Bot Drop Curious

[Lazy]

You have my comments backwards Lazy ... I used the example of waiting until 100's of people had died or been hurt before acknowledging a problem. I did not say put a new drug out to the public after 15 trials. I said take note of problems after a small sample and investigate Before putting something new out to the public.

i don't think a 10% success rate of getting a bot is comperable to a medicine. any medicine with a 10% success rate and a 90% mortality rate isn't going be be fielded so your comparison is, frankly, nonsense.

let's use a more realistic example: let's pretend you're asking people to pick a number from 1-10. you ask 15 people. no one picks #4. are you going to accuse the people you're asking of being broken?

[Deemure]

Wow, some people should take a classor two in statistics. Without getting into standard deviations, actual vs. predictive, and a few more complex aspects. It really is quite simple. If the drop rate as stated is 10%, then each raid, assuming randomness, has a 1/10 chance of bots with replacement. So in fifteen raids, the probabiliy of zero bots is 9/10 to the fifteenth power. With some rounding, this is roughly a 14% chance of no bots. There is a 85+% chance of at least one bot, and 10 to the 15 power chance of 15 bots. Math is math people. If the bot drop rate is actually 10% then the 'dry periods" mentioned including the 15 Kinnik mentiond, and the 24 I have personally experienced since the merge statistically are way out into the 5 or 6 standard deviation of probability. Not likely at all. In fact close to impossible if the drop rate is 10% and the drops are random. Therefore there are two and only two possibilities here. Either the drop rate is closer to 2% (no, I won't bore you with the statistics and how to verify statistical premises) or the drops are not random but are based on other factors as well (including an formula that tries unsuccessfully to mimic randomness). In either case some clarity would be appreciated. I know that 14% chance is only about 3 standard deviations, still probable but highly unlikely. The 30+ mentioned and the 24 are down into the 1% probability. Statistics do not play favorites, but it is increasingly likely that fc does.

[Pennypacker]

Wow, some people should take a classor two in statistics. Without getting into standard deviations, actual vs. predictive, and a few more complex aspects. It really is quite simple. If the drop rate as stated is 10%, then each raid, assuming randomness, has a 1/10 chance of bots with replacement. So in fifteen raids, the probabiliy of zero bots is 9/10 to the fifteenth power. With some rounding, this is roughly a 14% chance of no bots. There is a 85+% chance of at least one bot, and 10 to the 15 power chance of 15 bots. Math is math people. If the bot drop rate is actually 10% then the 'dry periods" mentioned including the 15 Kinnik mentiond, and the 24 I have personally experienced since the merge statistically are way out into the 5 or 6 standard deviation of probability. Not likely at all. In fact close to impossible if the drop rate is 10% and the drops are random. Therefore there are two and only two possibilities here. Either the drop rate is closer to 2% (no, I won't bore you with the statistics and how to verify statistical premises) or the drops are not random but are based on other factors as well (including an formula that tries unsuccessfully to mimic randomness). In either case some clarity would be appreciated. I know that 14% chance is only about 3 standard deviations, still probable but highly unlikely. The 30+ mentioned and the 24 are down into the 1% probability. Statistics do not play favorites, but it is increasingly likely that fc does.

Yes, each time you post something against Funcom on forum they add an extension to the droprate formula decreasing that forum names accounts chances of drops.

[Gatester]

I've had 7 supples drop in a row from 6 raids.

I've had a streak of 63 raids and 0 bots drop.

Probability is not guaranteed, it is merely expected. Just because something is unlikely to happen does not mean it will not happen.


I've also posted my 3 bags of ICE with 0 bot drops ratio several times...and just now realized I've been saying 71 instead of 63. Why have none of us noticed this

[Archaeopteryx]

Well if anyone is bored and got nothing better to do....i can give you the xl file im using to record my drops (raid loot).
It calculates everything just needs the data gathered

...Ask ppl to post here or in another post what the general drops each time they kill it and input it into the file.
At least like that we'll have some relevant %

[Lazy]

Wow, some people should take a classor two in statistics. Without getting into standard deviations, actual vs. predictive, and a few more complex aspects. It really is quite simple. If the drop rate as stated is 10%, then each raid, assuming randomness, has a 1/10 chance of bots with replacement. So in fifteen raids, the probabiliy of zero bots is 9/10 to the fifteenth power. With some rounding, this is roughly a 14% chance of no bots. There is a 85+% chance of at least one bot, and 10 to the 15 power chance of 15 bots. Math is math people. If the bot drop rate is actually 10% then the 'dry periods" mentioned including the 15 Kinnik mentiond, and the 24 I have personally experienced since the merge statistically are way out into the 5 or 6 standard deviation of probability. Not likely at all. In fact close to impossible if the drop rate is 10% and the drops are random. Therefore there are two and only two possibilities here. Either the drop rate is closer to 2% (no, I won't bore you with the statistics and how to verify statistical premises) or the drops are not random but are based on other factors as well (including an formula that tries unsuccessfully to mimic randomness). In either case some clarity would be appreciated. I know that 14% chance is only about 3 standard deviations, still probable but highly unlikely. The 30+ mentioned and the 24 are down into the 1% probability. Statistics do not play favorites, but it is increasingly likely that fc does.

done 5 raids since merge, got 4 bots since merge
ground raids only... why do people make these topics still?

so 15 raids from kinnick, + 24 of yours + 5 of his gives us 41 raids with 4 bots or about 10% drop rate. looks fine from a statistics standpoint. or did i miss something?


EDIT: also, the mistake you're making is the Gambler's Fallacy. it doesn't matter if you do 1 raid, 10 raids or 1000 raids. each raid has a 10% chance of dropping a bot regardless of what the drop rates for the previous raid were.

[Deemure]

so 15 raids from kinnick, + 24 of yours + 5 of his gives us 41 raids with 4 bots or about 10% drop rate. looks fine from a statistics standpoint. or did i miss something?


EDIT: also, the mistake you're making is the Gambler's Fallacy. it doesn't matter if you do 1 raid, 10 raids or 1000 raids. each raid has a 10% chance of dropping a bot regardless of what the drop rates for the previous raid were.

If you would like, I would be happy to refer you to several books on probability and statistics since you seem to be light on understanding the difference. While it is true each event has a 10% chance. The composite probability changes with each addition to the string of events. So your probability of of bots in two raids for instance is 1/10 for the first and 1/10 fo rthe second. There are only two outcomes for each raid bots or no bots with probability of 1/10 and 9/10 respectively. The combined probability for two raids is 1/10 X 1/10 for two successful (1 in 100), 9/10 X 9/10 for no bots (81%) and 18% for one and only one bot from the two raids (for simplicity assuming one bot drop).

Again, please refer to some authoritative probability and statistics reference. Cherrypicking across event lines is meaningless, unless you are somehow suggesting that the conditions are identical. A fact not supported. Also the large number of bots from a few raids (4 bots 4 raid is a one in 10,000 probability) which again supports the hypothesis that the drops are not random.

I have had issues with the randomness of items in anarchy for years and have made several posts questioning how close to randomness they actually achieve with their generator. This seems to be another indicator that that generator doesn't work in a true random nature. There are too many instances outside the normal statistical deviations. If that is the case, then the question of how and why is does not approach statistical randomness is a concern. if it isn't random, then what is the criteria for determination?

[Lazy]

If you would like, I would be happy to refer you to several books on probability and statistics since you seem to be light on understanding the difference. While it is true each event has a 10% chance. The composite probability changes with each addition to the string of events. So your probability of of bots in two raids for instance is 1/10 for the first and 1/10 fo rthe second. There are only two outcomes for each raid bots or no bots with probability of 1/10 and 9/10 respectively. The combined probability for two raids is 1/10 X 1/10 for two successful (1 in 100), 9/10 X 9/10 for no bots (81%) and 18% for one and only one bot from the two raids (for simplicity assuming one bot drop).

while this is true, this doesn't affect the probability of the next raid being 10%. just because you do 100 raids with no drops doesn't guarantee a bot on the 101st raid.

Again, please refer to some authoritative probability and statistics reference.

no u

Cherrypicking across event lines is meaningless, unless you are somehow suggesting that the conditions are identical.

both are city raids. how much more identical do you need them to be?

A fact not supported. Also the large number of bots from a few raids (4 bots 4 raid is a one in 10,000 probability) which again supports the hypothesis that the drops are not random.

sure they are. if you don't mind me quoting wikipedia, "Gambler's fallacy arises out of a belief in the law of small numbers, or the erroneous belief that small samples must be representative of the larger population. According to the fallacy, "streaks" must eventually even out in order to be representative.[7]"
there's nothing that says for every 9 raids with no bots you will get a bot on the 10th, or that for every bot you get you will get 9 raids of no bots directly after. but over the course of 10k raids you will have gotten about 10% drop rate. this is why i and everyone else has been saying that to be taken seriously you need a large sample size.

I have had issues with the randomness of items in anarchy for years and have made several posts questioning how close to randomness they actually achieve with their generator. This seems to be another indicator that that generator doesn't work in a true random nature. There are too many instances outside the normal statistical deviations. If that is the case, then the question of how and why is does not approach statistical randomness is a concern. if it isn't random, then what is the criteria for determination?

you and everyone else. i've personally gone 2 bags of ice with 0 bots pre-patch. does this mean the drops are bugged? probably not. just means i've been horrendously unlucky.

[Liwa]

i don't think a 10% success rate of getting a bot is comperable to a medicine. any medicine with a 10% success rate and a 90% mortality rate isn't going be be fielded so your comparison is, frankly, nonsense.

My example was to encourage investigating a problem early rather than waiting for a large sample with problems. I never said it was broken, I asked because of a recent change, i.e. patch and merge, was it possible something changed.

let's use a more realistic example: let's pretend you're asking people to pick a number from 1-10. you ask 15 people. no one picks #4. are you going to accuse the people you're asking of being broken?

15 people picking or thinking of a number from 1 to 10 is not even random ...

If you have 15 jars each containing 10 chips/cards or something with all the numbers from 1 to 10 and each of the 15 people pick a number from their own jar, that would be random.

And the probability of none of them picking a 4 is (9/10) to the 15th power.

And yes, the probability of a 16th person picking from a 16th jar any particular number from 1 to 10 for that draw is 1 in 10. That does not change the fact that the probability of having a string of 15 picks with no 4's is still (9/10) to the 15th power.

Kinnik

[Belph]

Independent events is what you're missing (I think) so you can't raise the odds to a power....

e.g. Flipping a coin 10 times won't mean on the 9th time I have a (1/2)^9 chance of getting a head if none were thrown before... It's still 1/2

Though the chance of you randomly doing 9 flips and no heads landing would indeed be (1/2)^9, subtle difference

[Solox]

Independent events is what you're missing (I think) so you can't raise the odds to a power....

e.g. Flipping a coin 10 times won't mean on the 9th time I have a (1/2)^9 chance of getting a head if none were thrown before... It's still 1/2

Though the chance of you randomly doing 9 flips and no heads landing would indeed be (1/2)^9, subtle difference

What is the probability of a coin toss landing heads? Most people, including Patrick Hawley, would say, “1/2!” but Adam Elgar replies differently; in the case of the “Sleeping Beauty Problem”: “1/3.” In this scenario, the subject, Beauty, is put to sleep on Sunday, and then a fair coin is tossed. If the coin toss yields heads, then Beauty is awakened on Monday and Wednesday. If the toss yields tails, Beauty will instead be awakened on Monday, Tuesday, and Wednesday. Throughout the duration of the experiment, Beauty is injected with a drug that removes memories of being awakened; thus, if she is awakened on Tuesday, she will not remember Monday’s disturbance. Thus, Elga’s answer of 1/3 heads is his response to a more complex version of the elementary question: “When the subject is awakened on Monday, to what degree will she believe that the outcome of the coin toss is heads?” Elga, a “thirder,” interestingly—but incorrectly—believes that the probabilities of being awakened on Monday with heads, Monday with tails, and Tuesday with tails are equal. Patrick Hawley, a “halfer,” refutes Elga’s argument, instead correctly stating that the answer to the problem is 1/2, since the probability of Beauty waking up on Monday is “1,” rather than less than “1″ as Elga states. Although Elga and Hawley create similar analyses at points, Hawley is ultimately correct in saying that the halfer’s conclusion possesses more merit than the thirder’s.

Elga incorrectly believes that the answer to the Sleeping Beauty problem is 1/3 heads and 2/3 tails, rather than 1/2 heads and 1/2 tails. To explain his reasoning, he first presents three “waking” possibilities: “H1: Heads on Monday,” “T1: Tails on Monday,” and “T2: Tails on Tuesday.”1 He states that if the researchers wake up Beauty and tell her it is Monday, she is living either the H1 or T1 circumstance; thus, H1 and T1 both have a probability of 1/2. If Beauty wakes up on either Monday or Tuesday and learns that the coin landed on tails, she is in either T1 or T2; therefore, T1 and T2 both have a probability of 1/2.2 Essentially, if H1 = T1 and T1 = T2, then, according to Elga’s “highly restrictive indifference principle,” H1 must equal T2.3 Therefore, H1 = T1 = T2. The likelihood of each waking must be 1/3, since they must yield a sum of 1. He also illustrates that Beauty’s thoughts should change from Sunday to Monday.4 On Sunday, she should believe indisputably that there is a 50% chance of heads and a 50% chance of tails, considering the coin is fair. Yet, when she is awakened on Monday, she should have confidence that there is a 1/3 chance each of H1, T1, and T2.5 Elga attributes the change in credence to the fact that when she wakes up on Monday, she considers her “temporal location as relevant to H,” in contrast to Sunday, when her temporal location was not relevant. If she is not told explicitly that it is Monday, she is aware that her place in time is somewhere in the midst of the experiment. According to Elga, this is relevant information that should cause her belief to change from a 1/2 chance of heads or tails on Sunday to a 1/3 chance of heads, and a 2/3 chance of tails on Monday.

[Manicmouse]

Epic copy paste there.

[Lazy]

Though the chance of you randomly doing 9 flips and no heads landing would indeed be (1/2)^9, subtle difference

but so are the odds of any other head/tails combination

[Belph]

Lol true ;D

[Colrain]

Independent events is what you're missing (I think) so you can't raise the odds to a power....

e.g. Flipping a coin 10 times won't mean on the 9th time I have a (1/2)^9 chance of getting a head if none were thrown before... It's still 1/2

Though the chance of you randomly doing 9 flips and no heads landing would indeed be (1/2)^9, subtle difference

Yes, I think that may be what Kinnikinnick might have been trying to say. That the chance of getting "no heads" many many times over (if true randomness is operating) is very low. Not that the next toss must be a head after many many tails.

Colrain

[Deemure]

while this is true, this doesn't affect the probability of the next raid being 10%. just because you do 100 raids with no drops doesn't guarantee a bot on the 101st raid.




no u



both are city raids. how much more identical do you need them to be?



sure they are. if you don't mind me quoting wikipedia, "Gambler's fallacy arises out of a belief in the law of small numbers, or the erroneous belief that small samples must be representative of the larger population. According to the fallacy, "streaks" must eventually even out in order to be representative.[7]"
there's nothing that says for every 9 raids with no bots you will get a bot on the 10th, or that for every bot you get you will get 9 raids of no bots directly after. but over the course of 10k raids you will have gotten about 10% drop rate. this is why i and everyone else has been saying that to be taken seriously you need a large sample size.



you and everyone else. i've personally gone 2 bags of ice with 0 bots pre-patch. does this mean the drops are bugged? probably not. just means i've been horrendously unlucky.

When you are talking about statistics, there are standard deviation tables specifically dealing with small populations. Those predictive numbers are slightly different than for large populations. Your last comment, to me, shows the problem. In statistics there is no such thing as "horrendously unlucky." There are statistics. Either you understand and believe in the math, or you don't. You don't need 10000 events, you don't need 1000. For most statistical verifications (and that is what we are talking about here verifying the 10% drop rate and randomness, you need as few as 67 events. If the results do not reflect the original premise, then you need to do several times the 67 samples to accurately calculate the correct original premise.

Again we are not talking about whether a string or events allows one to predict success or failure on the next raid, we are talking about whether the drop rates are random and whether the drop rate is 10% as premised. The strings of bots and no bots, being outside the statistical models for 10%, indicate that either the rate is not 10%, or that the rate is not random across all levels, cities, professions, or other unknown (at this time) criteria.

[Taranide]

A poem. Whatever the odds, it always feels like spiritual drops.

[Utopian]

When you are talking about statistics, there are standard deviation tables specifically dealing with small populations. Those predictive numbers are slightly different than for large populations. Your last comment, to me, shows the problem. In statistics there is no such thing as "horrendously unlucky." There are statistics. Either you understand and believe in the math, or you don't. You don't need 10000 events, you don't need 1000. For most statistical verifications (and that is what we are talking about here verifying the 10% drop rate and randomness, you need as few as 67 events. If the results do not reflect the original premise, then you need to do several times the 67 samples to accurately calculate the correct original premise.

Again we are not talking about whether a string or events allows one to predict success or failure on the next raid, we are talking about whether the drop rates are random and whether the drop rate is 10% as premised. The strings of bots and no bots, being outside the statistical models for 10%, indicate that either the rate is not 10%, or that the rate is not random across all levels, cities, professions, or other unknown (at this time) criteria.

This.

However, we are not talking about a single random occurrence and probability of that occurrence (i.e. raid to bot drop %) - we have to factor two other criteria into the formula (there maybe more that FC is using; but we know at least these two):

1. That all general types have the same probability of spawn -and-
2. That each general has the identical drop rate.

Only if these two factors are identical can we apply a model to predict ALL raids with ALL bot drops and place that predictability within the standard deviation.

Since 1 and 2 don't appear (in my raids) to be identical, the ability to predict all raids seems flawed from the beginning. And this "assumes" these are the only other factors to be taken into account - which again may be a giant leap of faith with the FC coding.

[Elithas]

I wrote a haiku to summarize the last 4 pages:

Every few years
someone whines about drop rates
nobody changed them

[Deemure]

Indeed Utopian, each general type does not seem to appear with equal frequency, nor does it appear that each type of general has the same drop rate. In addition, I am curious whether or not profession may influence drop rates (in over 230+ raids since the merge) 6 of the 7 bots I have gotten were with my fixer, the other with a MA. Also 5 of the 7 we gotten from a Serenity Island City, 2 from Playa, and 0 from the new large city plot. So I am also curious whether location plays a role as raids. Unfortunately, I do not have sufficient fidelity of detail or even enough information for any kind of statistical analysis, or any of the 3 or 4 other mathimatical techniques that may provide some insight.