QUOTE (Sengir @ May 19 2013, 04:39 PM)

An initial dice pool of 10 means total of 55 dice rolled, without limits that gives 18.3... expected successes.

That is why I'm glad they stayed with the old extended tests, it's easy for the GM to do those numbers mentally and come up with a realistic threshold. Of course limits can skew those numbers badly, if the limit is significantly lower than the DP.
But just based on attribute and skill, 10 is hardly an average number. It's pretty high end. It was average in SR4 because there were a lot of various modifiers you could apply that had nothing to do with attribute or skill, and those things have been stated to go over to Limits now.
It could be that a skill rating of 6 is the average now, so 3 attribute and 6 skill is average for a professional, but right now we really have no basis for that king of assumption. And as Entropian noted, a skill 3/attribute 3 (ie pretty average professional) guy is going to be rolling an average of 7 successes. So he can barely even succeed at easy tasks.
Note: Unless the mechanics change drastically, data searching will fall under the same category of problems. You'll need a dedicated hacker with a big dicepool to accomplish more than a 2 second google search, which is absurd.
Honestly I just don't like the mechanic because of how important it makes every extra die you add. Each extra die adds your entire dicepool to the total. Going from 3->4 gets you 4 dice, while going from 9->10 gets you 10 dice. But thresholds show no sign of being tuned to account for that kind of very fast increase. So you have silly things like the average dude with 6 dice struggling to get 6-7 successes, but adding 4 more dice jumps that up to as you note 18 successes, and jumping that up 4 more dice (to 14 total) gets you all the way up to ~32 successes. And given extreme is 30+, getting anything beyond that is useless except to make you ridiculously fast at accomplishing an extended test.
Really we need more information on expected ranges and such before we can really make any real judgement on this. But at the very least the Threshold numbers should be tweaked. As it is they are just a straight copy/paste job of SR4A, and that's not really reassuring. Should have something like first figure out what your average dicepool should be, then figure out what they can hit on average. This is your average threshold. Go up and down from this by 3 dice for every other threshold.
So if we decide that say 9 dice is the average. 9 Dice can be expected to score 15 hits. So 15 is your average threshold.
So now we look at 6 dice for the Easy threshold. 6 dice can be expected to score 7 hits. 7 is your easy threshold.
Now we look at 12 dice for Hard. 12 dice are expected to score 26 hits. This becomes your hard threshold.
Now we look at 15 dice for Very Hard. 15 dice get on average 40 hits. This becomes your very hard threshold.
Now we look at 18 dice for Extreme. 18 dice get on average 57 hits. Since this is extreme, we'll round it off to an even 60, and call 60+ an Extreme threshold.
On the other hand, if 6 dice is your average, then 7 becomes average threshold, 15 becomes your Hard, and so on. The new easy threshold is a mere 2. Of course, you can see where these numbers get out of hand pretty quickly.
Personally I would like to see Limit factor in as a limit to how many times you roll, rather than how many successes you can get, on an extended test. So your average person might be 9 dice with a limit of 3 (average: 8 successes), while your extreme will be 18 dice with a limit of 6 (average of 31 successes), which gets us numbers closer to what we're used to, still makes limits useful/sane
Easy: 6 with 2 (average: 3.6, round it off to 4)
Average: 9 with 3 (average: 8 )
Hard: 12 with 4 (average: 14)
Very Hard: 15 with 5 (average: 21.6, round it off to 22)
Extreme: 18 with 6 (average: 31)
Alternate numbers without decrementing dicepools:
Easy: 4
Average: 9
Hard: 16
Very Hard: 25
Extreme: 36
Which also seems relatively reasonable, and can work out as well.
Alternatively, I wouldn't mind seeing a bigger overall change to how these work. Possibilities:
1) Simple change. You roll once for your Extended test. Divide the threshold by the number of hits, that tells you how many intervals it took (if this number is less than 1, you are awesome, and can take whatever fraction of the interval you managed). If the number of intervals is higher than your limit, you failed. This method works well with the last set of Threshold Values I listed above.
2) Bigger change. You roll your dice for each interval, but rather than tracking hits across each roll, instead the hits on your previous roll adds to your dicepool for the next roll, until you achieve more net hits than your threshold, or your number of rolls has exceeded the limit. For example an average extended test threshold might be 5. We have our average mechanic (6 dice) with a shop that gives him a Limit of 4. He goes to do an average test that has a Threshold of 5. First roll, he gets 2 successes, so his next roll he gets to use 8 dice. That roll he gets 3 successes, letting him roll 11 dice on his 3rd roll. On that roll, he gets 4 successes, letting him roll 15 dice for his 4th roll, where he gets the 5 successes he needs to succeed on the task.
This method keeps the massive amounts of die rolling in for Extended Tests, but deflates the DCs necessary a fair bit. Using the same metric as above (average = 9 dice, limit 3. Going up 1 stage is +3 dice + 1 limit, down 1 stage is -2 dice, -1 limit), the thresholds you'd want for this are:
Easy: 3
Average: 5
Hard: 9
Very Hard: 15
Extreme: 24+
This method has an advantage over the traditional extended test model in that it's more consistent with other rules. Basically it works almost as though you are doing a teamwork test with yourself every interval.
Unrelated, and can be applied to any possible option: I would love to see something that lets you cut down your interval one step (ie from weeks to days, days to hours, hours to minutes, etc) in exchange for bumping the difficulty up one step.
So say you have a Mechanic check that is normally easy, but has a 1 day interval. A typical mechanic will do it in one day, and not worry about it. But a really good mechanic with great equipment knows he can hit a Very Hard check with no trouble. So he drops the interval from Days to Hours, to Half Hours, to 10 minutes, in exchange for raising the DC from Easy, to Very Hard. This lets complete the task that would normally take a full day to somewhere between 20 and 60 minutes, depending on luck (it will take him 2-6 intervals most likely to achieve the very hard task).
This would be nice for a number of reasons, but at least one big thing would be in letting skilled hackers write their own programs in some reasonable timeframe while still making them very big projects for the average code monkey.
[Edited to remove annoying smileys.]