How does asking an procedure to run in ask patches produce a different output than running the patches as ask patches?

Question

I was working on a NetLogo project for school, when I ran into this issue with my NetLogo code. When I first made the code, I put the procedure death-check in the ask patches command with this being my program:

patches-own [num-live] 

to setup ;;this sets up all of the code
  clear-all
  ask patches [
    if random 100 < percent-alive [
      cell-birth
    ]
  ]
  reset-ticks
end

to go 
  ask patches [
    set num-live count neighbors with [
      pcolor = yellow
    ]
    death-check
  ]
  tick
end

to cell-birth 
  set pcolor yellow
end

to cell-death 
  set pcolor black
end

to death-check 
  if num-live < 2 or num-live > 3 [
    cell-death
  ]
  if num-live = 3 [
  cell-birth
  ]
end

And I got this output

Now if I put the ask patches inside the procedure, it comes out with a different output. Here is what I changed inside the code:

to go 
  ask patches [
    set num-live count neighbors with [
      pcolor = yellow
    ]
  ]
  death-check
  tick
end

to death-check
  ask patches [
    if num-live < 2 or num-live > 3 [
      cell-death
    ]
    if num-live = 3 [
      cell-birth
    ]
  ]
end

With this being the output

What I was wondering was why did this small change make such a big difference for my program?

Answer 1

The difference here is that in one instance, you are calling ask twice, whereas in the other you are only calling ask once. Consider two examples, which I'm calling go-1-ask and go-2-ask:

go-1-ask:

to go-1-ask 
  ask patches [
    set num-live count neighbors with [
      pcolor = yellow
    ]
    if num-live < 2 or num-live > 3 [
      cell-death
    ]
    if num-live = 3 [
      cell-birth
    ]
  ]
  tick
end

go-2-ask:

to go-2-ask
  ask patches [
    set num-live count neighbors with [
      pcolor = yellow
    ]
  ]
  ask patches [
    if num-live < 2 or num-live > 3 [
      cell-death
    ]
    if num-live = 3 [
      cell-birth
    ]
  ]
  tick
end

All I've done relative to your code is nest the code from each different death-check examples within the go procedure.

In go-1-ask, you are asking all patches, in a random order, to set num-live then to evaluate it. So, patch 43 will run the code, then maybe patch 71, and so on. In contrast, in go-2-ask, you are first asking all patches to assess their num-live. Only then, once all patches have performed that first check, are you performing the death-check. In short- the order of operations is different between the two methods, which accounts for the difference in behaviour.