Please don’t misunderstand me and conclude that I continue to follow this thread; I don’t, but I thought I’d just post a recent conversation that I had with the time traveler (TT) in the museum:
TT: OK, that was very informative, thanks, but I really must be getting back now.
Me: Well, I'm just pleased I could be of help, showing you, by example, how machines have developed very similarly to animals by incremental improvement over time.
TT: Actually, before I go, I don’t really understand how the design development process for the Mercedes car happened, that makes it analogous to evolution. It seems to me that there could have been some intelligence involved with forethought and intent, not to mention the ability to select beneficial changes over disadvantage changes artificially. Doesn't hat lead to a definite distinction between the two processes?
At this point in the conversation I decide, in the interests of simplicity and comprehension, to introduce a few white lies into my explanation. It goes like this:
Me: Well TT, what happens is this:
There’s a Mercedez factory down the road where the very earliest car was made (I’ll not go into detail about how that first model came about; nobody can really answer the same question about animals). Now the factory is run by two different types of very simple beings; one type assembles the cars from instructions (let’s call it code, if you like). These types are called ‘assemblers’. The other type occasionally make changes to the design. These types we’ll call ‘engineers’.
TT: So, do the assemblers think much about what they’re doing, and how they could assemble the cars differently, or better, for example?
Me: Oh no, they just blindly follow the code time after time, turning out the same car model exactly like and in the same way as the previous one.
TT: And what about the engineers; don’t they give any thought to the occasional changes that they make to the code?
Me: No, they don’t. They use a random number generator linked to the code which determines which part of the code to alter, in what way and to what degree. I guess it generates three separate factors somehow that are independently applied. Incidentally, not every part of the code can be changed in any way whatsoever and by any conceivable degree. The Random Number Generator is set with limits such that radical changes are generally not allowed, which I believe is similar to how evolutionary change works.
TT: So, after these alterations to the model’s code are made, and the assemblers start making the changes, what happens then; how does the company determine what alterations are good and what are bad?
Me: Well, the cars are simply sent out to the showrooms, just like the unaltered ones were, for sale. People come into the showroom and compare the new car with other cars out there in the marketplace. If they like the alterations, then that’s reflected in increased sales, and the production rate is increased to take advantage. If people don’t like the alterations then they tend to buy an alternative car that’s on the market instead. Sales fall, so the alteration is dropped and they go back to making the previous model and try a different randomly generated alteration.
TT: I see, so the success of the car over the long term is measured by its ability to survive in the marketplace against the competition?
Me: Yes, that’s right.
TT: OK, that all seems to make good sense, but it does seem a rather inefficient way of going about things.
Me: What do you mean?
TT: Well, presumably roughly the same number of bad alterations are introduced as good alterations. Is there not some way of assessing the likely success of the alterations before they’re put into production.
Me: Well, I suppose so, but how?
TT: Well here’s a thought: why don’t the engineers, instead of making random changes to the code, try to assess in advance what they think might be a good change, based on what they know about what people are looking for in a car, and only make those kinds of changes?
Me: I suppose so, but what will that achieve?
TT: Well, I realize that it will only end up with the same or similar results as just carrying on doing what they’re doing right now, but it would save a lot of abortive cost, and it would lead to a much higher likelihood of only increasingly beneficial cars being sent to the showrooms!
Me: You know what TT, you’re not as primitive as I figured you were. That’s exactly how Mercedez Benz actually develop their cars. They assess and predict what the future market requires and introduce alterations accordingly. They could, of course, just make random alterations and see what survives, as I described to you above, but they would soon be out of business.
TT: So design development really is no different from evolution when you look at it like that. The only real difference is that prior assessment and testing of possibly beneficial alterations to the code have been introduced essentially for convenience and commercialism; alterations that would not be necessary otherwise for designs to develop the way they do.
Me: Exactly TT. Hey, you’d better get going; you’re gonna be late for dinner!
OK; I'm outta here again.
