We agree that weight seems to be the primary factor in check success. I haven't done the testing, but some of my anecdotal (based on thousands of hours of gameplay, no research) thoughts:
* Point of contact, or angle of check is a big factor. A check right to the "chest" of a player will have a much higher success rate than if the player is moving away/sideways. Basically I think of it as a radius of contact for each player and the more in contact (overlap) they are at the time of impact, the greater the checking success. Picture a semi-circle in front of each player, and the more they overlap when they are in a check, the higher the success.
* In this way, agility is important as it gives you the ability to dodge quickly and cause your opponent to partially check your player. Or in my radius analogy, one player moves his radius of contact away, so only 50% of that semi-circle is in contact with the other players.
* CB has a higher success rate than plain C, or another way to put it I think the weight difference needed is smaller for higher success. Also, I'm pretty sure checking rating factors in here as Coffey (9 wgt, 2 chk) can't CB worth a damn, but Tikkanen (9 wgt, 5 chk) has tremendous CB success. I'd say Tikkanen can take out any 9 weight player with high frequency.
* Being near the boards is no mans land. Everyone dies there. Lindros can flatten Fleury.
Those are my two cents to add to this thread. I am intrigued about looking at the code though as I've gotten better at deciphering over the last few months with my '95 hack. To be continued...