@ThatGuyThere Actually, the average number of successes for someone rolling 9 dice 4 times is around 15 because of exploding contributing more than you think (http://anydice.com/program/bd0a. Thanks to Misadventure for pointing me to the site).
And while things like 9-again may not contribute too much to a roll by the time all is said and done they contribute a great deal more than you might think. After all, someone with 6 points in combined skill and attribute will roll 54 dice (assuming a +3 equipment bonus) so at the end of the day an extra 10% here and an extra 5% there adds up to a lot of successes.
But the real point of this thread wasn't to quibble about how much ability A helps or how many extra successes ability B will generate. It's just to figure out what the theoretical pools (and any roll modifiers) should be.
At present I think I'm more or less in line with your estimates. I'm actually ranging from 4-8 dice with the 4 dice being the skill level of your average entry level professional. These are the guys on the construction crew that do the real simple jobs. You have them handle the framing and drywall when the job is easy but in the spots where those tasks get tricky you have someone who's been doing the job for a couple of years now come in and handle it. 6 dice are your true 'average professionals' who are fully trained. They've been doing the job and 95% of the time if you want the job done you call these guys in. 8 dice are the true experts. Not necessarily world-leaders in their field but they are they guys who get called in to do jobs that even your normal professional would consider to be difficult.
Then what I did was I figured out the average number of successes for 1 to 6 rolls as an extended action using a rote action for 7 dice, 9 dice, and 11 dice. I assume that the vast number of extended roll activities your average professional does are rote actions. Sure, the tailor needs to modify the pattern for the custom fit jacket they are making and everyone's body if a slightly different shape but they've done so many jacket alterations, plus they have the time to sit down and work out any tricky areas that might exist, that it is still a rote action for them.
Then I figured out how many rolls you would expect someone normally doing that job to take. If your 'average professional' would take 1-2 rolls I used the average number of successes for 7 dice because the average person doing that job is a bit less skilled. If it was 3-4 rolls I used the average successes for 9 dice and if it was 5-6 rolls I used the average number of successes for 11 dice.
This seemed to work really well because the guys with the 4 die pool could nearly always accomplish the '2 roll' task (99.96% of the time). It might take them twice as long to get it done (because they had to make four rolls) but they would nearly always be able to, much like you would expect in real life. Meanwhile the odds of that person pulling off a '3 roll' tasks is about 50/50 (meaning that they were around 75% likely to succeed if they took 3 times as long as the 'fully trained' professional. Meanwhile the 'fully trained' pro with their 9 die pool would nearly always accomplish a '4 roll' task (99.21% of the time) but dropped to around 50% odds of succeeded at a '5 roll' task.