nav search
Data Center Software Security Transformation DevOps Business Personal Tech Science Emergent Tech Bootnotes BOFH

IBM maps the future with ODIS

Making what could happen less of a guess

By Martin Banks, 1 Sep 2006

In broad terms there are three key elements to consider in the development of any new application – or any new product, for that matter. The obvious one, certainly from a software point of view, is the technology to be used – the language, the target platform and operating system, and the rest.

But this is largely a set of tactical choices made after a couple of more strategic ones - what to develop and why - coupled with when to develop it and, sometimes more important, when not to bother after all.

The further out in time a company can start that strategic decision-making process the more equipped the business will be to react appropriately when the time is right.

And even though the common talk these days revolves around the ability to develop products rapidly, the real trick for many businesses is still to build up a strategic picture of what is coming down the track as far into the future as possible.

This is one of the capabilities now being developed at IBM's Almaden Research Labs in California. True, the work that underpins On Demand Innovation Services (ODIS) was initiated by major customers needing any sort of strategic view possible for 20 years into the future, and that remains IBM's big target market for the service, but the approach that lies behind it is one that can be extrapolated to other, more parochial environments as well.

According to Dr Ray Strong, the Science Lead at Almaden for ODIS, the idea came at least in part as a response to client requests to look further into the future than the three to five year views provided by the company's Global Technology Services. The demand was for 20 to 30 year views instead, and the initial response was that it could not be done, it would only be guesswork.

To meet the demand, the researchers came up with an alternative approach built round new techniques in futurology that bypass any attempt to predict what will happen in 20 years time. Instead, the approach they adopted was to build client-specific models based on defining potential sign-post events that would indicate which vision of the future was more likely to occur given the occurrence - or otherwise - of those events.

"The sign-post events are not dated but simply act as map points to different possible futures - the most likely, the less likely and the very unlikely. This is not a matter of dates but of visibility and feasibility," Strong said.

"IBM researchers use current knowledge across a wide range of issues - not just technology but also science, demographics, political trends and many other disciplines. Some of this is well enough understood that it could even be dated. This knowledge is combined with really wild ideas that may never happen - if only because no one can envisage how they might happen."

It is, however, possible to work out what else might be required – and that could be anything from a technology development through to a political change – for such a "wild idea" to actually become possible. That is the nature of the sign posts, which therefore require two specific properties to be included in the model.

The first is that they can be recognised as clear potential future events. "They have to be very precise and measurable," Strong said. "For example, if the sign post event is 'a new processor, only 5mm square that contains the equivalent of 1,000 Itaniums' it may not be specific enough. Is it just being written about as a research idea, is it a working prototype in a lab, is it now available on the market?

"Alternatively, you might want to make the sign post a relative measure rather than an absolute. For example, compared to a current known technology there is a leap forward of a defined magnitude in very specific parameters. These factors are important in determining whether the sign post event has happened or not."

The second property is that the sign posts must be actionable. Each one carries with it a recommended action, at the strategic level, which is accompanied by a cost/benefit analysis. This can then be set against the then current business plan and the business priorities re-evaluated accordingly.

"So businesses have things that are planned and ready to execute, and the strategic contingency plan with its own cost/benefit analysis. You can't compare them until the sign post event has occurred, but when that happens the current plan can be reviewed against the alternative," he said.

The actions taken can then be geared to the specifics of the sign post, so that a report of technology developments in a laboratory might lead to investing in the planning process for future commercial exploitation, while announcement of product availability might even oblige a need to find new markets as a planned project is now already miss-timed. Strong suggests that the ODIS approach is helpful with this issue of timing business and product developments.

"As many businesses or products fail for being too soon to market as for being too late. What we are offering here is not a crystal ball, but it is a method for systematically analysing the future and coming up with reasonably educated guesses about the time to do things."

Though ODIS is a service aimed squarely at the biggest businesses, its capabilities could become of growing interest to smaller operations as well, for an inevitable by-product of the process is the build up of a growing matrix of sign-post events that can be re-used.

"Over time we will get better at it and more efficient though I hope it never becomes a commodity," he said. ®

The Register - Independent news and views for the tech community. Part of Situation Publishing