How are Succeess/fail percentages calculated

[mattp169]

I am hoping Dawn will weigh on this thread, or possibly TYTONY.

ALot of people are complaining about how many fails they get on something at 90% etc.

Now when the game says 90%

Is that 90% for that attempt ONLY
or 90% for all combined attempts for that by the player ever or over a certain period of time
or 90% of all attempts of that server wide for the history of the server or over a certain period of time

Let me try and explain how that would all be different

If its 90% for just that attempt that means basically the software generates a number between 1 -100 and 10 of if its randomly generates a number between 1 and 90 you succeed otherwise you fail. If you do fail the process repeats the exact same way

If its 90% for all combined attempts for the individual player - if you fail the number generated to fail can no longer be generated on the next attempt, increasing your chances on the next attempt after a fail

If the 90% server wide then it would require generating a random number between sat 1 and 1 billion and any number below 900 million generated is a success, and if a fail occurs that failing number is removed from the pool for the next player who makes the attempt

If it is over a certain period of time in the latter 2 situations then the removed numbers will be put back into the pool after the time period expires

I believe the chances for the entire game are situation 1 - where the percentage applies to that attempt and that attempt only.

If that is the case that is what is driving everyone crazy.

Because when you see you have a 90% chance to succeed and fail 5 times in a row - you think I failed 5 out of 5 times how is that possibly 90%? And you are correct. This method of success failure determination can actually produce a 0% success rate at ant percentage below 100%. Because if its generating a random number between 1 and 100 to determine success, even at 99% success rate, the random number could be 100 every time you try thus leading to 0% success rate.


However using a method similar to the other 2 situations I described above would guarantee those success/fail rates to be correct over the long run

Now if the success/fail rate is determined by the first situation, some may argue that this would be dishonest or deceptive. I would argue its not. But I do think we should at least be told what type of method is used to determine the rates. This will eliminate a lot of confusion, anger and resentment.


SO please Dawn or TyTony, could you make some comment on this for us all??

[Klepto]

My guess is they just did the simple thing and did a 1-100 random number generation, then used the threshold of percentage to determine failures. That's what I would do as a developer. (e.g. percentage success is 90%, so any number above 90 is a failure, being 91-100).

Having said that, there are other ways to be more creative while still technically claiming a specific percentage, but that requires a lot more work.

[nolansco_5578848]

This is only a guess.

There is little doubt that tynon, and evony, 'tweek' the game parameters as the servers come out. In early evony for example attacking npcs gave regular 'finds' for heroes, but then became rearer than ducks teeth.. Tynon I think is doing the same thing, however I an not sure they update the published parameters, so the chance of say blue medals producing an upgrade is higher in server 1 than in server 99, although the published percentage success rate is not changed. As I say, i could be wrong, but it does seem to be the case with me. I know a 1 person smple is not mathematically significant, but thats what I think is happening

[mattp169]

Actually what you are saying points towards it being a server wide percentage over a specific period of time.

Because on a newer server you would have less people doing the upgrade then and older server with more people actively doing upgrades. SO you could be seeing 5 out of 50 fails which is 90% on a newer server while on an older server we have had 5000 fails out of 50k attempts. SO those 45k successes make people complain alot less

[Revan]

Also, they are likely using a pseudo-random number generator, so success/fail rates are not going to be evenly spread out throughout the day.