The quick are the dead



Picking it up from yesterday’s entry, I needed to confirm that the quick populations, those that produce new figures quickly, were as fragile as I thought, and for the reasons I thought. I started with m1.

There are 20 figures in m1. I thought that there would be one or two figures that when removed, would cause the entire population to die out. When I tested it I found out it was much worse than that. Out of the 20 figures, only three of them could be removed without killing everyone off. Mind you, if I removed several figures from the population, it will recover from the injury, replace the missing figures, and go trucking along as though nothing happened.

By contrast, one of the slow populations, 61, remains stable no matter which figure was removed. There are 61 figures in that population. I removed the first 31, it ran fine. I removed the last 31, it kept going. I removed the middle and it was still fine. I could remove all but the last three, or all but the first one and likewise, the population just kept making new figures, stabilizing at a new population size. 61 figures are too many to go through and remove one at a time, so I set it up to remove one at random and ran it several times. Most of the time, the population would replace the missing figure. Once or twice, the population dropped to 60 and staid steady at that size, rather than growing one more and stabilizing at 61.

I tested another quick population with a pop size of 20. This time it behaved more like I had expected. If you remove one figure, it will survived, replace the missing figure, and continue, except for two figures. If you remove either one of those two figures, everyone dies. It goes a long way to explaining how entire populations were getting wiped-out by a single mutation.

To be certain the trend holds, I tested a couple of other fast and slow populations against one another. In every case, the slow figures would survive more mutations on average. None of them were as spectacular as 61, but the trend is holding. The quick are the dead.

As the results show, not every slow population is as robust as 6 and 61. It’s not just how slow you are, but what kind of slow. Keep in mind that “slow” only refers to the rate at which a given population is creating new figures. For a given population at a given size over a given time, the quick and the slow execute about the same number of instructions.

I’ve a notion of exactly what kind of slow survives the best and why. Thus far, everything is in line with my reasoning, but to be certain will require a few experiments. This includes attempting to develop a method to generate a mutation resistant population deliberately.

Other projects and obligations are looming. I’d love to finish this up tonight or tomorrow, but I’m making no promises.

Here are the results. There are two sets, and the fast population is first, followed by the slow one it was being compared to. There are reasons to compare the ones I did with one another involving which experiment the given populations came from, but the details aren’t that important.

Set 1.
all1.pop: most 36, least 4, average 12.3

1:
mutations 15

2:
mutations 36

3:
mutations 9

4:
mutations 10

5:
mutations 4

6:
mutations 4

7:
mutations 9

8:
mutations 14

9:
mutations 7

10:
mutations 15

all3: most 2765, least 15, average 401.6, adjusted average tossing out 2765 139
1:
mutations 342

2:
mutations 23

3:
mutations 32

4:
mutations 220

5:
mutations 92

6:
mutations 234

7:
mutations 2765

8:
mutations 73

9:
mutations 220

10:
mutations 15

set 2.
skipsnap10: most 28, least 1, average 7.6

1:
mutations 14

2:
mutations 10

3:
mutations 6

4:
mutations 28

5:
mutations 1

6:
mutations 1

7:
mutations 5

8:
mutations 1

9:
mutations 9

10:
mutations 1

skipsnap7: most 61, least 2, average 26.9

1:
mutations 24

2:
mutations 23

3:
mutations 14

4:
mutations 61

5:
mutations 50

6:
mutations 54

7:
mutations 14

8:
mutations 5

9:
mutations 2

10:
mutations 22


Leave a Reply