Funny how people have posted to this thread who claim they know something about mathematics or statistics.

Put simply, **what it means to say** "you have a 20% chance of succeeding each time you try this" is that, if you had a **statistically significant** sample set, you should find that somewhere close to 20% of the attempts resulted in successes. This correlation should increase the larger your sample set.

So, if your sample set is 1 attempt, you aren't going to derive ANY meaningful data. Same with 10, 100, 1000. Ok, so maybe at 1000 you will start to get SOME meaningful data.

But imagine if you had 100,000,000,000,000,000,000,000,000,000,000,000,00 0,000,000,000,000,000,000,000 attempts recorded down, and in that enormous sample set, only **6%** of the attempts **actually** succeeded.

Assuming you had this huge sample set and this observed result of 6%, how accurate is the original claim that "you have a 20% chance of succeeding each time you try this"? Well, given the astronomically huge sample set and the extremely skewed results, your conclusion would be "the original claim is totally bunk".

That is just one extreme example, where I demonstrated that, with an enormous sample set whose results do not even come remotely close to matching the original claim, the **descriptive statistics** of the results do not match the **inductive statistics** of the prediction.

All we're looking at with SWTOR is a much, *much* smaller sample set, where the data is still nonetheless skewed. That doesn't mean "you can't derive any conclusions from it at all because it's random" -- there are different **types** of randomness, different "qualities" of randomness. If the randomness source were **ideal**, then the more data points you have, the less likely you are to observe results which do not agree with the inductive principle (the 20% chance) -- **assuming** that (1) the inductive principle is CORRECT, and (2) assuming that the random number generator produces true random numbers.

So, if we HAD a significant sample size and could demonstrate that the sum of all our player-experience data is statistically significant, and we found that the results were skewed, then we could conclude that either (1) the inductive principle is INCORRECT, or (2) the random number generator DOES NOT produce true random numbers.

While it is true that one player's results would not be enough data, I think it'd be neat to have some kind of collaborative spreadsheet on Google Docs where a ton of people record their RE attempt results. This would allow us to gather a significant sample set.

Of course, "significant" is in the eye of the beholder. What's significant? How many is enough? It really depends on how many chance events are occurring within a practical timeframe that you're willing to measure.

For example, if you are measuring probabilistic quantum events that occur thousands or millions of times per second, a sample size of 1000 is not statistically significant. But, if you are measuring a baseball player's chance of hitting a home run when he steps up to the plate, 1000 could be very significant as a means of coming up with an inductive principle.

What Bioware has done is they've given us an inductive principle (20% chance of success) without providing us the "source code" (the mathematical theory or empirical evidence) supporting that principle. So, basically, without a statistically significant number of rolls of the dice from their random number generator in a spreadsheet, we can't be sure that what they claim is true, unless we observe the results ourselves (which will take a long time to get a goodly amount of data).

Right now, "20% chance of success" means nothing. It may as well be 0.1%. Any inductive probability claim is only as true as the data supporting it.