The joke doing the rounds on social media compares big data to teenage sex: everyone's talking about it, only a few know how to do it, they all think everyone else is at it and so pretend they are too.
And humans fumbling about in the dark, so to speak, are the weakest link in the world of analytics and data, said Steve Brazier, CEO at Canalys.
More ReadingSHARE your big data scientist. They're too costly and rare to keep a whole one – HP execEMC partners fear rebate dive in new channel programmeVeteran vendors are shaping up for a bitter fight to the DEATHFormer Toshiba bigwig Andy Bass rocks up at Open SymmetryHitachi Data Systems hands out P45s ... to EXECUTIVES
Companies are ploughing money into big data projects "without knowing what they are doing" and handing projects to the IT department "far divorced from the business operations", he claimed.
"The biggest problem of analytics so far is humans. We all suffer from confirmation bias, we look out for data that we agree with and then we use it. But the data we don't agree with, we don't use."
Brazier added: "Getting human beings to ignore that [instinct] is very challenging, even highly trained professors of statistics consistently make basic errors on confirmation bias."
Nominet, the .uk domain name registry in the United Kingdom, agreed with him, describing confirmation bias as a "very real danger when organisations approach big data".
"It can mean meaningful patterns are missed altogether, or we interpret statistical anomalies as trend when dealing with smaller data sets," Simon McCalla, CTO of the not-for-profit-organ told us.
He said that typing queries into a database sequentially "often doesn't suit the way our brains are configured" because we are visual creatures who recognise patterns by seeing and touching them.
"As such, a major part of any analytics programme will be data exploration – not necessarily testing specific hypotheses but using tactile and visual interfaces to examine and interrogate data sets to show up trends we wouldn’t think to ask about," McCalla said.
Use your illusions
This has helped Nominet data nerds "unpick potential issues" that were spotted before a potential exploit in the domain name systems allowed "serious damage" to occur, said the CTO.
The "innate inability" of people to use logic is more than just a challenge in big IT projects, says Logicalis CTO Chris Gabriel - just walk through any High Street after closing hours on a Saturday night.
"If companies invest in analytics just to affirm a false premise then they are not only wasting their time and money, but, doing their organisations a significant if not catastrophic disservice at the same time," he said. People make decisions based on their predilections "which is why more and more we are seeing analytics projects coupled with collaboration projects", he added.
"Organisations want to spot a trend and then they want to share it, and get an answer based on a collective wisdom, not an individual bias." Big data is billed as the next big thing in IT, driving revenues for vendors and channel partners at a time when traditional tech sales are under immense pressure.
It will 'redefine humanity'? Really?
And in an industry known for marketing bluster, vendors haven't been shy in talking up big data.
According to Dave Coplin, chief envisioning officer at Microsoft - no we'd not heard of that job title before either - big data will "redefine humanity".
Toning that with a healthy dose of realism, the Microsoft man conceded the road ahead is bumpy, saying big data will challenge "established human traits, like destiny" [presumably the illusion of having one] but also outdated frameworks and regulatory approaches to managing privacy.
Most organisations concentrate on the re-use of existing data but were "scratching the surface" when it comes to joining up disparate data sets or creating multiple use data sets, said Coplin.
He agreed the "main challenge is the human element" as big data forces a change to scientific approach - moving from trying to work out why something happens to a "world of correlation based on sample sizes".
"Organisations are going to need a new breed of data scientist. They’re going to need people with data skills, but not just maths and stats, they need a blend of data science, computer science, design and storytelling.
"In the short term, I suspect many organisations will buy these skills in (they are in very short supply) but over time, a new generation of 'engineers' will emerge and likely be found in most organisations regardless of size".
Big data bods could be the accountants of tomorrow, he told El Chan.
OK. But what are we LOOKING for?
Like Microsoft, HP-owned Autonomy reckoned customers need to define big data before they can make sense of it, which seems like a sensible starting point.
Companies with reams of information need to consider many aspects of that data, said Autonomy CTO Fernando Lucini, including the "latent or hidden value within your data, and taking advantage of that".
But where Autonomy's argument becomes less clear is when the CTO disappears down a verbal hole last dug by US politician Donald Rumsfeld.
"Another important thing to consider is simply that you don’t know what you don’t know. It’s not just about knowing what questions you want answered. That’s part of it, but it’s one piece of the wider puzzle.
"What are really valuable are the unknown unknowns – the surprising things you can find out through big data analytics that you would never in a million years have thought to look for.
"But to get at those, you have to have a technology that’s capable of looking at everything, whether that’s in the form of highly structured information in a database, or social media data, or images and video files, and understanding what it means, and how it relates to everything else. That’s how you find the really interesting patterns and unlock that value," concluded Lucini.
That rather sounds like a vendor trying to convince channel partners to convince their customers they need refresh their IT estate before big data can work its magic.
Chris Roche, EMEA CTO at EMC, told us that confirmation bias is not the only obstacle facing big data.
"Confirmation bias, correlation versus causation, privacy, access to skills are all potential examples," he said.
"As with any transformational undertaking there will be challenges to overcome. Navigating the journey to an analytics focused organisation is no different," he added.
We asked Oracle, the purveyor of databases, what it is doing in the world of gigantic data, but were told "whilst we have reams of information on big data, this is not an area in which we are experts".
And here we come full circle back to those sexpert teenagers again... ®