it’s dropping exponentially because of Moore’s Law, but it certainly a fraction of what it was in the past.
There is however, some bad news - the Mathematical and Statistical skills required for a Big Data solution are in short supply. If you are a small company wanting to leverage Big Data, you will need what is called Data Science to be able to build and refine the algorithms that make a Big Data application work – and also to interpret the results you are getting. This is great news for those of us with Data Science skills, but it is a brake on the mass adoption of this technology.
In short, Big Data only works if your Business Case is so strong that you can justify the expense of buying-in Data Science skills, andso it is not something that is ready for everyone, right now.
An Internet of Things Application is even less ready for everyone, right now – as well as the shortage of Data Science skills, there are technical issues around sensors and networking and The Cloud, but also political and social issues.
I’m no expert in the technical issues that are being faced in making sensors small enough, connected enough and cheap enough to do the business. But if you are putting sensors into potentially everything on the planet and storing that information at a central point, there are certainly some sociological and political questions that need answers. And these, rather than the technical issues may be what slows down or halts the development of this technology in the future.
… but when you think you have a business case for Big Data and the Internet of Things – you should be talking to EXASOL