great question!
I have only ever seen a similar question asked in SWRPG, as few things in SWMinis have a decoy ability
(yet, in principle, the question could also apply to reserves/reinforcements).
So, I have the 'long' answer, and a shorter answer concept;
Long answer style;
When random counters are placed in Tannhauser or W40K, it's usually best for the arbiter/supervisor/gamesmaster to jumble and place
the tokens, rather than the player themselves.
Some folks like to 'roll a check for decoy' using a reflex check or perception check; in SWMinis, it could be a 'check' imposed based on the ATK value? Atk vs defense of the token?
You could use a random pigeon hole principle for the 9 square radius around the 'token';
assigning a 'square value' to a mini pseudorandomly (if the replacements are arranged into alphabetical order, or via cost from highest to lowest etc)
you could then roll a number of 'best of' dice (say, the placement of the token is the aggregate of the best of 3 for each figure) on a 2 D 6 table (or, any preferred method to produce the table of 9, if you prefer the probability distribution of some other die)
there are loads of Dice Apps that'll quickly let you do that, and can produce the results to an eigenmatrix (which you'd make as a 3 x 3 array).
that said, you could then also have a random assortment of tetris-like pieces,
and each piece is randomly chosen - the number of pieces required exactly matches the number of pieces brought in,
and the endpoints are subtracted (because an 'end' to the tetris piece will converge and overlap on the token square, though you could have fun combinatorially arraying that stuff)
- that'd be the "fairest" way to do stuff, so as sheer stochastic chance (biases of different systems notwithstanding) determines where those sudden reveal characters are.
to the second part : "choose random distribution of decoy counter at time of setup"
a couple of places have you covered
: The Electronic Journal of Combinatorics
http://www.combinatorics.org/and OEIS (formerly Sloanes) at
www.oeis.orgthe map is a 24 x 32 ish array, each gridsquare has a value you assign to it (and that value can be assigned pseudorandomly each game or each occasion it is required - if there's not already an app for that, there oughta be)
so, you can then do a "Markov Chain Monte Carlo" value assign method to randomly choose some squares,
you can literally do a slice,
you could apply random overlay shapes (which then have a 'choose' array per shape, and assign random values within the shape or from the total per board required)
this is less daunting when you take the limits approach, it's only x,y,z ! worth of places it could be hehehe.
Having experimented with these in the past, it can be pseudorandom and creepy
(especially if the GM is being 'impartial' towards the players, and has a lot of random checks to make)
The OEIS can be used, in a "Library of Biblioteca" style,
to pastiche randomly (MCMC style), through intervals of many different sequences;
if you've ever 'spoken' to Ramona AI or Evie Existor "AI", DeGaris contends that's primarily the method at play.
Short Answers
to reveal the characters, place the minis in a bag, and jumble it around a little,
then deploy those clockwise to the table (starting from closest to the enemy table edge to the farthest, whatever method you'd like)
to distribute decoy counters;
'blind' pick the locations?
toss the counters into the air a number of times per each 'pick' of the token (that one was the idea of an irish friend); you keep one of the tokens in the spot that looks random-y, and you clean the rest off and rinse-n-repeat
I hope this is of some use to you,
thanks for asking an AWESOME question