Peter Cochrane's Blog: My productivity has peaked

Written on the Ipswich to London train and dispatched to silicon.com via a free wi-fi service in a London hotel later that day

Ever since I started work in 1962 I have seen my effective output increase year on year. Before the explosion of ICT, I used all manner of techniques to minimise paper, cut back on duplication and improve my overall efficiency and effectiveness. These measures very nearly doubled my productivity each decade but with the arrival of ICT this figure gradually rose to tenfold per decade.

By merely taking advantage of the continually increasing processing, storage and communication power, I achieved a substantial increase. But it was by using a continually evolving raft of applications, coupled with adapting my working practices, that the whole really took effect. Through the latter half of the 1980s, all of the 1990s and until recently my progress was relentless - tenfold every 10 years.

Unfortunately I am now approaching stasis. Any amount of basic machine upgrading, and it continues apace, won't make a jot of difference, as I am now the fundamental slowdown agent. I just can't work any faster; I just can't find any new tricks to get another big productivity hike. It seems I'm to be stuck in a 2004/05 mode for some time.

So if raw processing power, storage and bandwidth can't help, what will? What is it I need to leap forward by another factor of 10? In a word: intelligence. In two words: machine intelligence. I need something that monitors my activities, anticipates my next move and automatically satisfies my needs.

Search engines are impressive and often useless at the same time. Knowing that there are 64,431,000 references for the term 'artificial intelligence' isn't a whole lot of use! What we need is a cognitive approach with search material retreated and presented in some context relative to our current end-objectives at the time.

When will we see such technologies? They are around now and in use by people engaged in companies, security and legal systems who have to be able to search vast numbers of documents quickly. Some of the technology is based on statistical inference techniques with some extended by artificial life. For the time being they remain far too expensive for most but sooner or later they will appear to a wider audience.

The other big winner is in the area of decision support based on modelling and prior history.

For my part, I'd like to start with a system that manages all my interfaces and remembers all my habits. An effective search engine would be next followed by modelling and decision support. I really do need symbiotic support!