"Prediction is very difficult, especially if it's about the future," said Niels Bohr, the father of quantum physics. He has a point. These days we are besieged by predictions from pundits, futurologists and analysts. They have little to lose by being wrong, but CIOs stand or fall on the strength of their own predictions.

I quite envy the futurologists, who come out with some great-sounding idea and in the 20 or so years it would take for people to realise you were talking utter rubbish, they will forget, and you will come out with the next great prediction for the next 20 years - my kind of job.

I once heard the great futurologist Ray Kurtzweil speak on understanding how the brain works. It's quite simple: get an army of nanobots and inject them into someone. They move around the brain, map all the synapses, and bingo - you know how the brain works.

Besides the nanobot creation problem, I have a bigger problem with this. Just because you know the wiring diagram doesn't mean you understand how something works. Give my dog a wiring diagram of my laptop and I don't think for a moment that he could understand how it works. This strikes me as an obvious flaw, yet the audience was perfectly happy on this one. There's a lesson here for budding futurologists: if you are going for it, go for it big time, nanobots and all.

Today I can't pick up any publication without reading about cloud computing. As someone who worries daily about
13 petabytes of customer information in the cloud, it's clearly already a reality for some. However, as I sit here in Silicon Valley, you would be forgiven for not believing that any self-respecting CIO who hasn't thrown all their systems out onto the street and rushed headlong cloud-wards is somehow a dinosaur.

Firstly, why move a working system? The old IT approach of 'it's working, don't touch it' means that a lot of traditional processes will remain firmly on the ground for a long time yet. The museum of IT is usually found working away inside corporates, and quite rightly so. So the first of my predictions is that cloud computing will really take off for new applications, not traditional ones.

Secondly: empowering non-IT. I believe a major driver for cloud computing is the desire of business units to take back control of critical systems, something that many CIOs will welcome. Marketing departments no longer want to wait three weeks to get their website optimised and IT could do without the hassle. In these situations, cloud-based solutions will thrive, allowing business units to work their magic without worrying about servers, backups and so on.

Related:

Thirdly, say 'no' to dogma. Pure cloudists like to imagine a fluffy, sunlit world up there. The reality is that a lot of the processes and information will originate from within the enterprise, which has its feet firmly on the ground. An example of this approach is collection to the cloud for legal hold and e-discovery. These systems have on-site solutions to collect information, examine it and ship the right stuff off to the cloud, while keeping irrelevant or hypersensitive stuff in the basement. I see this hybrid model as the architecture for the major systems.

As CIOs know, people are funny things. Companies are often happy to archive highly sensitive information in the cloud, but will reject the idea of cloud-based enterprise search. "The information is just too sensitive," they cry. But as soon as you point out that they have already put the same information in the cloud as part of the archive, the embarrassed silence of irrationality emerges.

In talking about the cloud, it is often seen as somehow out of touch to mention security of information in this world of 'mine's longer than yours' encryption key length. But let me give you two examples.

One of my first jobs was writing police systems and setting traps for bent officers looking up information for people. It always amazed me that the going rate for this was just £50. Fifty quid to commit a criminal offence, risk jail and ruin your career. Another example is road tolling in Thailand: a tollbooth collector could handle their weekly wage in just 10 minutes. The ingenuity of the human mind in finding ways to get the booth computer offline for those 10 minutes and pocket the cash was staggering.

The reality is that cloud computing is an excellent paradigm for the right tasks. The important thing for CIOs is to adopt pragmatism, not dogma, and use it in the applications that empower users. Unlike virtualisation, which was all about the IT department, cloud computing is about end-user functions, marketing, sales, legal support and production.

Next time someone looks unimpressed that you have not moved all your systems to this month's revolutionary idea from Silicon Valley, ask them if it's powered by nanobots. If it's not, it's just so yesterday.

Mike Lynch is the founder and CEO of UK software company Autonomy