I should probably start a separate thread on new use, but Monday is a bad day for me to be adding things to my list of things to do. I'll deal with it here.
In the example of shafting the arrow, the success or failure of the psionic accuracy might not matter--success would mean new use, failure would mean it's not new use--but the potential that you might botch is still in view. That botch might be anything--brain burn with a visual component that prevents you from being able to focus your eyes on the target without succeeding at a difficult will power check, psionic shockwave spreading out from you injuring everyone around you including, at the epicenter, yourself; psionic inaccuracy, penalizing every shot you make for a while; neural breakdown, chipping away at your mind and possibly driving you to madness; the list might include a number of other things.
When you are practicing, even if you are practicing by using psionic accuracy to attempt to shaft an arrow, we're not going to roll for it because it is practice, and because success and failure do not really matter, and because this is a controlled practice situation--and as a result, you cannot botch in practice, you can only learn.
The "new use" improvement is not supposed to be the norm but the unusual. It is supposed to be the thing that kicks in from time to time particularly with the guy who never "practices" but attempts to get better by taking risks and using his skills in unexpected ways. Certainly the guy who is constantly using his skills is "practicing" by doing--that's built into the Mary Piper advancement system, in which you are using the skills necessary to the job on a daily basis and do not roll them, but if as a pilot it suddenly falls to you to avoid the whirlpool or the iceberg or outmaneuver the incoming pirates, that's when you roll, and that's when you can try something outrageous that might fail or botch and claim in the process that you just did something you never did before (instead of simply using your ordinary skills in an extraordinary situation, which is not a new use). This constant ordinary use or practice of your skills builds skill ability level slowly; the other is the rare eureka moment that happens to people a couple times in their learning, that they suddenly see something about what their skill can do that they never considered before, and tried it, and it worked.
I expect "New Use" to reflect that rare eureka moment, and in most (but not all) cases to be a response to a pressure situation, and in all cases to be a situation in which the potential for botch is real.
Now, if you want to do something in "practice" and make it a new use, the way we do that is we announce that we are not in "practice" but we are "trying something new", and on that one attempt we roll the dice to see whether it succeeds, fails, or botches. If it fails, you didn't do it, and you don't get the credit; you can wait a reasonable time and try again, but you can't just keep trying until it works, because "keep trying" is part of the check in many skills (starting a car, for example--if it fails, it didn't start when you tried most of what you knew as an operator, and you're going to have to use such skills as you have acquired as a mechanic to get another shot, or wait and hope that it was just flooded or something). If it succeeds, you did it. If it botches, you face the consequences. However, if the botch consequences are not real in potential, it cannot be a new use.
Your bunkmate's interest, to this point, is merely in hearing the stories. You can tell that he is not certain whether to grant them credibility or take them as interesting fiction.
It is the sixteenth day out from Emerald, the one hundred fifty second day of this trip, and you are at the helm in the middle of the day when an alarm rises. There is a small fast ship closing from the starboard stern, and it does not look like it has good intentions.
--M. J. Young