Flight Drop Rate Change For Major November Update

[Navigata07]

Hey everyone, so with a discussion I was having with a few other players (@Agent_L, @Wired Parrott , and @Capt.S.), they inspired me to put together a suggestion to GI on their facebook page. The message I left is in italics down below. It has to do with updating the flight drop rates across the board and creating the new percentage spreads based on the information given below. My goal is to flood the GI inbox with this new idea, and hopefully they will at least address the current flight drops for something that resembles our suggestion.
If you want to be a part of this initiative, please do the following:
1) Take the percentages and statement in the message below. Re-word them into your own language style.
2)Send to GI either via email or post on their Facebook wall under their most recent post.

"Hey, I have a suggestion for Airport City that may be a great change for the major update coming at the end of the year.
Drop rates from flights have been dismal, and are actually discouraging when it comes to making the game fun. With that being said, many players from the Airport City Forum came up with a percentage spread that would make things fair yet challenging. Here is the recommended spread below:
always: 100%
very often: 75-100%
often:50-75%
seldom:25-50%
very seldom:1-25%
Also, get rid of the "rare item" that is selected at random for each player. A "rare item" is the complete opposite of random, so give all items in a collection an even chance in the ranges specified. It's been years since the drop rates have been modified; I believe now would be a key time for a drop rate overhaul. Please pass on to the development team. Thanks."

 

[Aardvark]

Also, get rid of the "rare item" that is selected at random for each player. A "rare item" is the complete opposite of random, so give all items in a collection an even chance in the ranges specified."

Is this where you have a collection (be it normal or event) where you just can't get one item to drop? I've seen people complaining, and it frustrates me too, but if you look at the numbers, you rarely see something more than 3 SDs out, which suggests it's just variance. We have no way of proving one way or the other whether this issue exists, and if it doesn't, then asking about it undermines the integrity of the rest of your post, and you get dismissed with the rest of the tin-foil-hat-wearing brigade.

 

[Navigata07]

Is this where you have a collection (be it normal or event) where you just can't get one item to drop? I've seen people complaining, and it frustrates me too, but if you look at the numbers, you rarely see something more than 3 SDs out, which suggests it's just variance. We have no way of proving one way or the other whether this issue exists, and if it doesn't, then asking about it undermines the integrity of the rest of your post, and you get dismissed with the rest of the tin-foil-hat-wearing brigade.

But if that is indeed the case, then why is it that the "rare item" can actually change over time? If each item has the chance of varying by the 3 SD, then I would expect the "rare item" to change far more frequently than it does. The only time that collections can be done quickly is if you happen to catch the point of change where one "rare item" is replaced with another.

 

[Aardvark]

It changes slowly because you have to dig yourself out of one hole before getting into another. Whatever you are running low on needs to even itself out before you start running low on another item. Given my assumption that all drops are independent events running from a basic model, this takes time.

It would also be a pain in the *** to start programming a regularly changing rare item, bugs would pop up all over the place. I just don’t think it stands up to logic.

If Ivan comes here and tells me I’m wrong then fair play, but until then, I’m sticking with the maths, and Occam’s razor.

 

[ArGi1981]

Statistics is not intuitive for most, so I did the math for one specific instance, to show how quickly the number of items in one collection can be skewed:
Let's say we have a collection with 5 items, and each item has a 5% drop rate during some specific action (e.g., plane landing, or cloud clicking, etc). You perform said action 1000 times. Statistics then tells us that each of the five items drops 50 times on average. However, the standard deviation is 1.5 in that case, so the actual number of items we can have will be in the range of 45 to 55 (with 99.9% certainty). So after only 1000 actions with 5% drop rate per item, we can already see a difference of 10 items between the highest and lowest number in one collection. Increasing the drop rate (e.g., 10% drop rate, of 15%) or the number of actions (e.g., 2000 actions, or 10000) will only enlarge this potential difference (for instance, for 1000 actions with 10% drop rate, we have a standard deviation of 9, so we can have 100 +/- 27 of each item in one collection: A potential difference of more than 50!).

Each item is drawn independent of one another, so for each player the exact numbers will be different: For some players it might be pretty much evened out, while for some players the difference between the highest and lowest number in one collection might be pretty large. Which item has the highest and lowest number will be random too. And since all items are drawn independent of each other, we can see shifts over time as well, but these shifts will have to follow the same laws of statistics as the initial observations. Hence, if after a 1000 actions we have two items at the extremes of the scale, it will take at least another 1000 actions to balance them out again.

 

[Navigata07]

Statistics is not intuitive for most, so I did the math for one specific instance, to show how quickly the number of items in one collection can be skewed:
Let's say we have a collection with 5 items, and each item has a 5% drop rate during some specific action (e.g., plane landing, or cloud clicking, etc). You perform said action 1000 times. Statistics then tells us that each of the five items drops 50 times on average. However, the standard deviation is 1.5 in that case, so the actual number of items we can have will be in the range of 45 to 55 (with 99.9% certainty). So after only 1000 actions with 5% drop rate per item, we can already see a difference of 10 items between the highest and lowest number in one collection. Increasing the drop rate (e.g., 10% drop rate, of 15%) or the number of actions (e.g., 2000 actions, or 10000) will only enlarge this potential difference (for instance, for 1000 actions with 10% drop rate, we have a standard deviation of 9, so we can have 100 +/- 27 of each item in one collection: A potential difference of more than 50!).

Each item is drawn independent of one another, so for each player the exact numbers will be different: For some players it might be pretty much evened out, while for some players the difference between the highest and lowest number in one collection might be pretty large. Which item has the highest and lowest number will be random too. And since all items are drawn independent of each other, we can see shifts over time as well, but these shifts will have to follow the same laws of statistics as the initial observations. Hence, if after a 1000 actions we have two items at the extremes of the scale, it will take at least another 1000 actions to balance them out again.

Yeah I'm definitely not a stats guy, but I followed your explanation for the most part. I guess for me, I look at it as very odd that at the beginning when all is even, it still takes quite an imbalance for some items to get into the hole that they get in. I definitely see your logic when one item is rare while the next rarest may be 5 or 10 items larger, but when you see items being in a larger hole than that, one can't help but question how random it really is, and how random the SD is as well. I'll leave that argument as is because as I said, Im not a statistician. However, a change in the drop rate spread is still needed. When you have "very seldom" items ranging from 0.3 to ~30%, yet have "seldom" drops being around 10%, it's a bit too wide and undefined of a band. No need to have exact numbers of course, but the bands should make sense

 

[ArGi1981]

and how random the SD is as well

I'm just gonna pick this one out and then leave it at that: The Standard deviation is not random, it is implied by the drop rate and the number of times you perform the action. There is a one-to-one relation between the probability and the sample size on one hand and the standard deviation on the other: SD = SQRT( n ) * p * (1-p) where n is the sample size and p is the drop rate. For instance, with a drop rate of 5% and 1000 actions, the standard deviation is SQRT(1000) * 0.05 * ( 1 - 0.05) = ~1.5. With a drop rate of 10% and 100 actions, the standard deviation is 0.9.

That said, I'm all with you on changing the drop rates, especially if it is for the better

 

[Navigata07]

I'm just gonna pick this one out and then leave it at that: The Standard deviation is not random, it is implied by the drop rate and the number of times you perform the action. There is a one-to-one relation between the probability and the sample size on one hand and the standard deviation on the other: SD = SQRT( n ) * p * (1-p) where n is the sample size and p is the drop rate. For instance, with a drop rate of 5% and 1000 actions, the standard deviation is SQRT(1000) * 0.05 * ( 1 - 0.05) = ~1.5. With a drop rate of 10% and 100 actions, the standard deviation is 0.9.

That said, I'm all with you on changing the drop rates, especially if it is for the better

Cool I appreciate the explanation. Still can't stand stats, but if the usual rules of statistics are indeed being followed, then no harm done. These drop rates need to go up though. Dont want it to be easy, but make it that you dont have to fly for a year just to complete a collection 30 times.

 

[Free2Play]

Also, get rid of the "rare item" that is selected at random for each player. A "rare item" is the complete opposite of random, so give all items in a collection an even chance in the ranges specified. It's been years since the drop rates have been modified; I believe now would be a key time for a drop rate overhaul. Please pass on to the development team. Thanks."

There's no such thing as a 'rare item that is selected at random for each player', as I've said numerous times. There's no need to skew anything because basic deviation of drop rates for several items with equal chances already works good enough.

When you have "very seldom" items ranging from 0.3 to ~30%, yet have "seldom" drops being around 10%, it's a bit too wide and undefined of a band. No need to have exact numbers of course, but the bands should make sense

We don't have "very seldom" items ranging from 0.3 to ~30%, the actual percentages behind text descriptions ('seldom', 'very seldom', etc.) haven't changed once since game's release and the gaps aren't that huge. But of course if you have several active bonuses that muliply the chances — and you're lucky — the drop rate can increase drastically on some occasions.

 

[Navigata07]

There's no such thing as a 'rare item that is selected at random for each player', as I've said numerous times. There's no need to skew anything because basic deviation of drop rates for several items with equal chances already works good enough.



We don't have "very seldom" items ranging from 0.3 to ~30%, the actual percentages behind text descriptions ('seldom', 'very seldom', etc.) haven't changed once since game's release and the gaps aren't that huge. But of course if you have several active bonuses that muliply the chances — and you're lucky — the drop rate can increase drastically on some occasions.

Thanks for the response. The main thing I was looking at in terms of the "very seldom" band had to do with the Stockholm flights. I took a small sample of flights just to see how the items spread would look in terms of drops. I can agree with the standard deviation statement (I'm not a stats guy, but it does make sense that one item would be lagging due to SD), but when you look at the drop rates for Stockholm in regards to the Knowledge Is Power collection vs. The item drops for the furniture factory, I saw something very different. Both are "very seldom", but the building drops were around 30% compared to the Knowledge Is Power collection whose highest item drop was about 5%. It was an interesting yet odd observation. Maybe it wasn't meant to be that way, but that's how it is in my game.

 

[Bob69bob]

The title of this thread is click bait if ever I’ve seen it. On top of that, you’re asking people to join a campaign based on misinformation regarding the statistics of drop rates. You think you know what’s happening with the rates but you are wrong. I had misgivings regarding rates for awhile but instead of organising a massive nuisance, I did some more research and asked specific details from some very long term players. Luckily, one of them had a perfect example* to prove me wrong as well.

*It’s buried in the forum somewhere. I will endeavour find it and post it here.

 

[Navigata07]

The title of this thread is click bait if ever I’ve seen it. On top of that, you’re asking people to join a campaign based on misinformation regarding the statistics of drop rates. You think you know what’s happening with the rates but you are wrong. I had misgivings regarding rates for awhile but instead of organising a massive nuisance, I did some more research and asked specific details from some very long term players. Luckily, one of them had a perfect example* to prove me wrong as well.

*It’s buried in the forum somewhere. I will endeavour find it and post it here.

Well most people ignored the attempt to be a nuisance, so no harm done. I have no problem with being wrong. However, there is also no harm in drawing attention to something I noticed. For all we know, it could be an abnormality only with my game, or it could be perfectly fine. When I see things like the screenshot below, I have to question it. The screenshot below is my drops from Stockholm for the Knowledge collection and the Furniture Factory building items. Both are under the "very seldom" title when it comes to drops, by the way. The Knowledge collection had a little help from adventure buildings, but the building items are all from flights. If a statistician can look at this and say that it makes perfect sense with a logical explanation, then I will never utter "drop rates" again as long as I play this game.

 

[Bob69bob]

Well most people ignored the attempt to be a nuisance, so no harm done. I have no problem with being wrong. However, there is also no harm in drawing attention to something I noticed. For all we know, it could be an abnormality only with my game, or it could be perfectly fine. When I see things like the screenshot below, I have to question it. The screenshot below is my drops from Stockholm for the Knowledge collection and the Furniture Factory building items. Both are under the "very seldom" title when it comes to drops, by the way. The Knowledge collection had a little help from adventure buildings, but the building items are all from flights. If a statistician can look at this and say that it makes perfect sense, then I will never utter "drop rates" again as long as I play this game.
View attachment 15437

The problem here is that your sample size is way too small to draw any correct conclusion. It’s the problem with almost every piece of data posted on these forums, including Teboho’s launch chest stats. Here’s a challenge - get a dice and roll it 50 times and keep stats on what you roll and put those aside. Now adding to those figures, roll it another 250 times and put those stats aside. Finally, adding to the 300 so far, roll another 700 times and record the data. You’ll soon see what I’m talking about.

 

[Navigata07]

The problem here is that your sample size is way too small to draw any correct conclusion. It’s the problem with almost every piece of data posted on these forums, including Teboho’s launch chest stats. Here’s a challenge - get a dice and roll it 50 times and keep stats on what you roll and put those aside. Now adding to those figures, roll it another 250 times and put those stats aside. Finally, adding to the 300 so far, roll another 700 times and record the data. You’ll soon see what I’m talking about.

I know about sample sizes being small, since we don't have the dedication to measure the stats long term. However, I will say that those numbers are all that I have ever accumulated since I started playing the game, which is almost 2 years ago. I've probably flown well over 1000 flights to Stockholm, so it may not be 10000 flights, but it's still a very good amount. In this case, the SD spread for both sets is actually very good....that's not my issue. My issue is when you have two item sets in the same drop percentage category, but one set is consistently about 5 times more frequent than the other, one can't help but wonder.
Either way, it doesn't look like it will be addressed, as it's generally accepted that it's perfectly fine. With that being said, I'll leave the issue alone and just keep grinding. Appreciate the feedback from all of you guys, and if you find that thread, please share it. Curious about it.

 

[Bob69bob]

Here are a couple of samples but not the ones I was looking for. I’m curious as well - which collections are you talking about that are so far out of sync?

 

[Navigata07]

Here are a couple of samples but not the ones I was looking for. I’m curious as well - which collections are you talking about that are so far out of sync?

View attachment 15442 View attachment 15443

Hey that's neat. How'd you do that?

Those seem to have the SD variation that was discussed earlier. I've let that go, and it makes sense. I was talking about the collection for Stockholm "Knowledge is it's Own Reward" vs the building items for the Furniture Factory. Both are "very seldom" drops, but the Furniture Factory drops are 5 times more likely to occur, even though they are in the same band. I was just curious as to if that truly is statistically sound.

 

[Bob69bob]

Hey that's neat. How'd you do that?

Those seem to have the SD variation that was discussed earlier. I've let that go, and it makes sense. I was talking about the collection for Stockholm "Knowledge is it's Own Reward" vs the building items for the Furniture Factory. Both are "very seldom" drops, but the Furniture Factory drops are 5 times more likely to occur, even though they are in the same band. I was just curious as to if that truly is statistically sound.

Ok, got it. The question is not whether the rates are different within a collection, but rather what does GI use to determine the descriptor “very seldom” and are the descriptor values set the same across all collections.

The pics are from someone who simply hasn’t collected them yet.

 

[Navigata07]

Ok, got it. The question is not whether the rates are different within a collection, but rather what does GI use to determine the descriptor “very seldom” and are the descriptor values set the same across all collections.

The pics are from someone who simply hasn’t collected them yet.

Yes exactly. The band is too broad, hence the confusion. To add to the confusion is the "seldom" drops from adventure flights, which is approximately about 10% without a bonus item (based on some limited data collecting). The bands just need to be re-defined, or at least re-labeled.

 

[Wired Parrott]

The descriptors are not a very good gauge to determine the actual frequency of drop rate.
Being that in your stockholm example one is a collection and the other a building item I am
certain that is the difference.
As for that rare item I will say it is much more noticeable in event collections when you don't
have many collected. In this situation virtually everyone gets a rare item which SD in itself does
not explain. However once you have collected many items things come together and everything
fits within the SD.
Not sure exactly what is going on but find it telling that many games will give every formula used
within a game but GI feels these should be a secret for some unknown reason. As for tin foil hats
nothing to do with those just observation from years of playing.
The worst spread I can find at this point is this

But after getting a good number of event collections all look similar to this.

 

[Navigata07]

The descriptors are not a very good gauge to determine the actual frequency of drop rate.
Being that in your stockholm example one is a collection and the other a building item I am
certain that is the difference.
As for that rare item I will say it is much more noticeable in event collections when you don't
have many collected. In this situation virtually everyone gets a rare item which SD in itself does
not explain. However once you have collected many items things come together and everything
fits within the SD.
Not sure exactly what is going on but find it telling that many games will give every formula used
within a game but GI feels these should be a secret for some unknown reason. As for tin foil hats
nothing to do with those just observation from years of playing.
The worst spread I can find at this point is this

But after getting a good number of event collections all look similar to this.

To be honest, I'm still not onboard with even the SD spreads I'm seeing, but I am by no means a statistician so I leave it alone. For instance, look at the screenshot below

These two collections from the event shows exactly what I have an issue with. The bottom collection has an SD spread I am ok with (although the tablet skews things quite a bit). The "missing item" is only a few counts behind the next lowest item. That's perfectly fine. But look at the top collection. Excluding the laser pointers, all the other collections have a two point spread. That collection doesnt look like how a normal distribution should look; clearly there is a skew. However, I've been told that it's perfectly normal in statistics, so I let it be.