Now that I have over 50 votes (53), let's see what is happening:
Vote option 1 (23 votes): 0-1 times equates to a +15 modifier over +0
Vote option 2 (10 votes): 2-3 times equates to a +10 modifier over +0
Vote option 3 (12 votes): 4-5 times equates to a +6 modifier over +0
Vote option 4 (5 votes): 6-7 times equates to a +3 modifier over +0
Vote option 5 (3 votes): 8-9 times equates to a +1 modifier over +0
The expected value (using the mean for each voting block) is 2.8 times (within option 2) so a +10 modifier (with rounding, this is discrete after all).
Now, originally I was only comparing "proficiency bonus" against no proficiency bonus. As many people have pointed out, my word choice might have mislead some of the voting choices with the poll question, but hopefully the explanation I gave in the OP went into enough detail to remedy this. With the added caveat about not considering expertise, which would otherwise double the proficiency bonus.
Either way, if you understood my intent in only considering proficiency bonus (all other things being 0 or equal between the two contestants), then the +6 currently used in 5E doesn't really offer enough of a difference between maximum proficiency bonus and no proficiency. A difference that more represents the expected results of such a contest would be +10.
I know bounded accuracy was meant to stop the treadmill effect, and it has nicely, but personally I always felt there was not enough of a difference between the two ends of the spectrum. I don't want a full 20-points like in prior editions, but 6 seems too little. We've been playing with a house-rule that maxes out proficiency bonus at +8 for a while now, and considered raising it again.
All that being said, if you think that ability modifier must be included due to the intent of the designers and voted with that in mind, then the current +11 max (+5 for ability, +6 for proficiency) is close to the poll results and works pretty well.
Thanks to all for voting and adding your comments. It has given me much to think on.