Scanning the oil is not only a difficult task, even if you could get the info 100% accurate, how would you take that data and convert it into a shot selection?Originally Posted by NewToBowling
What would be more effective is to start with a "systematic wild *** guess", and launch the ball.
Using CATS data, determine the actual path of the ball, and how it differs from the expected path.
Use that information to update the decision process on the next shot.
Also take into consideration, that last shot has now altered the pattern slightly.
After a half dozen "warmup" shots providing current data, combined with perfect recall of every shot made in the past, EARL's brain (if it had one) would be able to build a functional model of the oil pattern, and develop a plan of attack.
What EARL could do, that a human can't, is determine how much of a error in ball placement was involved in leaving the 8, 9, or 10.
Then stay in the center of the range that minimizes the potential of leaving the 8, 9, or 10.
A computer doesn't solve problems the way humans would solve the same problem.
A computer is programmed to take advantage of it's strengths, while hopefully avoiding situations that expose it's weaknesses.
Your answered your own question (see last sentence, kind of what I was getting at). Give EARL practice shots to find line. EARL would watch every other bowler to see how their balls react and how it's own ball reacted on previous tosses and adjust accordingly. With all that data input (along with data from all available bowling balls) I'm sure EARL can bowl close to 290-295 average because the laydown, speed, axis tilt, rotation, revs would be exactly the same (something humans can't replicate). 99% if not all spares (excluding crazy splits) would be converted with spare ball.
Last edited by NewToBowling; 10-13-2016 at 04:06 PM.
The measuring of other bowlers shots not something EARL should put much weight in.
EARL knows a lot more about EARL's shots because it is choosing parameters such as speed, rev rate, tilt, axis rotation.
For other bowlers those have to be measured. Subject to the inaccuracy of the measuring equipment.
Deciding what to try can be very accurate,
Throwing the ball to match what is being tried can be accurate, but it's a machine that has some error in it's tolerances.
Measuring what other bowlers have done is the lowest level of accuracy of the 3.
Weather prediction models can be made very accurate, if you know to conditions of the environment at any given moment.
The problem is in the measuring of the environment, a small error in measurement can make the model significantly wrong over time.
You've heard the phrase "garbage in, garbage out".
Garbage in this case is the difference between your estimate of a balls status (location, direction, speed, etc) and the actual status to infinite accuracy (i.e. reality)
Modeling future events from a current measurement is "garbage in, garbage out" on steroids, because the output from one cycle of the model is used as input to the next cycle.
Truth be told the biggest variance is the human element. We'd like to think that we can have the same laydown, speed, revs, mark every shot. But there are going to be differences no matter how minor on each and every shot. Take that out of the equation (which EARL should be able to do) and majority of the issue is resolved. Now if you had a pro bowler knowledgeable in lane/oil transition inputting data into EARL before each shot it would be a beast. Basically input data such as Target Mark=15 Board at Arrow, Speed=Unchanged, etc...next shot feet 2 boards left, target 2 boards left, etc
Posting Permissions
Bookmarks