Status report on SL Research (2016-06-28)

# Status report on SL Research (2016-06-28)

## Playing RRR
#### (Board is BYR-BBR-RYR-PBR (positions 1 to 12); played positions are 369.)

Syntax: A - (B): C (D)

* A - Amount of cancelled tiles
* B - How the cancellation was achieved
* C - Position of triggers in played queue
* D - Conclusions

* 0 - (0): No
* 1 - (1): Third (Trigger in 4)
* 2 - (2): Second (Trigger in 4)
* 2 - (1+1): Second (Trigger in 4)
* 3 - (3): First (Trigger in 4)
* 3 - (1+2): First (Trigger in 4)
* 3 - (1+1+1): First (Trigger in 4)
* 4 - (whatever): No
* 7 - (whatever): Third (Trigger in 10)
* 9 - (whatever): First (Trigger in 10)
* 12 - (whatever): No
* 15 - (whatever): No
* 18 - (whatever): No
* 21 - (whatever): No
* 24 - (whatever): Second (Trigger in 26)
* 23 - (whatever): Third (Trigger in 26)
* 27 - (whatever): First (Trigger in 28)
* 25 - (whatever): First and Third (Triggers in 26 and 28)

Trigger positions (up to position 30): 4, 10, 26, 28

* The only thing that matters is the amount of cancelled tiles. How you reach that amount is irrelevant.
* In this case, the three tested characters had 25% trigger rates.
* The triggers are on a queue you navigate by cancelling tiles. Cancelling one tile moves you one step forward on the queue.
* The queue works the same no matter the direction.
* No obvious periodicity noticed so far.
* Playing RR instead of RRR uses a different queue (so the queue is tied to the exact amount of tiles played one way or another, possibly to their identity too).

/////

## Playing BBBB
#### (Board is BYP-BBB-RYR-RYR (positions 1 to 12); played positions are 1456; this experiment takes place right after the previous board.)

Syntax: A: C (D)

* A - Amount of cancelled tiles
* C - Positions of triggers in played queue
* D - Conclusions

* 0: No
* 4: No
* 8: First (Trigger in 9)
* 5: Fourth (Trigger in 9)
* 6: No (Expected Third, but Third wasn't a known trigger rate)
* 6: Third (Trigger in 9, confirming previous failure to be a matter of trigger rate and unreached threshold i.e. unit 5 on this board has a lower trigger rate than 25%)
* 7: Second (Trigger in 9)
* 12: No

Trigger positions (up to position 16): 9

* In this case, two of the four characters have unknown but probably high trigger rates (1 and 5). 4 and 6 have known, 25% trigger rates.
* Just like with the previous experiment, one cancellation moves the queue by one step. It looks like color doesn't matter here.

/////

## Playing YYY and GGG
#### (Board is YGY-BYG-PGR-PYR (positions 1 to 12); played positions are 153 (YYY) and 268 (GGG); testing whether the queue is the same for two trios.)

Syntax: A: C (D)

* A - Amount of cancelled tiles
* C - Positions of triggers in played queue
* D - Conclusions

YYY

* 0: No
* 3: No
* 6: No
* 9: No
* 12: Second (Trigger in 14)
* 11: Third (Trigger in 14)
* 15: No
* 18: No
* 21: Third (Trigger in 24)
* 24: First (Trigger in 25)
* 23: First and Second (Triggers in 24 and 25)
* 27: No

Trigger positions (up to position 30): 14, 24, 25

GGG

* 23: First and Second (Triggers in 24 and 25)
* 0: No
* 3: No
* 6: No
* 9: No
* 12: Second (Trigger in 14)
* 15: No
* 18: No
* 21: Third (Trigger in 24)
* 24: First (Trigger in 25)
* 27: No

Trigger positions (up to position 30): 14, 24, 25

* In this case, all units have 25% trigger rate except for position 3 (22%).
* It looks like only the amount of cancellations and the number of tiles activated influences the triggers. Which suggests there are nine trigger chains in existence (one chain per possible amount of tiles triggered). Requires testing with a board of nine 25% trigger rates of the same color.

/////

## Hypothesis

The chains start at the beginning of the board (possibly at the beginning of the first board) and navigate through the board. Each tile adds (or substracts) a certain value to the chain's value (for instance, let's say "Pink adds 4": if there are three Pink tiles in positions 1, 2 and 3 of the first board, the chain's value in position 2 is 8 and in position 3 it's 12). The chain's possible values probably are cyclic (like angles).

Next need is to experiment on starting boards and see whether two identical (or nearly identical) starting boards get the exact same (or similar) chains.