Quote:

Now mind you this copy was digital, meaning it should be virtually identical to the original.

---

Quote:

They checked all of the connections and the impute voltages and everything looked fine.

---

OrgoneConclusion said:

(No source but my memory)

Years ago, some programmers did an experiment with a PAL (Programmable Logic Array - A PAL is a board with numerous logic gates that can be arranged in a zillion ways.) in a computer. I forget the specific goal, but the methodology was reasonably simple:

1. Apply a fixed series of imputs wherein the desired results are known.

2. Randomize the PAL.

3. Meausure the accuracy of the outputs. (Naturally they will be way off initially).

4. Create 16 copies of the configuration.

5. Randomly select one (or some small number) of gates to be randomly changed (mutation).

6. Measure the results of all 16 and keep the best (even if the original was better!)

7. Repeat steps 4 through 6 until no more improvement is shown after X number of iterations.

After some experimentation, the programmers got the result that they were looking for.

Then they made copies of the PAL and stuck them in another computer and the board failed to work. Now mind you this copy was digital, meaning it should be virtually identical to the original. They stuck the copy in the original computer and it failed to work. They checked all of the connections and the impute voltages and everything looked fine.

I will leave the answer of the puzzle until later in case sharp minds want to attempt to figure this one out. I will answer basic questions as best I understand them.

The implications of the answer and the genetic methodology are most interesting to the main topic being discussed of late.

---

Quote:

1. Apply a fixed series of inputs wherein the desired results are known.

---