This is a working document in which we will explore the idea of building (or otherwise manifesting) a self-replicating machine.
Maybe go back to the word automata rather than machine – it is more descriptive
Motivation behind the investigation
The plan is to build a Reprap. The approximate bill of materials (bom) is about Euro 500.00, which in relative terms is a large amount of money, to test out an idea. So before we go ahead to build the Reprap we need to look ahead to see whether the end result will meet our expectations. What can we reasonably expect from such a machine? To answer this question we need to review what the intention behind the Reprap is and more generally the intention of self-replicating machines. (the simplest self-replicating machine would be a plant
Definition of a machine
Lets define a machine as an object or construct that takes an input (or maybe no input) and produces an output by converting the output or maybe leaving it the same.
Then lets define a self-replicating machine as an object or construct that takes an input (or maybe no input) and converts or creates a copy of itself as the output.
Essential elements of a self-replicating machine
Firstly the essential elements of a self-replicating machine:
1.Replication triggered by environmental condition (machine intelligence - replication by itself is not a sufficient criteria).
2.Must be able to access raw-materials
3.Build components from raw-materials
4.Assemble components into an autonomous replicate of itself
So next we need to develop a syntax that that will contain the essential verbs and nouns to describe the process of self-replication:
ENVIRONMENTAL_IMPULSE triggers MACHINE (to replicate)
MACHINE gathers RAW_MATERIAL
MACHINE constructs COMPONENTS from RAW_MATERIAL
MACHINE assembles COMPONENTS into MACHINE (autonomous)
Questions that arise
1.When is something a raw material and when is it not - this will be important when looking at the Reprap - Is it good enough to say that a raw material must just be less complex than the machine that will be constructed from it. Is a motor a raw material, is a integrated circuit a raw material. These are complex items which would be difficult (virtually impossible) to manufacture.
2.Do we need to construct and assemble components or can we live with one component that constitutes the fledgling machine - this will be a consideration when dealing with computer applications?
3.Is an ability to respond to the environment a necessary criteria?
4.There is also the question about the environment - what does the machine need to live - again this is crucial when looking at at least at computer applications.
Identifying Examples of Self-replicating machines
So the next process will be to identify self-replicating machines and to see whether they fit into our syntactic mould:
1.Some biological example - this is presumably where the idea behind self-replicating machines originated - so I am not going to go into any details here.
2.Capital – (odd-ball)
3.Computer application
4.Virtual machine (Uli's Interest)
5.Reprap (Rudi's Interest)
There was another one that I have not really thought about - music - a melody within a melody and maybe even dreams within dreams ... Hmmm maybe look at those a bit later
Sort of "Goedel, Escher Bach" stuff ... paintings within paintings etc.
Self referential is somehow linked to self-replicating
But for now back to the syntatic mould. So lets start with capital (seems the most outrageous): So ...
Example One: Capital
CAPITAL takes PRINCIPAL_AMOUNT
CAPITAL earns INTEREST form PRINCIPAL_AMOUNT
CAPITAL makes INTEREST available_as CAPITAL (autonomous)
So from first looks would seem to meet the criteria - also has intelligence as it responds to the environment. Hmmm here need to define the environment. Has a tricky part though - the raw material which is the principal_amount is indistinguishable from the actual machine ... will need to come back to this later.
Example Two: Reprap
Rudi where are ye! I need your input - time to jump to the reprap - you tell me why I think this does not fit into the self-replicating machine mould ...
Example Three: Recursion?
Then I will tell you why I think recursion does not fit the mould ...
So we have defined a machine and also a self-replicating machine. We have not talked about the nature of the inputs in relation to the outputs. But they can be either more complex or less complex. In general we will concentrate on machines that convert less complex objects into objects of greater complexity. Save to say it is just as important to break object doing to their basic consituents as it is to build them. However in general self-replicating machines will be the former.
So now, can this machine be a thought or a concept? That is in terms of inputs and outputs?
Can a thought result in an output? Do we need inputs or can machines create outputs out of the blue?
This is somewhat allied to the idea of a catalyst. Is a catalyst a machine. In real terms it just hangs around. But it does result in a transformation or speeds up the transformation.
So a thought can trigger an emotion - a real physical manifestation - happiness or sadness.
Lets look at our virtual machine example. It exists in the mind of the computer. In essence it is defined by a set of instructions executed in some order.
However the virtual machine is really just a paradigm shift. It is just a redefinition of the operating system / application boundary.
Also the virtual although it does not have any real - can manifest through the hardware of the underlying host machine/operating system.
So a self-replicating machine most probably does not need to meet all the afore going requirements. In other words it does not have build components and then assemble them. All it has to do is to make a copy of itself using raw materials. We don't care how it does this.
Bibliography
(We will start comparing our thoughts with what is already out there in the literature)
Oh yes and still need to cobble together some references:
1. http://en.wikipedia.org/wiki/Self-replicating_machine
2. http://staff.bath.ac.uk/ensab/replicator/Downloads/report-01-04.doc
3. http://www.3dreplicators.com/ - I like the quote that goes with this one
