Feedback New relic drop table?

[thephantom]

The amount is "correct" so as long as it is somewhere among the spectrum of probability, which is already in your case. The game being strategy doesn't change the fact that this particular item is based on a chance. My personal thought is that if you are "expecting" a specific result when it comes to probability, then you set yourself up for disappointment.

I fully understand what you are saying but from my point of view these aren't "fixes" for something, they're suggestions to change behaviour. I also don't think you will find many people who are happy with their number of relics for the simple fact that generally when people are happy they just continue playing. When they're not, they speak up. I know because I've had weeks with many relics (more than "expected" by that %) and I didn't tell anyone about it. But then weeks come when I have less (less than the "expected" by that %) and you will then find me complaining about it to my guild or friends.

 

[thephantom]

It's way too easy to "read" wrong patterns when there are none, when it comes to probability. Just look at all threads related to that (not just about relics) and you'll see people claiming you get better prizes if you stop the summer event wheel at a specific place, you get more relics if your battles are not frequent/quick, you get more apples if you click on a tree during a specific time of the day, etc.

Everyone has theories on how a chance-based system doesn't work as intended or can be "pushed" into non-chance behaviour. The reality is that the behaviour is random & leaning toward a %, but the sample size you're using/studying can mislead one.

Manganite's graph example in a prior post shows quite easily how for one instance it can seem like you're getting too few relics, while another person can feel they're getting too many. Taking all the data together, the average would be the "expected" one.

 

[DeletedUser7779]

Like many others mentioned, it's all based on randomness and if it weren't it would be cheating by Inno. If I may borrow manganite 's code, do you really think they have something like this? lol

if (player == Whiskey-s && ToR lvl > 75) %relic = %relic/2
if rand()<%relic
r = rand();
if 0<=r<%silver
you get a silver relic;
elseif %silver<r<%silver+%gold
you get a gold relic;
else
you get a jade relic;
end
end

I agree that it might be not so likely to get less relics than the expected value every time but then again your sample data of only 10 weeks is not sufficient to draw any conclusions.
There've been weeks when I've gotten 1 single relic for all GE encounters and just like you I thought the ToR is bugged. But there were other weeks also (like this one) where I got 2 jade, 2 gold and a bunch of silver relics before even finishing lvl 2. So in the end it all evens out, just more patience

 

[Manganite]

But in week 6 you got 2 relics more than the expectation value...

You also get on average much more gold and jade relics than you should expect. So your own stats proof that the nature of randomness will give one in some occasions results above the expectation value, in others below.

If you don´t believe this, just make an easy experiment. Take a few dices and throw each of them 100 times. And for each dice you write down how often they show a 1 or 2. The chance for that result is 33.33% percent. You will notice that some dice more than 33 time will show a 1 or 2, others will not. Maybe none of them will show this result for 33 or 34 times.

And also write down the length of the sequences you get no 1 or 2 for each specific dice. You will see that this sequences can be vastly longer than the average length of 3 one would naively expect from a probability of 33.33%.

Edit: NormaJeane - 2017-09-22 - removed reference(s) to deleted post.​

 

[Lionhead]

Frankly. Reading about this day in and day out is beginning to be a bit tiresome.
If Mensa truely is right, one would think you´d be wise enough to let it go.
It seems to me several people here have provided more than enough info to conclude, that your ToR is working just fine.

 

[NormaJeane]

Just a kind reminder of one of the forum rules to some posters in this thread:
- Please keep all discussions on topic, and do not post the same point repeatedly.

By now I've seen rather a lot of the same statements over and over again, it is really no use at all to keep repeating these...
even although the words may differ every now and then

 

[NormaJeane]

First of all thank you for your concise clarification
My remark was btw certainly NOT specifically targeted at you, but to ALL here repeating the same over and over again

Both you and the OP of this thread have a valid question, however the answer is not one that can be given easily. This is part due to ongoing in-game changes and also part due to the fact that game-designers of any game will - as far as my experience goes - rarely disclose all specific details such as a table on chances of a certain game feature.

On the other hand, all players are of course free to share their experiences in trying to come up with answers
I fully agree with what Darkstar said in this thread:

As for the compiling of information - I'm not sure why you'd expect any data to be directly provided by someone. Look at all the guides around and you'll see 99% of them are compiled from players for players. Lots of players (whether in a moderator role or not), have put together the data by compiling information provided by others.

So instead of continuing the 'yes-no-game' here: let's try to come up with information to put together a table!

 

[qaccy]

Both you and the OP of this thread have a valid question, however the answer is not one that can be given easily. This is part due to ongoing in-game changes and also part due to the fact that game-designers of any game will - as far as my experience goes - rarely disclose all specific details such as a table on chances of a certain game feature.

I do have to wonder why this is. I know that you're only a moderator here on the forums, but I've noticed this behavior in many games. Is there any insight to share as to why official probabilities are rarely given? For example, relic prizes...how come there isn't a listed probability of getting each reward? It's a little confusing to say things like 'reduced the chance of obtaining FP and goods from relic prizes' when nobody even knows what the chance of getting those prizes are in the first place...but it does tell us that the odds of getting all prizes are not equal. Some people are going to form their own conclusions regardless of what data is given, but some folks like me would be really interested to have more numbers to look at! People looking for an axe to grind can take the current lack of information and claim things similar to the recent discussion in this thread (though not exactly, since the % odds of a relic appearing ARE explicitly stated), such as Inno having secret 'dials to turn' on a whim that can adjust the rate of how often prizes appear week-to-week or even day-to-day. Maybe each time the game's updated they adjust a few 'random' percentages? Maybe winning medals from a gold relic is 40% one week but then another week it gets changed to 20%? We don't really know if that's happening or not because there's no table posted!

 

[NormaJeane]

I do have to wonder why this is. I know that you're only a moderator here on the forums, but I've noticed this behavior in many games. Is there any insight to share as to why official probabilities are rarely given?

Perhaps this comparison may clarify the secrecy part a bit: "The Coca-Cola recipe".
Although most of the ingredients of the Coca-Cola recipe are known, the exact ingredients, the exact amounts per ingredient and the exact way in which they are mixed have never been disclosed to the consumers. For over 125 years the Coca-Cola Company has kept their recipe a secret.

Also: over the years the Coca-Cola recipe has undergone changes, several varieties were introduced and are being marketed today. Consumers will have preferences for certain varieties or may dislike them totally. Their overhaul of the main recipe in 1985 was a total disaster... it took them some time to realize this, but in the end they adapted to what their consumers wanted. Still: The Coca-Cola Company keeps on developing new varieties and keeps on refining their recipes.

BTW: I prefer fruit juices and never ever drink Cola of any brand

The problem is, i didn't write my stats to prove something. I wrote the stats from 3 worlds as interesting data. There the system brings exactly the written percentage not every week but in average of a few weeks. I don't expect this everywhere all the time, but it brings these stats though. Maybe accidentally.

Please keep on collecting your stats data!
Plus: let's hope others will add their own stats also, that will be the only way to get more insight in the workings of ToR at higher levels

 

[Manganite]

As I explained already above, one can calculate this very easily... Taken your stats after 9 weeks one get with a certainty of 99% that your ToR gives you relics with a drop rate in the interval 22.60% - 32.30%. Everything beyond this is pure speculation...

Edit: NormaJeane - 2017-09-22 - removed reference(s) to deleted post.​

 

[Manganite]

Please keep on collecting your stats data!
Plus: let's hope others will add their own stats also, that will be the only way to get more insight in the workings of ToR at higher levels

You can collect data for ages, if you have no clue how to interpret them it will not help. As one can observe in all the FOE forums this only leads to wrong expectations, confusion and silly claims about broken functionality of the game...

Up to now I saw lot´s of discussions like this one here, but none of them were based on data that had any statistical significance to support seriously any claim that something is not working as it is supposed to do.

 

[Manganite]

Of course a company should care for reasonable feedback of it´s customers. But why should they investigate a problem, that does not exist? I think they should spend their time on solving real problems or add new, interesting features.

 

[RedRed]

Is there any insight to share as to why official probabilities are rarely given

One possibility is because if they are secret you can rebalance them in any moment without make players angry (not immediately, at least; everyone here knows what happens on rebalancing). This obviously is totally in the rights of the devs but implies several things:

• as developper you can't stop conspiracy theories, because you can't show data to prove that your system is fair;
• as player, you have to trust the devs (and you cannot think to plan the return of an investment based on unknown drop rates; f.e. the bonuses of rain forest project, seed vault, etc, ...);
• if you want to make some statistics, you have to collect a lot of data, and if you want to join the data with other players, be sure to do it correctly (same level of ToR, timestamp the results - because it might change over the time);

 

[Manganite]

I'm not sure about anything, the only thing I say is, there's no indication that something is wrong with the ToR, because the data provided are within in a reasonable range that one can expect for your ToR.

Of course the data are borderline at lower edge of what is possible, but nevertheless they are not impossible or something else. For example you say, it is almost impossible to have in 10 weeks 9 times a result below the expectation value. But it is possible. The chance is quite low, I calculated something about 0.3%, but that is not equal to impossible. Just imagine we have let's say 1 million active players around the world. That means there're 3000 other guys around experiencing the same.

Edit: NormaJeane - 2017-09-22 - removed reference(s) to deleted post.​

 

[Manganite]

First, the point is, that if your data would be really the result of an error in the code, then it should happen for everyone. So I would expect (regarding the high rate of complains in other cases), that something similar would be reported also by other players.

Second, as far as I understood you think, that your ToR works with a lower drop rate compared to the number that is shown. But you think beside that, the algorithm is working as supposed to work. Therefor, your data would be something like average data, just for a lower drop rate, let's say 27% instead of 31.5%. But that would mean, that other players should observe even lower average drop rates, something like 23%. But as far as I know, no one reported something like that. And it would be really noticeable even without writing down the numbers. Also it would be quite unlikely that anyone gets the supposed drop rate or something slightly above. Never saw any discussion about such behavior, too.

Third, I can not really imagine how this error should occur. If you imagine how the whole think might be implemented, then I would say, most probably somewhere in the code is a table with all the drop rates, and both, the algorithm that shows the drop rate on your ToR and the algorithm that decides about the win of a relic should use the same table. Why should they maintain 2 of them?

I guess one can also exclude the random number generator by itself, cause it works fine throughout the whole game.

Developers would be really stupid to produce an error of the kind you assume. As I showed you above, the code to make the decision of winning a relic or not is just one line of code...

Fourth, being critical is not a bad attitude by itself, but personally, without evidence for the opposite I would always assume things are working as supposed. Especially if everything I know about that issue is compatible with that assumption.

 

[Manganite]

That would be really stupid to maintain two list with drop rates and forgetting to update one of them... Stupid and inefficient...

As I wrote, the rest of the code is simple, just a few lines. I reprogrammed the whole relic selection algorithm to generate some data for proper statistics and it is really hard to make something wrong with that...

 

[qaccy]

Perhaps this comparison may clarify the secrecy part a bit: "The Coca-Cola recipe".
Although most of the ingredients of the Coca-Cola recipe are known, the exact ingredients, the exact amounts per ingredient and the exact way in which they are mixed have never been disclosed to the consumers. For over 125 years the Coca-Cola Company has kept their recipe a secret.

Also: over the years the Coca-Cola recipe has undergone changes, several varieties were introduced and are being marketed today. Consumers will have preferences for certain varieties or may dislike them totally. Their overhaul of the main recipe in 1985 was a total disaster... it took them some time to realize this, but in the end they adapted to what their consumers wanted. Still: The Coca-Cola Company keeps on developing new varieties and keeps on refining their recipes.

BTW: I prefer fruit juices and never ever drink Cola of any brand



Please keep on collecting your stats data!
Plus: let's hope others will add their own stats also, that will be the only way to get more insight in the workings of ToR at higher levels

That's an interesting way to look at it. Is a list of probabilities really comparable to a closely guarded trade secret recipe that a multi-billion dollar company is built upon, though? I personally don't think so. Continuing the example, I think it's fair to say that the outcome would be radically different if Coca-Cola suddenly had its secret recipe (including preparation) revealed to the public, compared to a small list of % chances here. Even if the numbers were later changed, could they not just be publicly updated as well? Some of us are data-minded, but your wording implies that numbers ARE in fact changed regularly behind the scenes, which actually renders collecting data irrelevant as if the conditions are regularly changing, there can't be any reliable numbers compiled by the players. A player-made average over the course of any period of time could be wholly inaccurate due to being too small or simply because Inno could be changing numbers and we'd never know it. So what's the point in gathering data in an environment like this? I do thank you for taking the time to respond, though!

 

[DeletedUser8079]

So... if there's missing relics, one thing the devs could double check is what happens when a relic is already in a location. One often appears off in the fog. If another appears and would have been in that location, what happens? Obviously it should just go into another spot, but the code should be double checked.

The devs certainly aren't infallible. It was only a couple months ago that the "You have a relic on this map" indicator in the GE broke for a bit. Mistakes happen

 

[Manganite]

That's really nice data. But it proofs nothing. You can make a simple hypothesis test. The expectation is a probability of 0.315, this is the null hypothesis. Your expectation is, that this is not true and the real value is somewhere below 0.315. The question is now, what is the threshold to throw away the null hypothesis if you play 64 encounters in the GE every week. Let's assume you want to test this against an error of 5%. Then you have to calculate the threshold for the number of relics so that the cumulative binomial distribution stays below 5%.

For N = 64 and p=0.315 this gives you a value of 13 relics. That means, if you got more than 13 relics in a single week the null hypothesis is true with a certainty of 95%.

You provide 20 samples for this test. And as you see only in the last week you got only 13 relics. In the 19 other weeks the value is higher, so you cannot dismiss the null hypothesis of p=0.315. In exactly 95% of your samples the null hypothesis has to be accepted. And that's exactly what one would expect.

The problem is, you have a wrong expectation how randomness works. Every week is an independent sample of the probability distribution behind the GE. Your chance to get a certain amount of relics is every week the same. It does not increase if you got some weeks a number below the expectation value. It stays always the same. And the chance to get less is quite high, e.g. the chance to get less than 20 relics in 64 trials is 43.6%!

Edit: NormaJeane - 2017-09-22 - removed reference(s) to deleted post.​

 

[Manganite]

Sorry, but I gave a mathematical evidence. Read this article in the wikipedia:

https://en.wikipedia.org/wiki/Null_hypothesis

The problem is, that you think, that your tiny 20 samples out of hundred of thousands of the other players are of any evidence for your hypothesis. I showed you step by step, that it is not.

Edit: NormaJeane - 2017-09-22 - removed reference(s) to deleted post.​