¹⁶. We not only introduced the selective reject mechanism, we removed the normal reject mechanism. We had long and difficult discussions about whether we should retain the normal reject mechanism to reduce the chance of destabilizing existing protocol software base (we anticipated most developers would start from a LAPD implementation). In the end, we felt that if the normal reject mechanism was retained, there may be little incentive for the developers to use add the selective reject mechanism, then our intent of a more efficient error recovery mechanism would be foiled.