|
|
| Author |
Thread Statistics | Show CCP posts - 0 post(s) |
chrisss0r
   |
Posted - 2009.01.14 21:21:00 -
[1]
i want a shuttle-based ship with a module that can shut off every tank module of targetted ships. (3-4) from 200km
it wouldn't be overpowered for the ship would not be able to tank, deal damage or do anything else.
|
chrisss0r
|
Posted - 2009.01.14 21:29:00 -
[2]
 Originally by: chrisss0r
i want a shuttle-based ship with a module that can shut off every tank module of targetted ships. (3-4) from 200km
it wouldn't be overpowered for the ship would not be able to tank, deal damage or do anything else.
oh yes and i want a 95% chance to be successfull. for nothing that is chance based can ever be overpowered!
|
chrisss0r
|
Posted - 2009.01.14 22:58:00 -
[3]
ECM is not overpowered since falcon alts equal cash for ccp.
no way they gonna nerf them anytime soon |
chrisss0r
|
Posted - 2009.01.15 00:59:00 -
[4]
Edited by: chrisss0r on 15/01/2009 01:01:04
Edited by: chrisss0r on 15/01/2009 01:00:29
What people always miss out and what really makes the computing of permajamchances difficult (and no i don't mean difficult in like people cannot compute chances....) is the fact that you can try a single jammer and if you don't succeed you can add another.
This has 2 major effects:
1. If the falcon misses a cycle on you you don' get a full cycle of not beeing jamed. You earn the second of not beeing jammed till more jammers are applied which is why there are many more "permajams" experienced than on paper.
2. To compute the jamming chances for every single jammer u add you need the bayesian probability calculus.
it's a bit complicated to explain it to such fools as you are but i'll try:
The moment you put a jammer on a ship and get a success or a fail u have gathered information. This results in not adding a second jammer in the moment of success or adding a second jammer in the moment of fail.
The fact that you don't add your space jammers in a moment of success leads to the higher propabilities of permajamming
This game has as many stages as a falcon has jammers free and together with (1) it is why so many more permajams occur than the (1-jamchance)^number of jammers fumula returns
The formula is still true but only after you have all your jammers applied.
So to speak the chance of jamming someone if highly biased into the beginning of a fight, and most of eves fights don't last long enough
So please shut the **** up if you have no clue
http://en.wikipedia.org/wiki/Bayesian_probability
|
chrisss0r
|
Posted - 2009.01.15 13:40:00 -
[5]
Originally by: Pac SubCom
Originally by: Camilo Cienfuegos
Until then, it's a moot point.
The "Bayesian method" doesn't influence the final outcome. It's still (1-p)^n, no matter which timeframe your calculations are based on. The real moot point is the pretension that wasting half of your jammers is what CCP created their balance around.
For instance, if the jamming probability of a single jammer is 30%, the Bayesian method does not convert this into 100% on one ship + 30% on a second with two jammers, and this is what I suspect he wants to say. If I win the lottery, this doesn't mean the probability to win is 100%.
This is indeed true. As i stated above the formula is still true once u won't add more jammers.
CCP calculated the chances of beeing jammed or not jammed around the simple formula and it would be the right formula to us if u could only decide in the beginning of the fight how many jammers to apply to a target and could only apply them all at once.
The fact that u can add the jammer one after another converts jamming from a fix chance like it's calculated by all the people here into a staged one and on each stage u can use the information gathered (yes/no) for your decision to deploy another jammer or not.
The result is, that you always have the "maximum" number of jammers free (depending on the outcome of the jammers you have already applied) to break a lock when your deployed jammers fail. Resulting in ALOT more "permajams" than the usually used formula would suggest.
On more than one target the outcome is even more drastic since the falcon can supply the optimal number of jammers on each target instead of, and only in that case the simple formula would be correct. Deciding in the beginning of the fight who to apply how many jammers to and not beeing able to change that decision.
I know the difference is not easy to get but it's a large impact on the question whether or nor someone will be able to lock anything the first maybe 120 seconds after a falcon arrives.
Numbers will not be provided. Not because i talk some blubberish that does not change anything but because calculating this pile of crap is a huge pile of work i don't wanna invest. We are talking about dozens of working hours.. |
chrisss0r
|
Posted - 2009.01.15 14:53:00 -
[6]
Edited by: chrisss0r on 15/01/2009 14:55:06
Edited by: chrisss0r on 15/01/2009 14:53:12
Originally by: Yarissia
Originally by: chrisss0r
The fact that u can add the jammer one after another converts jamming from a fix chance like it's calculated by all the people here into a staged one and on each stage u can use the information gathered (yes/no) for your decision to deploy another jammer or not.
The result is, that you always have the "maximum" number of jammers free (depending on the outcome of the jammers you have already applied) to break a lock when your deployed jammers fail. Resulting in ALOT more "permajams" than the usually used formula would suggest.
Only against multiple targets. If you only jam one target it doesnt matter for the chance of being jammed if you turn on one jammer after the other or all at the same time. This is because the chances of jamming are not correlated to each other. In other words your off-racials jammers dont mind if your racial got a jam or not they will still jam with the same chance.
The only thing you will get if you jam-look if jam worked-turn on next jammer is more cap and obviously a couple of free jammers.
Against one target this doesnt really matter.
This is not correct. Against a single target it is an advantage to have jammers free in the event of having the applied jammers to target miss a cycle so u can jam him again before he can get a lock. By adding one jammer after another you can determine the exact number of jammers needed to jam a target and thus keep the maximum number of jammers free.
As i stated before this biases the likelehood of jamming a target into the period the first cycles of jams occur.
Murina:
My argument is not simply that more jammers are better. You miss the point completely and should just shut up instead of showing off your lack of understanding.
If the falcon has 6 jammers, faces 6 opponents and puts 1 jammer on each of them the bayesian calculus is not needed at all, so the usual formula applies. one more reason why falcons kill small gangfights. The less targets the bigger shift of permajamming propability into the period till all jammers are applied
|
chrisss0r
|
Posted - 2009.01.15 15:02:00 -
[7]
Edited by: chrisss0r on 15/01/2009 15:03:01
Originally by: Murina
1. against a single target having jammers free makes no difference and no sense at all as you will either get a jam or not and holding a jammer back just in case you do not is pointless.
Well now i really know why you think the falcon is balanced. You just suck at getting everything out of ecm. And please stop pretending to understand what i'm saying everyone with better knowledge in statistic calculus just laughs at your counter arguments |
chrisss0r
|
Posted - 2009.01.15 17:02:00 -
[8]
Originally by: Murina
It makes no sense against a single target not to just put all your jams in it in he first place as its not like theirs any other ships around to save any for.
In fact if you click one and wait a few secs to see if it hits then another then another it could give the target ship a chance to lock one of your ships thats within range and fire or assign drones ect ect before you get through your entire rack, especially if its the last jam you assign that gets him.
While if you just sling them all on him str8 away the odds are exactly the same for a jam but if one does activate he does not have a chance to lock and shoot ect ect one of your gangs ships.
How hard is that to understand?.
This is indeed true but only viable as long as you can be sure the fight won't last longer than 20 seconds. For every next jam cycles you will have wasted free jammers u would have had in many cases thus increasing the probability of your target to get a lock.
This post again shows that you did not understand what i'm talking about.
Bayesian probaility calculus is not about changing chances. The jam chances for The single jammer are still the very same but it's about gathering information.
instead of wasting all your jammers on the single target and allowing him a 20 seconds logspan in case they all should fail you just apply as many as needed. Jamming is not a single point event when u deploy your jammers 1 by 1. It get's staged and every new stage allows you to chose if you should apply another jammer and this is resulting in alot more permajams until all jammers are applied. Read the wikipedia article and freaking try to understand it.
Statistic calculus does not use the bayesian formula just for the simple joy it brings while havin the same results as your simple calculus.
|
chrisss0r
|
Posted - 2009.01.16 10:03:00 -
[9]
Originally by: Pac SubCom
Edited by: Pac SubCom on 16/01/2009 08:12:56
His point is that that "a high efficiency jamming method results in more permajams than you think." With "than you think" being the core (non-)argument. It has no further descriptive power as regards the discussion.
Nonetheless it can be a useful tactic. But it doesn't change the jamming probabilities, therefore it doesn't change permajamming probabilities.
This is not Correct. People mix up absolute probabilities and conditioned probabilities. If it did not change anything why would it be a theory used? I dont provide numbers cause it's a ****ing ****load of work and i don't feel urged to invest so much time so u guys have a number.
I'm not saying "bayesian calculation proves falcon is overpowered" how could it?
i have no idea how big the influence is (hard to calculate), i'm just stating you guys compute permajamming chances the wrong way.
i personally feal the falcon is overpowered but that's nothing pure to numbers. let's say you have a situation where u compute a permajamming chance of 40% while the real permajamming chance (in the beginning of the fight and while more jammers free to deploy) is 43%. would that prove falcons are overpowered? no. Would it prove falcons were overpowered if it was 69% ? No. That's why i don't put the effort in calculating it. it would neither prove if falcons are overpowered nor if they are not. My statement is all the peeps who want to bring math arguments for or against the falcon fail cause they calculate the jamchances wrong.
I will once again try to explain it on a simple example.
You have a falcon with 6 jammers and 2 target ships. jaming chance is 50%
relock time is set to the amount of player reaction time on the falcon to make calculation easy, which means if a ship trys to lock and the falcon has a jammer free i can throw it into before the ship got a lock. this is restrictive but if you fu.ck around you can come op with a more realistic model yourself.
So: Classical Calculation
Falcon deploys 3 jammers to each target at once
Chance target 1 is jammed: 1-(o.5^3)=0.875
Chance target 2 is jammed is equally= 0.875
BAyesian calculation:
Falcon put's one jammer on each target and waits the result:
Well and now it's where things get complicated:
Given target 1 is jammed by the first jammer, the Falcon has 5 jammers free and the jamming chance of target 2 becomes (1-0.5^5)=0.96875. (the falcon flame brigade: don't use this number, you will use it wrong. It does not say falcons have 100% jamchance on ship 1 and 96.875% on ship 2...)
I'm pretty sure you may have noticed that while jamming the second ship with a higher probability you have only 50% on the first ship. Thats indeed true and is the reason why overall jamming chances don't change (since the jammers are uncorrelated) but it does not matter. stage one of the game (1 jammer applied to each ship) is played and done. Ship one beeing jammed is not a stochastic event anymore but a given condition in this stage of the game.
The whole thing is staged (if the first jammer fails on ship one and you apply another one which succeeds you have another 4 to deploy to ship 2) the whole thing can be looked at from both sides.
If the first jammers on both ships fail the game is reset with the falcon now having 4 jammers and starting again.
The fact that you can gather information by staging your jammers is why you have to use baysian calculus.
i kno this will result in missunderstandings and flames since people don't get the concept at all ( JAM CHANCES DON'T CHANGE; FAAAAIL) or just missinterpret it like sayying it's the proof that ecm is not stacking nerfed  (lolwhat?) or will demand NUUUUUUMBERS. Here i have to say FUC:K YOU again. i will not create a dynamic probability model just to find out how high the probabilty of a blinky thing in an online game is. Gimme the money that is common for dynamic statistic calcs and i'll reconsider
|
chrisss0r
|
Posted - 2009.01.16 10:06:00 -
[10]
sorry amira but he's right. You can't fool people who who a topic well into believing you also know it well.
it is kinda obvious that you and murina have not the slightest idea what the hell i'm talking about so please give in.
|
|
|
chrisss0r
|
Posted - 2009.01.16 11:57:00 -
[11]
Edited by: chrisss0r on 16/01/2009 11:59:06
Originally by: Pac SubCom
Edited by: Pac SubCom on 16/01/2009 10:45:49
Originally by: chrisss0r
FUC:K YOU
You get the same efficiency by just spreading n jammers over n targets,
Sorry this is bull****. You did not get it at all.
And this is not some "fancy numbers" i came up with but scientific reality.
"bayes yields around 2.5m google hit's. Alot of people who like to do complex calculations that come to the same effect as simple calcs. Just for the fun of it, you know
|
chrisss0r
|
Posted - 2009.01.16 12:29:00 -
[12]
Originally by: Pac SubCom
Edited by: Pac SubCom on 16/01/2009 12:21:24
Originally by: chrisss0r
Edited by: chrisss0r on 16/01/2009 11:59:06
Originally by: Pac SubCom
Edited by: Pac SubCom on 16/01/2009 10:45:49
Originally by: chrisss0r
FUC:K YOU
You get the same efficiency by just spreading n jammers over n targets,
Sorry this is bull****. You did not get it at all.
And this is not some "fancy numbers" i came up with but scientific reality.
"bayes yields around 2.5m google hit's. Alot of people who like to do complex calculations that come to the same effect as simple calcs. Just for the fun of it, you know
You describe is a method to avoid the waste of jammers. Good job, but it doesn't change the probabilities of the Falcon permajamming me a single bit because of the chances that he will not primary me and use his [yes/no] algorithm on me. Should he spread jammers equally, my gang will lose the same damage over time than if he made sure to permajam one or two targets.
Spreading jammers might be even more effective because a relocking period of many ships can reduce gang dps more than permajamming few.
If you would provide numbers to prove your point or some mathematical proof, I would reconsider. I suspect you see the whole thing narrowly, ie the permajamming of primaries as a psychological effort to drive them to the forums to whine, while I look at ECM as a whole (damage reduction of the whole enemy gang), and that is why we diverge. You are correct, just as I am.
because you calculate from a point of view where a falcon ahs no free jammers anyways. The bayesian calc comes into account when there are many more jammers than targets. if there is as many targets as jammers the simple formula accounts or, as you noticed correctly the fail/notfail method will provide a more probable jam on a few targets while the falcon does not try to jam the rest and gets toasted.
Now consider solo/small gangs and you'll see why it is important to factor the real jamming probabilities instead of the simple ones which are wrong in that case for as long the falcon has free jammers to apply
|
chrisss0r
|
Posted - 2009.01.16 18:33:00 -
[13]
Originally by: Lilith Velkor
Edited by: Lilith Velkor on 16/01/2009 18:32:13
Originally by: chrisss0r
because you calculate from a point of view where a falcon ahs no free jammers anyways. The bayesian calc comes into account when there are many more jammers than targets. if there is as many targets as jammers the simple formula accounts or, as you noticed correctly the fail/notfail method will provide a more probable jam on a few targets while the falcon does not try to jam the rest and gets toasted.
This ffs.
His point isnt if the falcon can 'permajam' (I'll just use this term, bare with me) or not, it is how many other ships the falcon might take out after 'permajamming' the primary target.
Now you dont need any fancy calculations, just a programming language of your choice (Excel prolly can do with macros), a few dozen lines of code and you get a pretty good approximation (in about 1 hours work).
For those too lazy to do it themselves, in short: falcon with all caldari racials vs only caldari BSs will take one out of the fight almost entirely, and severely hamper a second one. That is a scenario that gives all good cards in the game to the falcon pilot though... while I still think the actual result is balanced.
asked a friend to code it for me. see the other thread
|
|
|
| |
|