Player Features and Ideas Discussion

CCP constantly speaks of immersion - but there is a jarring disconnect every time an invention job is completed.

There is eye candy ranging from: "This job was so easy you feel you could do it again in your sleep" to "You feel a bit out of your league, succeeding at this job requires a fair bit of luck."

It is difficult to comprehend how one can get both messages while inventing the *same* BPC. If it was so easy I could do it in my sleep, why do I need luck to succeed on any other BPC of the same type?

I propose that CCP adds a mechanism to account for experience.

This mechanism should:

• Account for experience gained from successes and failures
• Track experience within each specific BPC
• Track experience across all invention attempts

For any given attempt at invention, the BPC-specific and BPC-agnostic experience figures would serve as additional modifiers to current probability calculation.

What Gizznitt said. Right now spreadsheets for invention are fairly complicated due to the i.i.d. probabilities associated with the invention process. Adding in some sort of probability process that actually has a memory would make such cost-benefit calculations more complicated and talk about banging your head against the wall then...

You'd have to deal with conditional probabilities and you'd have to learn how those conditional probabilities work through trial and error and that would likely mean you'd have to do a number of rounds of invention, gather the data and then try to infer what the conditional probabilities are.

Consider this fairly simple example. You are going to draw two cards from a standard deck of cards (52) without replacement. What is the probability that you draw a queen on your first draw and jack on the second?

The queen part is easy, that is 4/52 or 1/13. But the second one is less easy. Now you only have 51 cards, and the probability that you will draw a jack, given you drew a queen is 4/51. The probability that you get a jack and a queen is now, 1/13*4/51 = 4/663.

But that isn't quite how the process you are describing will work. Suppose the problem is amended as follows:

You are drawing cards from a standard deck. If you draw a queen you draw again, but without replacing the queen. But if you don't draw a queen you replace the card. Now, what is the chance you'll draw a queen and then a jack? Now you'll have to do two probability calculations. The first is as above, 4/663. The other is 12/13*1/13 or 12/169 (i.e. the probability you don't draw a queen, 12/13. And the probability of drawing a jack from a full deck, 1/13)--assuming I did the math right .

And you'll have to keep going and keep a record of this so that your cost benefit calculations contain the appropriate probability. And you can't use the probabilities above, BTW. That is for a deck of cards, CCP would come up with the new probabilities and how each success will change your chances of future successes. Too estimate those you'll need an actual history of successes and failures, then you'll have to take that data and come up with a statistical model to estimate the conditional probabilities. Also, my guess is it wont be a simple process like with the card example above since eventually if you keep drawing cards you'll get to a point where there are no more queens or jacks and the probabilities will be trivial.

Basically, it is much easier to work with a memoryless process.

Calculating Invention costs is simple when you understand the probability function (currently its a Geometric distribution).
Then determine the expected number of trials to ascertain a success (currently = 1/p).
Then determine the Variance (currently = (1-p)/p**2 = Sigma squared).

Assume success rate is normally distributed with the above mean and variance.
95% of the time it will take less than Mean + 2 Sigma attempts to create a BPC.
99.7% of the time it will take less than Mean + 3 Sigma attempts to create a BPC.

Multiply by cost of a single BPC run...

Your suggestion makes calculating the expected number of trials and variance non-trivial, as your "experience level" will continually improve your success rate as you "learn how to invent the item".

You could still use the above/current formula as a excellent estimate to price your BPC's, you'll just end up with more profit in the end.

Now, lets actually review your suggestion:

Increase the probability of success on an invention job based on your character's historical attempts inventing said BPC:

Where is the middle ground? Does the old character, with 10k+ attempts/failures/successes, get a big advantage over the new guy just starting out? I personally think the current setup, where skills modify the results somewhat, allow older players to have a small skill-based advantage, while still maintaining an equal playing field because the new guy can "catch up" by devoting time to skilling.

Why is this dev work needed? You claimed it was because you feel a lack of immersion from the messages that pop up. It would be much less time & effort to simply remove the messages, than create a counter for each BPC, on each character, that tracks your historical invention rates. It would be much less effort than creating a new invention algorithm that is balanced so new players and old players can compete.

These are the points I was trying to make!

You are an employer that is employing different groups of scientists to invent your BPCs for you. Sometimes you get the MIT grads, sometimes you get University of Phoenix grads.