055 DDG Large Destroyer Thread

[vesicles]

Dunno. I'd assume they compile it from a variety of sources, from both test launches and combat launches. What I do know is that missile Pks are always given for one missile, not one salvo of missiles.

Please see below (quoting my above post)

How was Pk calculated from actual launches? It makes absolutely no sense that someone just randomly chooses a missile and launches it. then they magically come up with a number of probability/accuracy. Launching one single missile means you either hit it (probability of 1) or miss it (probability of 0). how do you derive any other probability values? They must launch numerous missiles against numerous targets. With a large enough data set (satisfying the power analysis), they then calculate mean and probability. at this stage, mean is effectively probability. they might say such probability is for estimating a single missile launch, that number is derived from large data set and should stand for large number of launches without further manipulation.

Now, will your next missile actually hit its target? No one knows, just like no one knows if a professional basketball player will his his next basket. However, if he shoots enough shots, his overall success % will be close to his probability calculated from his previous performance. In that sense, launching numerous missiles will actually get you close to your calculated probability of hitting a single target. Thus, it is actually beneficial for missile launching systems to fire against more targets if you want it to perform to its potential.

 

[vesicles]

again you are falling for what is call a gamblers fallacy. Each individual coin toss is 50/50 chance and prior coin toss has NO bearing on the next result. Even if you throw 5 tails in a row does not mean your next attempt will likely be heads.

It's actually not 50/50 chance. you can try it at your desk now. It's likely you get heads 5 times in a row. Like Ironman said, each individual attempt is independent and has nothing to do with any other attempts and is actually independent of the probability. Probability comes from huge data sets and many many attempts. In order to achieve 50/50 of a coin toss, you might need to toss your coin 200 times (hypothetically), where your initial 50 attempts might have 40 heads and 10 tails, your second 50 attempts might give you 40 tails and 10 heads, your third 50 tosses might give you 45 heads and 5 tails and your final set will give you 5 heads and 45 tails. As you can see, the data spread is very uneven. You only get 50/50 when you pool all the data together.

This is also why casinos always want to keep you in there as long as they can. The house is always favored, probability wise. However, to achieve that probability, the house will need to play many many attempts. that's why they want you to stay there as long as they can. that's also why you can win big at casinos even when the probability is stacked up against you. Just don't play for long. the more you play, the more likely the outcome will be closer to the probability, which means you will lose...

 

[kwaigonegin]

It's actually not 50/50 chance. you can try it at your desk now. It's likely you get heads 5 times in a row. Like Ironman said, each individual attempt is independent and has nothing to do with any other attempts and is actually independent of the probability. Probability comes from huge data sets and many many attempts. In order to achieve 50/50 of a coin toss, you might need to toss your coin 200 times (hypothetically), where your initial 50 attempts might have 40 heads and 10 tails, your second 50 attempts might give you 40 tails and 10 heads, your third 50 tosses might give you 45 heads and 5 tails and your final set will give you 5 heads and 45 tails. As you can see, the data spread is very uneven. You only get 50/50 when you pool all the data together.

This is also why casinos always want to keep you in there as long as they can. The house is always favored, probability wise. However, to achieve that probability, the house will need to play many many attempts. that's why they want you to stay there as long as they can. that's also why you can win big at casinos even when the probability is stacked up against you. Just don't play for long. the more you play, the more likely the outcome will be closer to the probability, which means you will lose...

hmm i thought that's what I was saying LOL.. individual chances on a singular event has no bearing on the cumulative total probability.

 

[Iron Man]

hmm i thought that's what I was saying LOL.. individual chances on a singular event has no bearing on the cumulative total probability.

Well if that's what you're saying, then you are still not understanding it. Individual chance on a singular event DOES have bearing on the cumulative statistical chance (just not on other individual chances). If the coin toss/missile Pk is .50 to get heads/hit or tails/miss, then the cumulative probability of 5 tails/misses out of 5 attempts is only 3.125% chance. If the coin is weighted to favor heads or the missile Pk is higher, say 0.7, then the cumulative probability of 5 tails/misses out of 5 attempts drops to only 0.243% chance.

 

[vesicles]

hmm i thought that's what I was saying LOL.. individual chances on a singular event has no bearing on the cumulative total probability.

thats not what I was saying. I said each individual attempt is independent of each other and is independent of the probability. However, that does not mean the probability is independent of individual events. Cumulation of individual events gives rise to the final probability.

 

[kwaigonegin]

thats not what I was saying. I said each individual attempt is independent of each other and is independent of the probability. However, that does not mean the probability is independent of individual events. Cumulation of individual events gives rise to the final probability.

I agree.. but as it pertains to missile launches is there truly enough 'real' launches to come to such a conclusion? What is a fair figure to arrive with fair amount of mathematical certainty to derive a statistical agreement of a particular probability figure?

 

[vesicles]

As I have been saying, no need to further manipulate whatever probability values people provide. It is a value already derived from large data sets and effectively indicates the mean. The more attempts you have, the closer your final cumulative outcome will be to the probability value.

It is also useless to predict how the next single missile launch will perform. It is not what we are arguing. All this mess started because someone claimed that more missile launches will hurt the success % of the missile launch system. So the question at hand is how probability changes as a function of large number of attempts. We know that as data set gets large, probability effectively becomes the mean. And we know that probability was most likely calculated from accumulating data from numerous attempts. Thus, whatever probability value we get will be % success the missile launch system will have when facing saturating attacks (assuming the system has enough missiles).

 

[vesicles]

I agree.. but as it pertains to missile launches is there truly enough 'real' launches to come to such a conclusion? What is a fair figure to arrive with fair amount of mathematical certainty to derive a statistical agreement of a particular probability figure?

that is for the engineers to decide. I believe a power analysis is designed to do exactly that, figuring out how many test launches would be enough to derive accurate enough conclusions.

 

[kwaigonegin]

that is for the engineers to decide. I believe a power analysis is designed to do exactly that, figuring out how many test launches would be enough to derive accurate enough conclusions.

My guess is not enough because unlike flipping a coin launching a missile even a dummy one is very $$$$!

It's based on the 'human/salesman' interpretation and pitch of the statistical average and not a TRUE mathematical average.

As you all have mentioned, true probability can only be derived from HUGE data sets and events.

If I test launch 10 missiles and I hit 8 targets I can legally claim and put my accuracy at 80% HOWEVER mathematically speaking it would be erroneous to factually state that if it is tried it 1 million times, I will hit it 800K times.

 

[vesicles]

My guess is not enough because unlike flipping a coin launching a missile even a dummy one is very $$$$!

It's based on the 'human/salesman' interpretation and pitch of the statistical average and not a TRUE mathematical average.

As you all have mentioned, true probability can only be derived from HUGE data sets and events.

If I test launch 10 missiles and I hit 8 targets I can legally claim and put my accuracy at 80% HOWEVER mathematically speaking it would be erroneous to factually state that if it is tried it 1 million times, I will hit it 800K times.

That's what the power analysis is for. It is a well established mathematical method for estimating size of data set. Engineers will be able to use available parameters to calculate the minimal number of launches needed to make conclusions.

More data is not always a good thing. For instance, if you want to compare two types of missiles, too many data points will artificially decrease standard errors, which might give you the impression of a data set tighter than what it actually is. That may mislead you into concluding that missile system A is better than missile system B when in actuality there would be no difference between the two.