MOPEKS - The Future

MOPEKS^® The Future

"Making predictions is very difficult - especially about the future" Attributed Variously

General

In general, new, original and exciting ideas receive a very negative reception. On the other hand, old, unoriginal and boring ideas also tend to be treated with hostility. So, statistics are not on my side and the chances are that the future of MOPEKS will not really be affected much by the initial public reaction.

So, the future of MOPEKS, if any, is likely to be in my hands. So what are my plans and expectations after that depressing statement? Well, there are three aspects to this, namely, short term, medium term and long term.

Short Term

By which I mean to the end of 2014

The Launch

The launch will be very low key. I will do everything possible to avoid drawing attention to the site, short of hiding it completely. That way I will be able to respond to any problems encountered by early users. If there are no early users then maybe enough publicity to attract a handful!

Housekeeping

"If debugging is the process of removing bugs, then programming must be the process of putting them in" Edsger W. Dijkstra

Anybody who has ever written a big and complex program will be aware that programming errors come with the territory. MOPEKS is no exception. I have done my best but have no doubt that there are errors waiting to be revealed.

Modifications

Anybody developing a new product for launch (whether hardware or software) knows that there never comes a time when everything works perfectly and every conceivable improvement has been made. So, the launch date is a compromise between the desire to get it out there on the part of the management and the desire to add just one more improvement on the part of the engineers.

In the case of MOPEKS I am the management and the engineers.

Kinetics

This is the 'K' in MOPEKS. The intention is that in the Darkroom you will be able to turn film, video clips or a series of still images into Dynamic Trials in a real life environment and hence derive the mathematics (ie Methods) involved in such a process. This is currently under development.

Possible Minor Program Improvements

The ability to retrieve future value by using something like 'ax <== NextStep(R.Xaxis)'
Have a standard range of colours
Have a universal storage area for variables accessible by all routines
Enable users to change the Breeding Pool size
Be able to modify properties of the Environment during steps of a Simulation
Allow statements of the type 'if ax < Method17(R, S) then goto 5'
Allow users to locate data in a different area of their hard drive
In the Library, down and up arrows should change the entry highlighted
In the Simulator Setup form allow the user to use Methods rather than simple criteria to determine Success, Failure and Rewards
Give users choice of number of copies to run when finding a Method rather than current choice of one or the number of processor cores

The Website

I have spent a lot of time on this but there is still plenty of room for improvement. Currently, it runs best on Chrome (54% of market) and Firefox (27% of market) but it needs to work better on Internet Explorer (12% of market). Opera, Safari and other Browsers which have 7% of the market between them, will just have to take their chances!

The above percentages are for October 2013 and are taken from stats provided by a site which caters to computer programmers (who are likely to be a major proportion of readers - assuming there are any). See 'Browser Stats' for the latest position.

Medium Term

I am taking this period to be from January 2015 to December 2017

Possible Major Program Improvements

Abilty to create, divide and destroy objects during the course of a Process or Simulation
Increase number of permitted objects from 7 to maybe 16 or 32

Open Source

If enough people are interested I may well make MOPEKS open source

Long Term

January 2018 onwards.

"In the long term we are all dead" John Maynard Keynes

Possible Profound Changes

Currently, MOPEKS is very much controlled by the user but there is no reason in principle why it should not start to take a larger role in its own destiny.
Extend the role of Kinetics in MOPEKS so that it can try to make sense of the real world in its own time.
Make it mobile, with GPS
Turn a stereo image into object descriptions
Extension of MOPEKS beyond Numerical Mathematics - see below

Extension of MOPEKS beyond Numerical Mathematics

Currently MOPEKS generates Methods which are based on numerical mathematics. In other words numerical values eg 28, 37.624523 or 1000. It can, in principle, find a Method to fit a set of numbers which may, for example, be the distance between two Objects or the square root of a series of numbers.

At its most sophisticated it finds a mathematical Method which enables an "Intelligent" Object to achieve its objectives eg escape from a predator. That method will consist of precise numerical values eg walk for 10 steps directly towards Object A, then run 5 steps at an angle 39 degrees to the North etc.

In every case, the Method relates purely to numbers.

The next step beyond numbers is something I am having difficulty grasping but maybe the principles of MOPEKS can be extended to entities such as words or something beyond logic and numerical values? For example, in MOPEKS itself, you may have noticed that if you input the name of a MOPE, MOPEKS automatically abbreviates it down. So, for example, the Object Name "Long Named Object" is abbreviated down to "LngNmdO". Obviously this is done by a computer program (which I wrote)

Similarly, other people have written programs that can solve crossword puzzles, play scrabble or find anagrams of people's names eg "I am on a march" is evidently an anagram of "Chairman Mao". It has also been pointed out that "useless twat" is an anagram of "Ed Miliband" - but only if Ed Miliband types it.

[The anagram of Chairman Mao is especially clever because Mao was famous for leading the Long March. He lost power when he decided to allow a third of the population of China to starve to death (400 million people) so that they could catch up with the USA. This was regarded as a step too far. But you knew that already.

Ed Miliband is famous for Stabbing his Brother in the Back to gain leadership of the UK Labour Party and being generally useless - apart from being able to stab people in the back, of course. But you knew that too.]

Word based MOPEKS

So, let's explore this idea. Take the word 'EXAMPLE'. This consists of 7 separate letters but we could express them as Objects, namely:

Object1.Handle = "E"
Object2.Handle = "X"
Object3.Handle = "A"
Object4.Handle = "M"
Object5.Handle = "P"
Object6.Handle = "L"
Object7.Handle = "E"

Er ... hang on a minute! Yes, MOPEKS can, in principle, handle word games in its exsisting form (subject to extension to cover more Objects). So, the abbreviation problem mentioned above, or the anagram problem, could be formulated as a Dynamic Trial to find a Method which will carry out the process of moving and changing these objects - as in the animated gif below. This only shows one trial but you easily make up a few more on the same basis.

[I say in principle, because as currently written MOPEKS could not realistically tackle this problem. MOPEKS is really a demonstration of principle - bear in mind that the first airplane had no toilets.]

No. I honestly did not realise that when I started writing this section. Rather in the way that I had not realised that MOPEKS incorporated some of the fundamental aspects of Object Oriented programming until I began to write a section on the subject (see 'FAQ --> General --> Q4').

Really, of course, I should not admit that and pretend that this was all part of my grand plan. It wasn't.

Left click for ANIMATED gif. Then Save Image to Disk and Step through as a Movie if you wish - see 'FAQ ⇨ General ⇨ Q9'

To Boldly Go ...

Is it possible that computing can go beyond numbers and logic? Well, my first reaction was that it may be possible but then I decided it was an illogical idea! Surely, all computation must be based on logic and numbers?

Then I returned to where all of this started - the human brain. It clearly exhibits features of both an analogue and a digital computer. It is like a photon - sometimes it looks like a particle and sometimes it looks like a wave. In each case, trying to decide exactly what a brain or an electron "really is" is a waste of time. It is a "thing" that exhibits certain characterstics and carries out certain functions. That's it.

So do analogue computers use numbers and logic? Well, I guess they are logical in the sense that if set up in a certain way and fed a specific stream of data they will always produce the same output, subject to measurement error. They clearly do not directly use numbers - they use real world phenomenon which can be measured using numbers but that is not the same thing.

I have also read that the actions of any analogue device can in theory be emulated by a digital device. Whether this is a law or just a conjecture I am not sure. The ultimate analogue device is the real world. Could you simulate on a digital computer an explosion in a watch factory? Probably not with any degree of accuracy. The problem is just too big.

So, maybe the next major step forward for MOPEKS is to think in terms of analog computing and if, or when, a methodology comes to mind, emulate it using a digital computer if possible. If emulation is not a realistic possibility then it may be back to the workshop.

In this context, here is something I wrote many years ago.

Words and some events are surfaces in "Meaning Space". They have weights which the brain can adjust eg in learning a foreign language. A good analogy is to think of rocks in the sea over which different shaped seas sweep and are changed in shape. This may well iterate as the shape of the "rocks", ie "the world model", changes.

Quite how you would implement that I really don't know.

Rough Sets and Fuzzy Sets

These may also be relevant in the hunt for a future route of exploration. Clearly, much data in the real world is imprecise and any technique which copes with such imprecision may be relevant.

There is a good tutorial on Rough Sets, here and an academic paper by the originator of the concept, which also deals with Fuzzy Sets in passing, here.

The general idea is to take a collection of pretty rough data eg medical records and find a discernability function which specifies the minimum set of attributes necessary to differentiate each class of outcome.