There should be some constraints set on ladder templates rather than just pursuing variety. If you're going to rank people based on results across all templates (without having any weighting policy, either), then they should be fairly consistent. It's like having a ladder where you rank people based on their proficiency in card games in general- but someone's poker skill doesn't necessarily reflect their blackjack skill, and the two should be measured separately.
IMHO, MA, LD, Commanders, and a few other settings should be kept off the RT ladder since the skills that help you do well with those settings aren't the same skills that would help you conquer a template like Strat 1v1.
I disagree. I think the theme of the RT ladder is finding the most flexible player who can preform well at a wide variety of situations while under time pressure. There are other ladders designed for specialists, the RT is designed for generalists.
As stated the membership issue isn't much of a disadvantage and can easily be overcome by friending members, the issue here is for what percentage of the ladder population is autopilot a pay to win option, and does it matter if it only affects 1-10% at the bottom?
Lets suppose two real life friends have just started Warlight. Person A is bad, but better than person B. B buys the membership and proceeds to beat A using autopilot much to his amusement. Does A have a valid complaint that it is unfair? Once this has happened one time, how can A ever know when he plays B if he's playing the bot or the player?
In person A's position I think a fair proportion would simply quit.