k; now I'm going to break down the rough justifications for doing this technique.
If the probability of executing a set of techniques is locally independent; that is to say, given that learning and adaptation from preceeding attempts at completing a technique a certain small number of times does not increase the probability of later attempts being successful
*, the odds of completing a string of techniques wherein one would succeed 90,85,80,...,65,60% of the time are presented in the following table, where columns represent a set of repititions of a given set of techniques with a given % probability, rows represent repetitions. The values of the table are in percentages. In mathematical terms, and
you don't need to understand the rest of this sentence to proceed, this is the application of the law of total probabilities to the conditional probability of player success, given the hypothesis that repeated successes are locally pairwise disjoint, and dependent on a probability function who's terms are purely the net ability of the player for the duration of the attempts.
(* In reality, learning is literally in play, so the individual probabilities are actually arguably higher, as the odds of later success covary with preceeding successes when viewing a large sequence of events. This is due to player adaptation, but with a small enough number of attempts the effects of these adaptations are minimized; metaphorically, we're looking at trees, not a forest)
Code:
95.0000 90.0000 85.0000 80.0000 75.0000 70.0000 65.0000 60.0000
90.2500 81.0000 72.2500 64.0000 56.2500 49.0000 42.2500 36.0000
85.7375 72.9000 61.4125 51.2000 42.1875 34.3000 27.4625 21.6000
81.4506 65.6100 52.2006 40.9600 31.6406 24.0100 17.8506 12.9600
77.3781 59.0490 44.3705 32.7680 23.7305 16.8070 11.6029 7.7760
73.5092 53.1441 37.7150 26.2144 17.7979 11.7649 7.5419 4.6656
69.8337 47.8297 32.0577 20.9715 13.3484 8.2354 4.9022 2.7994
66.3420 43.0467 27.2491 16.7772 10.0113 5.7648 3.1864 1.6796
This is to say, for example, if a player tries to execute a technique, or sequence of techniques, where their base-line average % success rate is 85% (like L-canceling), they will succeed at doing an l-cancel 5 times in a row only 44% of the time (85% column, 5th row). This model does not take into account tilting, pressure, or the psychology of a match, so it is not accurate for tournament conditions, but for solo-practice I personally find it does well enough as an indicator of whether or not I
actually am in conscious control of my character. This model also does not account for variable success-fail conditions, for example, a frame perfect wavedash vs a wavedash which is dirty, but functional. In this regard the overall model requires the player to have clear-cut conditions for success or failure, and is therefore less-than-great for situations with perceived leeway (dash dancing away can be a good example, a lot of players don't have a set basis for a 'successful' average speed across the stage, vs a slow clumsy drift of the dash dance 'box').
However,
the net conclusion one should reach from that table, is that the difficulty in completing a set of actions a certain number of times has to do with your general aptitude of being able to perform that set of actions once. If you only sort-of know a technique or set of techniques, even if you think you know it well, it will show under attempts at repetition alone.
Picking the techniques at random provides a set of conditions to which the player is not previously acclimated, and forces the player to be able to block (merge into a single motion or thought) only the exact movements required. In the world of percussion, exercises like these are generally known as a form of 'gridding', mixing rudimentary elements of playing into a set of variations. Here is the US Army Hellcats snare line doing a grid exercise of a single hybrid rudiment (the herta), note that the tolerances for variances in timing for each member are less than 1/60th of a second, as that audibly makes the sound of the whole line, with the given instruments, as being dirty (****ty, bad, flammy, unpolished, unprofessional). Should anyone need evidence I can personally electronically produce a sound file containing this exercise played by two drums, with and without random timing variations of 1/60th of a second, for reference, and edit it into this post.
EDIT: there are times where, unfortunately, the Hellcats actually do flam their warmup; the instruments they normally play on are traditional marching snares, vs Scottish-style marching snares (modern drum corps snares) for which their drum-pads are suposed to emulate. Traditional marching snares distribute the sound over a longer period of time (have more sustain) and therefore can be played less accurately. As such I've attached an additional video of Santa Clara Vanguard warming up to present their higher accuracy as they play modern marching instruments.
The point is to break up the patterns of what you're working on until they are purely the atomic elements which require execution, and varying linking in-and-out of those atomic elements, until one can move freely around their entire option space at will, efficiently, and as required. The cerebellum, being a set of neural connections which seek to streamline activity as simply as possible, will chunk together sets as practices as their own distinct total acts; by challenging the mind to execute sets of these components at random, a player usurps this tendency to group the entire habits of practice, in favor of the desired ability to act out these technical components naturally, in any situation that may arise, without any need of extra thought on execution or concern of success.
Example strings include (generally the parts of these should be selected at random unless personal notes from a tournament or player give you a specific combination to work towards):
- Dash, wavesurf forward, dash, pivot ftilt.
- Dash into run, cactuar dash back, jump dair (l cancel)
- double shine into dair (l-cancel) into shield.
- dash shieldstop frame-perfect wavedash (oos) back, dash back shieldstop frame-perfect wavedash back, repeat changing directions, paying attention to accuracy of intended player position vs resultant. (HARD MODE: have friend(s) or (a) spamming CPU(s) attack your shield in a given position (or positions))
- Dash short hop turn around laser (without accidentally pivoting or analog jumping backwards) repeated any number of times with specific pre-determined laser heights or variations thereof.
More complicated ideas include:
- dash dance (no accidental runs or standing turn-arounds) until a friend yells go! at random (they can do this over voice chat), then combo a waiting 20XX 4.05 STAY Fox cpu (or an old 20XX lvl 1 cpu) to at least 50% (have a % condition of success), ledgestall 3x (bonus points if fully invincible), repeat (use 20XX save states to reset the cpu %s). Try this with other CPUs too.
- dash dance/platform waveland while trying not to get hit near ledge until new 20XX cpu grabs ledge, wavedash back ledgehog, repeat until you get used to getting kills on stalling cpus of varying types, and the means of bating them.
- Playing melee Horse with varying previously-stated (OUT LOUD) techniques, with a practice partner or friend, or even live-documenting the activity over skype/video chat. Set modifiers to the game of Horse, such as a number of repetitions required, or 'someone must lose' mode where the repetitions or acts are attempted one after the other until someone fails.
- in front of a friend, try to do a set of different options out of shield or at the ledge, and have them judge your comparative unpredictability.
- in a similar vein to the above, practice shield pressuring while mixing in varying options, including grab, against either a friend or a 20XX 3 cpu. If you're doing this alone, and can record your own gameplay, watch back at what you're doing and take notes at your own perceived predictability and which options beat out what you're doing.
But to sum up:
The way you practice creates a set of habits of both what you want to do, and how you do them. By fighting these habits you improve your innate reflexive knowledge of the game, as well as your ability to achieve what you've learned otherwise.
This is for individual practice though, and as a general rule, principles of gameplay are best tested in friendlies or in bracket, and not to be based on ones performance in an individual practice or analysis setting, regardless of the attempts made to best use that time.
