Angry Birds Short Fuse Level 27-12 Walkthrough

always had known since i had begun playing angry birds free that patterns of self-desruction were the same upon restarting the game and going to the level directly, in that case angry birds free level III-2. i thought this had long been addressed by rovio, until ernor63 pointed out rovio hadn’t (for level 26 6, see the comments there).

my idea of what is going on is that some random number generator (if you’re not familiar with programming, think of a random number generator as a built-in dice, with the shared property that the next value should not be predictable from the previous ones, but a difference being that in the case of random number generators the sequence of outcomes after reset is determined completely by some initial value, the seed) is used to construct levels. this random number generator has a fixed seed value upon program restart. moreover, this random number generator is called a fixed, large (maybe linear in the number of elements in the level?) number of times upon level initialisation and the generated random numbers determine small variations in the placement and properties of the elements within the level, and in that way also determine whether the level (or part of it) will then self-collapse or not: after initialisation subsequent behaviour of standard elements is completely fixed since the underlying physics engine, box2d, is deterministic.

under the assumption that the above hypothesis is correct, it was reasonable to also expect that going to another level and leaving it immediately has a fixed effect upon the random number generator, in the sense that it is then also called a fixed number of times. the point is that probably this number is different from the one for just reloading (the level one is interested in playing), since probably both levels have different numbers of elements in them, bumping the random generator a different number of times in this way, leaving it in a different state, leading to different subsequent sequences of generated random numbers. in this way, this probably leads to different self-collapse patterns, which is a good, thing when one is exploring, trying to find self-collapse sequences (always trying the same sequence does not bring anything; the pattern will always be the same).

this then led me to do the experiments described above for finding self-collapsing sequences; i just methodologically tried restarting ab, going to one of the sf-levels, leaving it, going to 26-12 and reloading it a small number of times to see whether it would lead to self-collapse after each. this worked fine, in that it produced several useful partial self-collapse sequences, but many other experiments could be done as well; i just did the simplest ones i could think of.

i hope the above makes sense. but irrespective of that: one can try this method for any level where one thinks that (partial) self-collapse brings an advantage, but all this may be very device dependent (as was pointed out by ernor63), as probably the random number generator is dependent on some api which is subject to change at any moment, and guarantees only a distribution of the numbers generated or something such, but certainly not the concrete sequence that will be generated upon ab restart.