Saturday 19 April 2014

So the Tasmanians forgot how to fish!

I was interested to read Adam Benton's post The curious case of the people who forgot how to fish on the Evoanth blog. Basically some 30,000 years ago Tasmania was a peninsula off the coast of Australia and there is good evidence from cave sites about 20,000 years ago that Australians had established themselves on the peninsula. However about 12,000 years ago rising sea levels meant that Tasmania became an island and the population became isolated. Adam argues that the evidence points to them having abandoned fishing (common on the mainland) and lost the skills to make some tools.

Whether his interpretation of the facts is correct (and some of the comments suggest other interpretations) there may have been many comparatively small groups of early humans which became isolated for varying periods. This separation may have been physical, as in this case, or may have been due to different groups exploiting very different environments. Different groups could end up with distinctive tools and cultures and this account reminds me that skills obtained by one group in onne environment are not automatically passed on to later groups in different environments.

Sunday 6 April 2014

Leo Computers Diamond Jubilee Reunion



 I have just returned from the Leo Computer Society's Diamond Jubilee Reunion which was held in the Hall of the Middle Temple and there was a good turnout with about 150 people who had worked with the pioneering Leo Computers, together with some friends. In my case I had programmed a Leo III computer at Shell Mex and BP, Hemel Hempstead between 1965 and 1967, before moving on to work with English Electric Leo Marconi between 1967 and 1970 (when it had become ICL).

As always, at such events, there were short presentation from some of the pioneers and there was a chance to exchange stories with old collegues, although as time goes by there are the inevitable news of someone who has passed on since the last reunion. At the previous reunion I had not found anyone who had worked at Hemel Hempstead at the same time as I was there - but this time I met two - and also several regulars from the time I worked at Computer House, Euston.

Of course, when I was working in the 1960s I only saw a small part of the action and the reunion gave me a change to talk to people who had worked on the design of the hardware through to programmers and systems analysts who had worked on different applications for different organisations - a stimulating experience.

There was some original hardware on display, and photographs of the Leo computers - and a good display on manuals. On the historical front I was interested to here that the old Intercode assembler program is now up and running so that anyone who has a program from that date can now run it.  There was also an appeal to see if anyone had a listing of the CLEO compiler and it would then be possible to compile historic CLEO programs. 

CLEO (Clear language for Expressing Orders) was an interesting high level language which could be considered as a hybrid between CODIL and Algol - and it was the first language I ever used. I was one of the first programmers to use CLEO as their first language in 1965 and quickly became an expert in it. However for some application I later learnt to use Intercode. At the time John Ruddle was the chief programmer at SMBP and when the company was planning to move from batch processing using magnetic tape storage to a system with online capabilities he wanted me to work on the operating system requirements of the computer that would replace the Leo IIIs. My interest in computers was at the human end - and I objected the the proposed move - which is how I came to work on the options from transferring the complex sales contract application onto the new machine. This decision had a big effect on my life as my investigation into the sales system led directly to my later research into CODIL.
[For a detailed account of my early programming experiences see Working with Leo III Computers at SMBP, 1965-1967]

The Death of Individuality - How original ideas can be suppressed

My approach to human intelligence is that we have the same logically rather crude symbolic pattern recognition and pattern processing language as animals. The difference is that we have learnt cultural techniques for speed learning. As a result I was excited to read Alex Pentland's opinion piece in the New Scientist this week entitled "The death of individuality". He writes:
If somebody else has invested the effort to learn some useful behaviour, then it is easier to copy them than to learn the effort to learn it from scratch by yourself. If you have to use a new computer system why read the manual if you can watch someone else who has already learned to use the system? People overwhelmingly rely on social learning and are more efficient because of it. Experiments such as those from my research group show us that, over time, we develop a shared set of habits for how to act and respond in many different situations, and these largely automatic habits of action account for the vast majority of our social behaviour.
A key factor of my evolutionary brain model (diescussed on this blog) is that once cultural information becomes more important that trial and error learning it pays to parrot your elders, and accept what they say without question. One of the tools that is passed from generation to generation, and is continually refined, is language. The better the language the faster you learn - so the more you can learn. This explains the explosive growth of human culture that started perhaps 100,000 years ago. The brain didn't change - instead we reached a tipping point in language development which made accelerated learning possible. So Alex's suggestion that we tend to work in copycat mode fits very well with my model.

However Alex's model also explains how the establishment-questioning research into CODIL actually started - and also why I had difficulty in promoting the research.

So how did CODIL begin. My chemistry Ph.D. involved extensive searches in the chemistry literature looking for patterns which could be related to a particular theoretical model. My first job as an information scientist also involved scanning large quantities of information, looking for, and reporting on, exceptional situations which might affect my employer's research and development program. In 1965 I moved into computers - working on one of the most complex sales accounting system that had been implemented anywhere in the world. After a year I was asked to look at the problems of transferring the contracts system (the most complex part of the whole process) to a new computer with (by today's standards) very crude online facilities. At that date very few people had any real experience of online working and I approached the problem with the experience I have gained looking for patterns in complex manual information systems. Without thinking I worked with the assumption that the world of contracts is too complex and too dynamic to be accurately and economically predicted in advance. What is needed was a system which will work symbiotically with the sales staff  and I suggested how it could be done.

Unintentionally I had stepped outside the society of "computer specialists" who thought they were very clever because they could make computers do things that ordinary people could not do. My boss was expecting a rule-based system - because that is how everyone knows computers work. What we need, he said, was a comprehensive explicit definition of all the valid contract structures - and I had come up with an contract description language based on how the sales staff viewed the problem - leaving the systems analysts and programmers with comparatively little to do. This was obviously ridiculous.I didn't realise it at the time but what I was suggesting was a pattern rather than a rule based approach - because I was used to looking for patterns in complex open-ended real world situations. I was told the idea was "research" and I should forget it - and under normal circumstances that would be the end of it. After all it is easier, and far more profitable career-wise, to conform with the people around you expect ...

However at this point I was offered a job by a computer company to do market research on the next generation of computers in an environment where new ideas were welcome. It rapidly became clear that many large computer systems had similar problems and I realised that it might be possible to generalize my earlier idea to provide a "white box computer." The proposal was to build a central processor which had a user-friendly symbolic assembly language and could work in partnership with humans to handle a wide range of opened real world tasks. The idea was actively supported by the UK commercial computer pioneers David Caminer and John Pinkerton until a company merger to form ICL resulted in the sales division being closed down!

In fact the work continued for some years - but in a computer environment - which proved to be the wrong environment for blue sky research which questioned the foundations of computer science. In his article Alex later says:
Our culture and the habits of our society are social contacts, and both depend primarily upon social learning. As a result, most of our public beliefs and habits are learned by observing the attitudes, actions and outcomes of peers, rather than by logic or argument. Learning and reinforcing this social contract is what enables a group of people to coordinate their actions effectively.
 The more successful something is the more people will go along with it - and in terms of new ideas affecting society there can be very few, if any things, more successful than the computer. We have now reached the stage where virtually everyone knows that computers are black boxes which have to be programmed by very clever people. By now most adults of working age will have been taught about computer programming - even if they never succeeded in doing it. And of course the "very clever" computer experts will control the funding of research. You might think (in an ideal world) a user-friendly white box pattern processing machine would be an advantage over the current user-unfriendly black box rule-based stored program computer. However any individual making this suggestion will be going against the strongly held society view of the most successful technology ever. Because my research was done in a computer environment I was surrounded by people who were strongly committed to the society view of computers. If I could have my life again one of the lessons I might want to retain is "Don't have unconventional ideas - because asking questions which do not conform to society's preconceptions can be a very dangerous thing to do"

Of course now that I have retired I no longer have to be a copycat. I am free to do the blue sky research which was impossible when I was working in the rat race for funding when I had to earn a living in the computer industry. Now I far better understand the theoretical aspect of what I had been doing in term of how the brain works on one side - and how in mathematical terms the ideas can be related to the theoretical aspects of computer science. Simply by re-examining my past work I can see how my human-computer language can be modified to map onto a network of neurons and it should be possible to explain how simple and logically not very sophisticated processes at the neuron level are sufficient to support powerful language and intellectual ideas as long as there is an effective copycat speed learning mechanism and enough brain capacity.

But of course I am an individual - and Alex is talking about the death of individuality. I am now free to think what I will, and to explore mental paths which were not open to me when I had to conform to earn a living. But society behaves as if original ideas only arise if the originator is trapped within some narrow specialism, is working in a well established academic box, can easily get his work peer reviewed by the establishment experts, and  has plenty of funding. The problem is that I am now outside every box that society looks in to find new and original ideas. I am 76, no longer have links with a university, and my only resources (apart from some 20 years of research data from the 1970s and 80s) is a personal computer, a small pension, and what I can find on the internet.

Alex Pentland has published a book with the subtitle "How Good ideas spread" - but his article describes a mechanism that means that individual ideas which do not conform to society's expectations can just as easily be lost.

Wednesday 2 April 2014

Relaxing with another Limerick

The boastful old man was aware
That the top of his head was quite bare
"It should be quite plain
That the size of my brain
Means that no room is left for my hair."