I take no offense at people preferring Plum's suggestion. The problem with it is that the tech window already is timing based, and altering the mechanics behind how shield inputs are read during hitstun in the manner required for that fix could very likely interfere with the 15 frame input window for techs.
OK, I am posting my planned initial alterations for throws. The first phase of gathering all the throw formula and hitstun is nearly complete. I will then optimize them based on both how the animation looks and bring all of the throws up to an acceptable level. I will then proceed to determine high% combo and kill potential of throws. That will allow me to create the most idea throw game. I aim to nerf nothing, assuming hitstun at .46, unless a serious CG is present (ICs are an exception to this principle as we have already altered their desynch). I am approaching this from an objective, data based perspective, to remove personal bias from the process. The changes will not be omgwtf alterations, if a throw has been nerfed I will attempt to determine the exact minimum required fix so as to have no overly nerfed options (see CF d-throw ;p )
My exact proposed methods for altering individual throws are subject to change if I see a more ideal approach (there are several ways I could improve Link's throws for example, and they are substantially more tricky than one would imagine). This sort of waffling will be minimal, I can assure you. When I am done, we will have an extremely solid and well understood system which will bring us far closer to the goal of a "gold set".
The thread with the data and proposed changes is linked in the welcome wagon thread, but it is buried under closed threads, active discussions, etc. I update it EVERY DAY and welcome dialogue with players extremely familiar with their characters options in B+ (a lot of technical knowelege and experience is required to really understand what I'm doing. Its very hard for me to overstate the complexity of what I am attempting). Nonetheless, I am making everything public so that the whole process is open, transparent, and subject to peer review (the cornerstone of successful data based endeavors). Check it out:
http://www.smashboards.com/showthread.php?t=257152 it is completely worth your time.