I am posting from the experience of making about 25000 skill checks, with supposed success probabilities ranging from 40% through to 25%.
In laymans terms this means I would expect between 4 in 10 and 1 in 4 successes each time.
The core of the problem and where the model is WRONG/UNFAIR is that the designers/statisticians have created a system that deliberately skews to lots of positive results in a row combined with even more negative results in a row.
So while the average probability ought to be correct, the distribution is deliberately untypical in order to 'excite' the player with lots of ridiculously improbably sequences (I have over 8 examples of more than a million to one against results, and hundreds of sequences which are over a thousand to one against)
Not only does this create in my case chains of up to 25 fails in a row, because it is a fake STRUCTURED probability system, at lower success targets (eg converting a level 8 enchant into a level 9 enchant) the strings of failures are so large that the average is unsustainable, and in this case drops from 20% to just 12%.
Again speaking plainly this means that the skill checks are not random in any meaningful case, and the proportion of players experiencing dramatic losses and gains is highly 'unfair' in the same way that playing the lottery is unfair.
Moreover it is very easy to blow over 100,000 AD on wards to protect against material loss for a result which is worth only 50,000, and which cost 40,000 to start with, leading to a massive loss. Any player desperate for an item (as with lockboxes, enchants, gifts or whatever...) will basically undergo a sado-masochistic experience because of the HAMSTER decisions made by the designers.
And the maths boffins really need to look at the calibrations they have made, because they are likely to cheese off and impoverish people who depend on a fair system.
The probability percentages stated are actually more wrong the lower the 'average; success ratio is supposed to be, and so they have not properly checked these ratios when they elected for a fake probability/atypical distribution of results.
My rant over, as much because I don't expect any designers to really care, but it is disappointing if they don't confirm or explain the wild style of model they decided to adopt, and think of changing it because of the way it pisses people off.
0