IRC Mini Tournament #20 - Sunday 13 July 2003

This time it wasn't a real tournament, rather some fun. The players were divided into two teams (actually there was an odd number of players, so one of them - Fizmo - was playing in both teams - he "was the jumper" as he said). Players were writing a warrior as a team, but no-one knew who he was teamed with. They even hadn't seen the whole code! They only knew the one instruction someone wrote before them. Generally, the position of the instruction in code was unknown, only the person who began knew his (her? no, there was no woman ;) instruction was the first one in the code (and the first executed). Using labels was forbidden.

As mentioned, five players took part (in alphabetical order): bvowk, Fizmo, Hurkyl, PhilB and willvarfa. In team A were: bvowk, Fizmo, Hurkyl; in team B: Fizmo, PhilB, willvarfa. Coding progressed very slowly due to handworking whole the 'send-receive-check' job (next time a simple IRC script may be useful), but after 6 instruction per team warriors seemed to be finished. They looked like below:

;name A Warrior
;author Team A
;assert 1

start spl   #235, 235   ; Hurkyl
      mov.i }-4,  >-1   ; Fizmo
      jmp   -2,   <-20  ; Hurkyl
      seq   @-21, <-21  ; bvowk
      jmp   -1,   <-22  ; Hurkyl
      seq   @-23, <-23  ; bvowk
      end   start
;name B Warrior
;author Team B
;assert 1

start spl   1,     <-1234 ; Fizmo
      mov.i -2,    >2     ; PhilB
      spl   -2,    >-10   ; willvarfa
      mov.i #3044, {3753  ; Fizmo
      djn.f -4,    #0     ; PhilB
      jmp   1,     {1     ; willvarfa
      end   start

As you see, Fizmo and Hurkyl wrote 3 instructions, bvowk, PhilB and willvarfa - 2. Looks a little unfair. Also the order of coding seems to be not so random and... democratic (?), but all of it was an effect of some random causes (bvowk had to disappear for few minutes and he had phone call and Fizmo was playing for both sides and... BTW it must have been confusing for Fizmo to think all the instructions were parts of the same warrior ;)

Warrior A is a simple clear, I think it's obvious. It could have been finished after three instructions, but players didn't know that, so they started to construct a... Q-scan (?) that was never executed (someone told it could have been changed by moving different instruction instead of JMP -2, <-20, but do you believe anyone would have done it coding blindly ;)

Warrior B is a... hell knows. Looks like double clear, but clears also with imp-like bomb, splits in a weird way, has two gates... According to what was said on IRC channel both warriors scores closely in wilfiz, that is 42 for Warrior B and 48 (?) for Warrior A. Someone correct me if I'm wrong, but that wasn't the aim of this tournament. Warriors created by teams A and B were fighting each other with standard pMARS rules (-s 8000 -p 8000 -c 8000 -l 100 -d 100) in a hundred fights (-r 100). And here are the results:

pmars -r 100

A Warrior by Team A scores 98
B Warrior by Team B scores 164
Results: 20 42 38

Which means team B won. Congratulations! And thanks!

Organizer of this crazy piece of fun:

Lukasz Adamowski