Time for another instalment of my NI Digital Experts interview series. This week we’re featuring one of the Northern Irish technology scene’s more colourful and enduring figures: Jason Bell.
I first got speaking to Jason via Twitter and then we met at a few networking events. We quickly realised we shared an overtly cynical attitude to the vacuous tripe that emerges from Silicon Valley’s startup culture, and want to resist the adoption of that culture in the Northern Ireland tech scene.
When it comes to big data and machine learning, I know no one more qualified than Jason. He wrote a book about machine learning which has helped me immensely in coming to grips with the topic, even though I can’t even begin to understand the mathematics behind it all.
Like myself, Jason is not native to Northern Ireland, but he’s been here so long he might as well be part of the furniture. A much-loved and widely respected figure, Jason’s views and opinions are always worth listening to. And he’s also a pretty damn good bass player.
So let’s hear it from Jason and his career in digital:
Tell us a bit about yourself and your journey in to the digital industry.
My journey started early, I was nine years old and the headteacher at Fishergate Primary School in York, Mr Male, had this brand new Sinclair ZX81 hooked up to a television. The programming book looked huge to me at the time but I found more interest and challenge in getting numbers to add and subtract than playing games on it. A year later my parents bought me a Spectrum and I taught myself programming. At secondary school during breaks and lunches a friend and I were usually hacking away on a Research Machines Z80, huge thing it was. In 1988 the last words of my computer studies teacher was “you’ll never get a job in computing”….
Leaving school at sixteen I went straight into what was considered cutting edge, it was called an ITeC (Information Technology Centre) in York. I managed to do the entrance exam though I didn’t have the required GCSE requirements to get in. So the ITeC did the BTeC NVQ III one day a week and the rest of the time I was out on work placement, one of the first to do it this way back in 1988. I was offered a job while I was studying at the ITeC, being the only person interested in hardware and not COBOL programming. I’m still very good friends with the very first customer I supported back in 1989.
The nineties were a mix of jobs in support, retail and supply chain where I learned a lot about business and customer support. I also learned a lot about stock pattern prediction while working in a record shop (called 4Play, I kid ye not). In 1997 I landed a job at the Press Association working on the Sporting Life website and that opened my eyes to web things and backend data stuff. My boss, Colin, taught me Perl on the back of a KitKat wrapper.
Since then I’ve worked with various companies, some established and some startups. You learn a lot along the way, what to do well and what to avoid, that’s not just about the programming but people, customers and everything that goes with it.
I had a go at the startup thing myself with uVoucher which is a customer loyalty system for retailers, it’s still there waiting for the right day plus I’ve added the machine learning and AI stuff in now. It’s fairly common knowledge that I love Dunn Humby and the Tesco Clubcard, I worked for a customer loyalty mining company in 2002 in Harrogate, not for long mind, but that really woke me up on the power of data. I’ve stuck with data ever since and am more than delighted that I’m with Mastodon C who are a big data company in London.
Quite simply I’ve learned that your journey is really down to your attitude and your network. My BTeC isn’t worth the paper it’s printed on, it’s 26 years old and it’s safe to assume the industry has gone through 52 six-monthly iterations in that time. I’ve been fortunate enough to learn plenty from various positions that has brought me to the place where I am now. Still don’t, and probably never will, have a degree.
You’re known throughout NI as a ‘big data’ expert, but also as a mean bass guitar player. If you could pick just one, would you rather be a musician or a big data expert?
I made a choice way back in the nineties to concentrate on the computing and data work. That all stemmed from a conversation I had with bassist Brendan Rothwell, he told me that work pays for being a musician but you are at the core a musician. He was 100% right. The profession of musician is just a precarious as being a startup founder, actually the probability of making money is slightly better but only just.
Most of the brilliant musicians I know have some form of other job paying for the gear, the rehearsal spaces, the CD duplication and so on. Being a musician and being in the music business are two very different areas.
I love both, please don’t make me choose, they compliment each other.
The term ‘big data’ gets thrown around rather carelessly these days, and as such its value and meaning has declined. What do you think is the true heart of ‘big data’ and why should we care about it?
I’m fine with the term “Big Data” as long as it’s used in the right context. To be honest I care little about it now compared to five years ago when I was talking about it in NI and getting blank faces all round.
I lost all faith in the media’s version of Big Data when it jumped on the ‘Target Baby Story’ while Charles Duhigg’s book “The Power of Habit” was being publicised. That had nothing to do with Big Data, it was just good maths. The true heart of big data is the scale, if it can’t be processed on one machine then is it big data? I feel the tech press needs to work on a new set of cliches as the old ones are wearing really thin now.
I’m at the point now where I actually care little about the term big data, it’s just data to me. Even walking around StrataConf there wasn’t a huge amount of reference to the term “big data”, the assumption is that it’s big all the time and are more bothered about the insight that can be gained from it. There was far more emphasis on streaming data than big data. The real conversation was about time recorded and time processed for streaming data applications. The next round of blog posts from me will be about streaming workflows I think.
The rise of ad blocking and the increasing focus on consumer privacy has also placed scrutiny on big data and how it is applied. How do you think businesses can best balance privacy issues with effective consumer data mining?
I think it’s rather simple customer communication. Businesses need to be upfront with customers about what they intend to do with their data. Whether that’s to improve the algorithm quality, find upsell potential or store the stuff about them, a customer should be aware of this on signup to the service. Data is money when used right, a business has the responsibility to treat the data, and the customer, like gold.
Look at the Tesco Clubcard as an example, people are quite happy to swipe or scan their card at the checkout in return for money off coupons. Simple. A win for the customer and a win for Tesco who analyse and monetise the data to their suppliers. It gets on shaky ground when you predict what the customer wants a bit too close or beyond their comfort zone. This is the exact problem Target had with the baby club story, so they purposely put in random items to let the customer think “ah, you don’t know that much about me”. Customers weren’t aware data was going to be used in this way really, the question came about much later on.
With the right data companies can do some clever things. Let’s not forget that storage is cheap, put it on Amazon S3 for example, gigabytes for pennies a month. You can sit on that data for years and then think about mining this historical to create a model for the current, within reason. Just don’t freak the customer out.
You’ve literally written the book on Machine Learning, which is a phrase that’s currently doing the rounds in the SEO community in the wake of RankBrain. Does Machine Learning technology live up to the hype?
I don’t think it’s hype at all, some of the machine learning algorithms we use are fifty years old, I think the hype has come out of the notion of the amounts of data available along with the scale and speed we can process, cars that drive themselves, Uber’s surge pricing, robot workers that will steal our jobs and that kind of gig.
With the majority of machine learning algorithms out there you can at least verify the numbers that are being produced. The hype I’ve seen really revolves around deep learning and neural networks which become black boxes we throw data at and expect answers to come out of, we’re expected to accept that as truth because the system says so.
Some of it’s really simple to do, word2vec is fun if you have a document corpus large enough. RankBrain doesn’t need context just documents and it will work out related words from there. Some of the algorithms are head melters if you’re not used to them, Support Vector Machines are a hard slog to figure out if you’re not used to mathematics (I’m not, I’m a software developer).
Machine learning is commonplace and was before the tech press made it all in vogue. Now’s a good time to be near it. I landed lucky with writing the book, right people, right place and right time.
You and I share an aversion for vacuous Silicon Valley-style entrepreneurial hokum. How can we best fight the spread of meaningless tech startup waffle in Northern Ireland?
Well it comes from experience, I worked out in Sunnyvale in 1999 before craft beer became a thing. I found out very quickly that the hype in the area was insane, even down to basic writing of HTML, the quality was poor to say the least. So I came away from there with a dose of realistic caution, tech waffle is not something I handle well, common sense usually prevails. It’s a whirlwind over there, 17 hour days with some delightful curry sent in for the evening slog.
In 2009-2011 I got caught up in the wave of startup culture in Northern Ireland, it was a raw and exciting time, there was a feeling in the air that something good could really happen. People may disagree with me but now it feels a little more fragmented. There’s a Cathedral Quarter vibe, a Titanic Quarter vibe, there’s a UU/QUB vibe and a Catalyst Inc. vibe, loads of different vibes going on. They’re all doing great things but there feels a lack of community cohesion. Perhaps it’s because I’m not in Belfast, I don’t know, I may well be missing something or perhaps I have better clarity for not being in the eye of the storm.
Quite a few founders have contacted me for advice, I have no idea why apart from talking with a large dose of common sense as I do, I certainly don’t have a unicorn in the shed, but it’s alarming the number who can’t be bothered to come over Derry direction, best way to kill a conversation is for said founder to say, “No rush, let us know when you are in Belfast next.” One large startup is still waiting for me to go to Belfast 18 months after the initial phone call, I think they’ve pivoted by now.
A few blog posts I wrote have raised eyebrows I know, to me the startup culture here is like Maslow’s Hierarchy of Needs. It appears to be a numbers game to me, jobs promoted, startups put through a certain programme. I struggle to listen to anyone telling me “it’s great to be in a startup” when it’s usually someone in a full time job for some agency telling me that may have no idea at all. My only paper qualification is in accountancy so I know how to read a balance sheet and do the accounting ratios, so as a hobby I read startup accounts and boy they can be an eye opener. You can talk the talk all you want but your liquidity ratio is far more interesting.
The only redeeming feature out of all this startup lark is that got me very interested in probability and betting odds. Once you realise your startup is a number in a roulette wheel you’ll see how random it can be. The lucky ones are the ones who know how to create those chances, as Francis Bacon was quoted, “A wise man creates more opportunity than he finds”.
Lastly, give us one website or app that you feel is vastly underrated and deserves a wider audience.
Simple, if you’re in Northern Ireland and you’re involved in the tech sector and content creation then make sure you’re on Digital Circle. Matt Johnston was and still is a networking marvel.
About Jason Bell
Jason Bell has been active in the technology industry over the last 28 years. He is author of Machine Learning: Hands On For Developers and Technical Professionals and specialises in data engineering and machine learning. He doesn’t take this startup lark too seriously. You can read his blog at http://dataissexy.wordpress.com.
Jason on Twitter: @jasonbelldata